XE8805/05A Sensing Machine Data Acquisition MCU with Zooming ADC and DACs XE8805/05A - SX8805R Sensing Machine - Data Acquisition with 16+10 bit ZoomingADCTM and buffered DACs General Description Key product Features The XE8805A is a data acquisition ultra lowpower low-voltage system on a chip (SoC) with a high efficiency microcontroller unit embedded (MCU), allowing for 1 MIPS at 300uA and 2.4 V, and multiplying in one clock cycle. * The XE8805A includes a high resolution acquisition path with the 16+10 bits ZoomingADC and two buffered DACs. The XE8805A is available with on chip ROM (the SX8805) or Multiple-Time-Programmable (MTP) program memory. Applications * * * * * * * Portable, battery operated instruments Current loop powered instruments Wheatstone bridge interfaces Pressure and chemical sensors HVAC control Metering Sports watches, wrist instruments Rev 1 January 2006 * * * * * * * Low-power, high resolution ZoomingADC * * * 0.5 to 1000 gain with offset cancellation up to 16 bits analog to digital converter up to 13 inputs multiplexer Low-voltage low-power controller operation * * 2 MIPS with 2.4 V to 5.5 V operation 300 A at 1 MIPS over voltage range 22 kByte (8 kInstruction) MTP 520 Byte RAM data memory RC and crystal oscillators 5 reset, 22 interrupt, 8 event sources 8 bit and 16 bit buffered DACs 100 years MTP Flash retention at 55C Ordering Information Product Temperature range Memory type Package XE8805MI028* -40C to 85 C MTP XE8805AMI000 -40C to 85 C MTP LQFP64 die XE8805AMI028LF -40C to 85 C MTP LQFP64 SX8805Rxxx -40C to 85 C** ROM *Not for new designs **Extended temperature range on request www.semtech.com XE8805/05A Sensing Machine Data Acquisition MCU with Zooming ADC and DACs TABLE OF CONTENTS Chapter 1 Chapter 2 Chapter 3 Chapter 4 Chapter 5 Chapter 6 Chapter 7 Chapter 8 Chapter 9 Chapter 10 Chapter 11 Chapter 12 Chapter 13 Chapter 14 Chapter 15 Chapter 16 Chapter 17 Chapter 18 Chapter 19 Chapter 20 Chapter 21 Chapter 22 (c) Semtech 2006 XE8805/05A Overview XE8805/05A Performance XE8805/05A CPU XE8805/05A Memory System Block Reset generator Clock generation Interrupt handler Event handler Low power RAM Port A Port B Port C Universal Asynchronous Receiver/Transmitter (UART) Universal Synchronous Receiver/Transmitter (USRT) Acquisition Chain (ZoomingADCTM) Voltage multiplier Signal D/A (DAS) Bias D/A (DAB) Counters/Timers/PWM The Voltage Level Detector XE8805/05A Dimensions www.semtech.com XE8805/05A 1. General overview CONTENTS 1.1 1.1.1 1.1.2 Top schematic General description XE8805 vs XE8805A 1-2 1-2 1-4 1.2 1.2.1 1.2.2 Pin map Bare die LQFP-64 1-4 1-4 1-5 1.3 Pin assignment 1-6 (c) Semtech 2006 www.semtech.com 1-1 XE8805/05A 1.1 Top schematic 1.1.1 General description The top level block schematic of the circuit is shown in Figure 1-1. The heart of the circuit consists of the Coolrisc816(R) CPU core. This core includes an 8x8 multiplier and 16 internal registers. The bus controller generates all control signals for access to all data registers other than the CPU internal registers. The reset block generates the adequate reset signals for the rest of the circuit as a function of the set-up contained in its control registers. Possible reset sources are the power-on-reset (POR), the external pin RESET, the watchdog (WD), a bus error detected by the bus controller or a programmable pattern on Port A. Different low power modes are implemented. The clock generation and power management block sets up the clock signals and generates internal supplies for different blocks. The clock can be generated from the RC oscillator (this is the start-up condition), the crystal oscillator (XTAL) or an external clock source (given on the OSCIN pin). The test controller generates all set-up signals for different test modes. In normal operation, it is used as a set of 8 low power data registers. If power consumption is important for the application, the variables that need to be accessed very often should be stored in these registers rather than in the RAM. The IRQ handler routes the interrupt signals of the different peripherals to the IRQ inputs of the CPU core. It allows masking of the interrupt sources and it flags which interrupt source is active. Events are generally used to restart the processor after a HALT period without jumping to a specified address, i.e. the program execution resumes with the instruction following the HALT instruction. The EVN handler routes the event signals of the different peripherals to the EVN inputs of the CPU core. It allows masking of the interrupt sources and it flags which interrupt source is active. The Port B is an 8 bit parallel IO port with analog capabilities. The URST, UART, and PWM block also make use of this port. The instruction memory is a 22-bit wide flash or ROM memory depending on the circuit version. Flash and ROM versions have both 8k instruction memory. The data memory of this product is 512 byte SRAM. The Acquisition Chain is a high resolution acquisition path with the 16+10 bit fully differential ZoomingADCTM. The VMULT (voltage multiplier) powers a part of the Acquisition Chain. The signal D/A (DAS) is a 16 bit D/A based on sigma-delta modulation. It includes a stand-alone amplifier that can be used for analog output filtering. The bias D/A (DAB) is an 8 bit low frequency D/A. It includes a stand-alone amplifier that is used to drive large currents. It can be used to bias a sensor. The Port A is an 8 bit parallel input port. It can also generate interrupts, events or a reset. It can be used to input external clocks for the timer/counter/PWM block. The Port C is a general purpose 8 bit parallel I/O port. The USRT (universal synchronous receiver/transmitter) contains some simple hardware functions in order to simplify the software implementation of a synchronous serial link. (c) Semtech 2006 www.semtech.com 1-2 XE8805/05A DATA MEMORY INSTRUCTION MEMORY VPP/TEST CPU COOLRISC816 VBAT VSS 8 X 8 MULTIPLIER B U S C O N T R O L L E R address control PORT A PA(7:0) datain dataout PORT C PC(7:0) 16 CPU REGISTERS RESET RESET BLOCK WD reset control ACQUISITION CHAIN ZoomingADC CLOCK GENERATION/ POWER MANAGEMENT RC OSCIN OSCOUT VREG AC_R(3:0) POR XTAL VREG TM AC_A(7:0) clocks test control VMULT VMULT TEST CONTROLLER DAS_OUT DAS_AI_P DAS_AI_M DAS_AO DAS Signal D/A 8 DATA REGISTERS irq IRQ HANDLING DAB_R_P DAB_R_M DAB_OUT DAB_AI_P DAB_AI_M DAB_AO_P DAB_AO_M DAB Bias D/A EVN HANDLING evn COUNTERS TIMERS PWM VLD PB(7:6) UART PB(1:0) PA(3:0) USRT PB(7:0) PB(5:4) PORT B Figure 1-1. Block schematic of the XE8805/05A circuit. The UART (universal asynchronous receiver/transmitter) contains a full hardware implementation of the asynchronous serial link. The counters/timers/PWM can take their clocks from internal or external sources (on Port A) and can generate interrupts or events. The PWM is output on Port B. (c) Semtech 2006 www.semtech.com 1-3 XE8805/05A The VLD (voltage level detector) detects the battery end of life with respect to a programmable threshold. 1.1.2 XE8805 vs XE8805A The XE8805A has a new RESET pin function. The action of the RESET pin of the XE8805A resets the clock registers too and creates an additional short delay. See the RESET chapter for more information. 1.2 Pin map ( 428.5, 4453.4) ( 593.5, 4453.4) ( 758.5, 4453.4) ( 923.5, 4453.4) (1088.5, 4453.4) (1252.9, 4453.4) (1418.5, 4453.4) (1588.5, 4453.4) (1753.5, 4453.4) (1923.5, 4453.4) (3114.6, 4453.4) (3293.5, 4453.4) (3458.5, 4453.4) (3628.5, 4453.4) VSS OSCIN VSS OSCOUT RESET VMULT VREG VSS_REG VSS VBAT DAS_AO DAS_AI_M DAS_AI_P DAS_OUT 1.2.1 Bare die (52.6,4123.5) PA(0) AC_R(0) (3958.4, 4118.5) (52.6, 3908.5) PA(1) AC_R(1) (3958.4, 3858.5) (52.6,3693.5) PA(2) (52.6, 3478.5) PA(3) VSS (3958.4, 3603.5) AC_A(0) (3958.4, 3343.5) (52.6, 3263.5) VBAT (52.6, 3048.5) PA(4) AC_A(1) (3958.4, 3088.5) (52.6, 2833.5) PA(5) AC_A(2) (3958.4, 2828.5) (52.6, 2618.5) PA(6) AC_A(3) (3958.4, 2573.5) (52.6, 2403.5) PA(7) (52.6, 2188.5) PC(0) AC_A(4) (3958.4, 2313.5) VBAT (3958.4, 2058.5) PC(1) (52.6, 1758.5) PC(2) AC_A(5) (3958.4,1798.5) (52.6, 1543.5) PC(3) AC_A(6) (3958.4, 1543.5) AC_A(7) (3958.4, 1283.5) AC_R(2) (3958.4, 1028.5) AC_R(3) (3958.4, 768.5) VPP/TEST (3958.4, 508.5) PC(5) (52.6, 683.5) PC(6) (52.6, 468.5) PC(7) 4100 ( 398.5, 47.6) ( 533.5, 47.6) ( 668.5, 47.6) ( 798.5, 47.6) ( 933.5, 47.6) (1063.5, 47.6) (1198.5, 47.6) (1328.5, 47.6) (1463.5, 47.6) (1593.5, 47.6) (1728.5, 47.6) (1858.5, 47.6) (2042.4, 47.6) (3363.5,47.6) VSS (3498.5, 47.6) DAB_AI_P (3628.4, 47.6) DAB_AI_M (52.6, 898.5) (2683.3, 47.6) DAB_AO_M VSS PC(4) PB(0) PB(1) PB(2) PB(3) PB(4) VBAT PB(5) PB(6) PB(7) DAB_R_P DAB_R_M DAB_OUT DAB_AO_P (52.6, 1328.5) (52.6, 1113.5) 4600 (52.6, 1973.5) Figure 1-2. Die dimensions and pin coordinates (in m) (c) Semtech 2006 www.semtech.com 1-4 XE8805/05A 1.2.2 LQFP-64 PC(7) 15 NC AC_R(3) PB(0) PB(1) PB(2) PB(3) PB(4) PB(5) PB(6) DAB_R_P DAB_R_M PB(7) 20 VPP/TEST 25 DAB_OUT DAB_AO_P DAB_AO_M DAB_AI_P DAB_AI_M 30 NC The XE8805/05A is delivered in a LQFP-64 package. The pin map is given below. PC(6) PC(5) 35 AC_R(2) PC(4) AC_A(7) PC(3) PC(2) AC_A(6) AC_A(5) AC_A(4) 10 PC(1) PC(0) 40 AC_A(3) PA(7) AC_A(2) PA(6) PA(5) AC_A(1) 5 AC_A(0) AC_R(1) PA(4) PA(3) 45 PA(2) AC_A(0) PA(1) 1 PA(0) NC OSCIN OSCOUT RESET 60 VMULT NC VREG VSS_REG VSS VBAT DAS_AO DAS_AI_M DAS_AI_P DAS_OUT NC NC 50 NC 55 NC Figure 1-3. LQFP-64 pin map Package pin 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 name PA(0) PA(1) PA(2) PA(3) PA(4) PA(5) PA(6) PA(7) PC(0) PC(1) PC(2) PC(3) PC(4) PC(5) PC(6) PC(7) PB(0) PB(1) PB(2) PB(3) Package pin 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 name VPP/TEST NC AC_R(3) AC_R(2) AC_A(7) AC_A(6) AC_A(5) AC_A(4) AC_A(3) AC_A(2) AC_A(1) AC_A(0) AC_R(1) AC_R(0) NC NC NC NC DAS_OUT DAS_AI_P (c) Semtech 2006 www.semtech.com 1-5 XE8805/05A Package pin 21 22 23 24 25 26 27 28 29 30 31 32 name PB(4) PB(5) PB(6) PB(7) DAB_R_P DAB_R_M DAB_OUT DAB_AO_P DAB_AO_M DAB_AI_P DAB_AI_M NC Package pin 53 54 55 56 57 58 59 60 61 62 63 64 name DAS_AI_M DAS_AO VBAT VSS VSS_REG VREG NC VMULT RESET OSCOUT OSCIN NC Table 1-1. Bonding plan of the LQFP-64 package (LQFP 64L 10x10mm thick 1.6 mm) 1.3 Pin assignment The table below gives a short description of the different pin assignments. Pin Assignment VBAT VSS VSS_REG VREG VPP/TEST RESET OSCIN/OSCOUT PA(7:0) PB(7:0) PC(7:0) AC_A(7:0) AC_R(3:0) VMULT DAB_OUT DAB_R_x DAB_Ax_y DAS_OUT DAS_AI_x DAS_AO Positive power supply Negative power supply Connection for the mandatory external capacitor of the voltage regulator High voltage supply for flash memory programming (NC in ROM versions) Resets the circuit when the voltage is high Quartz crystal connections, also used for flash memory programming Parallel input port A pins Parallel I/O port B pins Parallel I/O port C pins Acquisition chain input pins Acquisition chain reference pins Connection for the external capacitor if VBAT is below 3V Bias D/A output Bias D/A reference (x=P: plus, x=M: minus) Bias D/A amplifier IO (x=I: input, x=O: output ; y=P: plus, y=M: minus) Signal D/A output Signal D/A amplifier inputs (x=P: plus, x=M: minus) Signal D/A amplifier output Table 1-2. Pin assignment Table 1-3 gives a more detailed pin map for the different pins. It also indicates the possible I/O configuration of these pins. The indications in blue bold are the configuration at start-up. The pins CNTx pins are possible counter inputs, PWMx are possible PWM outputs. (c) Semtech 2006 www.semtech.com 1-6 XE8805/05A PWM0 PWM1 USRT_S0 USRT_S1 UART_Tx UART_Rx X X X X X X X X X X POWER PU OD DI X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X CNTA CNTB CNTC CNTD DO AO AI PA(0) PA(1) PA(2) PA(3) PA(4) PA(5) PA(6) PA(7) PC(0) PC(1) PC(2) PC(3) PC(4) PC(5) PC(6) PC(7) PB(0) PB(1) PB(2) PB(3) PB(4) PB(5) PB(6) PB(7) DAB_R_P DAB_R_M DAB_OUT DAB_AO_P DAB_AO_M DAB_AI_P DAB_AI_M VPP AC_R(3) AC_R(2) AC_A(7) AC_A(6) AC_A(5) AC_A(4) AC_A(3) AC_A(2) AC_A(1) AC_A(0) AC_R(1) AC_R(0) DAS_OUT DAS_AI_P DAS_AI_M DAS_AO VBAT third 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 33 35 36 37 38 39 40 41 42 43 44 45 46 51 52 53 54 55 I/O configuration second first function lqfp-64 pin X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X TEST X X X X X X X X X X X X X X X X X X (c) Semtech 2006 www.semtech.com 1-7 XE8805/05A POWER PU OD DO DI AO AI VSS VSS_REG VREG VMULT RESET OSCOUT OSCIN third 56 57 58 60 61 62 63 I/O configuration second first function lqfp-64 pin X X X X X X X Pin map table legend: blue bold: configuration at start up AI: analog input AO: analog output DI: digital input DO: digital output OD: nMOS open drain output PU: pull-up resistor POWER: power supply Table 1-3. Pin description table (c) Semtech 2006 www.semtech.com 1-8 XE8805/05A 2 XE8805/05A Performance 2.1 Absolute maximum ratings 2-2 2.2 Operating range 2-2 2.3 2.3.1 2.3.2 Supply configurations Flash circuit ROM circuit 2-3 2-3 2-3 2.4 Current consumption 2-5 2.5 2.5.1 2.5.2 Operating speed Flash version ROM circuit version 2-6 2-6 2-6 (c) Semtech 2006 www.semtech.com 2-1 XE8805/05A 2.1 Absolute maximum ratings Table 2-1. Absolute maximum ratings Voltage applied to VBAT with respect to VSS Voltage applied to VPP with respect to VSS Voltage applied to all pins except VPP and VBAT Storage temperature (ROM device or unprogrammed flash device) Storage temperature (programmed flash device) Min. Max. Note -0.3 VBAT-0.3 VSS-0.3 -55 6.0 12 VBAT+0.3 150 V V V C -40 85 C Stresses beyond the absolute maximal ratings may cause permanent damage to the device. Functional operation at the absolute maximal ratings is not implied. Exposure to conditions beyond the absolute maximal ratings may affect the reliability of the device. 2.2 Operating range Table 2-2. Operating range for the flash device Voltage applied to VBAT with respect to VSS Voltage applied to VBAT with respect to VSS during the flash programming Voltage applied to VPP with respect to VSS Voltage applied to all pins except VPP and VBAT Operating temperature range Capacitor on VREG (flash version) Capacitor on VMULT 1. 2. 3. Min. Max. Note 2.4 3.3 5.5 5.5 V V 1 VBAT VSS -40 0.8 1.0 11.5 VBAT 85 1.2 3.0 V V C F nF 2 3 During the programming of the device, the supply voltage should at least be equal to the supply voltage used during normal operation. The capacitor on VREG is mandatory. The capacitor on VMULT is optional. The capacitor has to be present if the multiplier is enabled. The multiplier has to be enabled if VBAT<3.0V. Table 2-3. Operating range for the ROM device Min. Voltage applied to VBAT with respect to VSS Voltage applied to all pins except VPP and VBAT Operating temperature range Capacitor on VREG Capacitor on VMULT 1. 2. 2.4 VSS -40 0.1 1.0 Max. 5.5 VBAT 125 1.2 3.0 Note V V C F nF 1 2 The capacitor may be omitted when VREG is connected to VBAT. The capacitor on VMULT is optional. The capacitor has to be present if the multiplier is enabled. The multiplier has to be enabled if VBAT<3.0V. All specifications in this document are valid for the complete operating range unless otherwise specified. (c) Semtech 2006 www.semtech.com 2-2 XE8805/05A Table 2-4. Operating range of the Flash memory Min. Retention time at 85C Retention time at 55C Number of programming cycles 1. 2. 2.3 Max. 10 100 10 Note years years 1 1 2 Valid only if programmed using a programming tool that is qualified Circuits can be programmed more than 10 times but after that, the retention time is no longer guaranteed. All qualification tests have been done after 10 cycles. Supply configurations 2.3.1 Flash circuit The flash version of the circuit can be run from a supply between 2.4V and 5.5V (Figure 2-1). The capacitor on VREG has to be connected at all times (value in Table 2-2) to guarantee proper operation of the device. The capacitor on VMULT is only required if the circuit is to be operated below 3V. VBAT VREG 2.4V - 5.5V VMULT Cvreg Cvmult VSS Figure 2-1. Supply configuration for the flash circuit. 2.3.2 ROM circuit For the ROM version, two possible operating modes exist: with and without voltage regulator. Using the voltage regulator, low power consumption will be obtained even with supply voltages above 2.4V. Without the voltage regulator (i.e. VREG short-circuited to VBAT), a higher speed can be obtained. 2.3.2.1 Low power operation In this case, the internal voltage regulator is used in order to maintain low power consumption independent from the supply voltage. The capacitor on VREG has to be connected at all times (value in Table 2-3) to guarantee proper operation of the device. The capacitor on VMULT has to be connected only when VBAT<3V. (c) Semtech 2006 www.semtech.com 2-3 XE8805/05A VBAT VREG 2.4V - 5.5V VMULT Cvreg Cvmult VSS Figure 2-2. Supply voltage connections for low power operation of the ROM version. 2.3.2.2 High speed operation In this case, the internal voltage regulator is not used. The operation speed of the circuit can be increased with increasing supply voltage but the supply current will also increase. The capacitor on VMULT has to be connected only when VBAT<3V. In this case, the supply voltage can decrease down to 2.15V. Beware however that the zoomingADCTM will not run below 2.4V (see Figure 2-4). In this configuration, the circuit can not be used above 3.3V. VBAT VREG 2.15V - 3.3V VMULT Cvmult VSS Figure 2-3. Supply voltage connections for high speed operation of the ROM version. (c) Semtech 2006 www.semtech.com 2-4 XE8805/05A acquisition chain voltage multiplier DAB CPU parallel and serial ports RC and crystal oscillator VLD Counters and PWM DAS (without amplifier) 0 2.15 2.4 3.3 VBAT (V) Figure 2-4. Operation range of the different circuit blocks 2.4 Current consumption The tables below give the current consumption for the circuit in different configurations. The figures are indicative only and may change as a function of the actual software implemented in the circuit. Table 2-5 gives the current consumption for the flash version of the circuit. The peripherals are disabled. The parallel ports A and B are configured in input with pull up, the parallel port C is configured as an output. Their pins are not connected externally. The pin RESET is connected to VSS and the pin VPP/TEST is connected to VBAT. The inputs of the acquisition chain are connected to VSS. Table 2-5. Typical current consumption of the XE8805 version (8k instructions flash memory) Operation mode CPU RC Xtal Consumption comments Note High speed CPU Low power CPU Low power time keeping Fast wake-up time keeping Immediate wakeup time keeping VLD static current 16 bit resolution data acquisition 12 bit , gain 100, data acquisition 1 MIPS 32 kIPS HALT 1 MHz Off Off Off 32 kHz 32 kHz 310 A 10 A 1.0 A 2.4V<>5.5V, 27C 2.4V <>5.5V, 27C 2.4V <>5.5V, 27C HALT Ready 32kHz 1.7 A 2.4V <>5.5V, 27C HALT 100 kHz Off 1.4 A 2.4V <>5.5V, 27C HALT 2 MHz Off 15 A 190 A 2.4V <>5.5V, 27C 3.0V, 27C 1 HALT 2 MHz Off 460 A 3.0V, 27C 2 1. PGA disabled, ADC enabled, 16 bit resolution 2. PGA 1 disabled, PGA 2 and 3 enabled, ADC enabled, 12 bit resolution For more information concerning the current consumption of the zoomingADCTM, see the chapter power consumption in the acquisition chain documentation which shows the current consumption of this block as a function of temperature and voltage and for different configurations of the PGA and ADC. The power consumption of the ROM version of the circuit is identical if it is configured as shown in Figure 2-2. In the high speed configuration, the current consumption will increase proportional with VBAT. (c) Semtech 2006 www.semtech.com 2-5 XE8805/05A 2.5 Operating speed 2.5.1 Flash version The speed of the devices is not highly dependent upon the supply voltage. However, by limiting the temperature range, the speed can be increased. The minimal guaranteed speed as a function of the supply voltage and maximal temperature operating temperature is given in Figure 2-5. speed (MIPS) 4 3 2 1 85C 45C 0 2 2.5 3 3.5 4 4.5 5 5.5 supply voltage VBAT (V) Figure 2-5. Guaranteed speed as a function of the supply voltage and maximal temperature. 2.5.2 ROM circuit version 2.5.2.1 Low power supply configuration In the low power supply configuration as shown in Figure 2-2, the operating speed does not depend highly on the supply voltage as shown in Figure 2-6. speed (MIPS) 85C 45C 125C 3.5 3 2.5 2 1.5 1 0.5 0 2 2.5 3 3.5 4 4.5 5 5.5 supply voltage VBAT (V) Figure 2-6. Guaranteed speed as a function of supply voltage and for different maximal temperatures using the voltage regulator. 2.5.2.2 High speed supply configuration In the high speed supply configuration of Figure 2-3, the guaranteed speed of the circuit is shown in Figure 2-7. (c) Semtech 2006 www.semtech.com 2-6 XE8805/05A 85C 45C 125C speed (MIPS) 4 3 2 1 0 2 2.2 2.4 2.6 2.8 3 3.2 3.4 supply voltage VBAT (V) Figure 2-7. Guaranteed speed as a function of supply voltage and for three temperature ranges when VREG=VBAT. (c) Semtech 2006 www.semtech.com 2-7 XE8805/05A 3. CPU CONTENTS 3.1 CPU description 3-2 3.2 CPU internal registers 3-2 3.3 CPU instruction short reference 3-4 (c) Semtech 2006 www.semtech.com 3-1 XE8805/05A 3.1 CPU description The CPU of the XE8000 series is a low power RISC core. It has 16 internal registers for efficient implementation of the C compiler. Its instruction set is made up of 35 generic instructions, all coded on 22 bits, with 8 addressing modes. All instructions are executed in one clock cycle, including conditional jumps and 8x8 multiplication. The circuit therefore runs on 1 MIPS on a 1MHz clock. The CPU hardware and software description is given in the document "Coolrisc816 Hardware and Software Reference Manual". A short summary is given in the following paragraphs. The good code efficiency of the CPU core makes it possible to compute a polynomial like Z = ( A0 + A1 Y ) X + B0 + B1 Y in less than 300 clock cycles (software code generated by the XEMICS Ccompiler, all numbers are signed integers on 16 bits). 3.2 CPU internal registers As shown in Figure 3-1, the CPU has 16 internal 8-bit registers. Some of these registers can be concatenated to a 16-bit word for use in some instructions. The function of these registers is defined in Table 3-1. The status register stat (Table 3-2) is used to manage the different interrupt and event levels. An interrupt or an event can both be used to wake up after a HALT instruction. The difference is that an interrupt jumps to a special interrupt function whereas an event continues the software execution with the instruction following the HALT instruction. The program counter (PC) is a 16 bit register that indicates the address of the instruction that has to be executed. The stack (STn) is used to memorise the return address when executing subroutines or interrupt routines. ST4 ST3 ST1 ST2 r0 r1 Data memory r2 data bus 22bit CPU CPU internal registers Instruction memory instruction bus PC program counter stack r3 i0h i0l i1h i1l i2h i2l i3h i3l iph ipl stat a Figure 3-1. CPU internal registers (c) Semtech 2006 www.semtech.com 3-2 XE8805/05A Register name r0 r1 r2 r3 i0h i0l i1h i1l i2h i2l i3h i3l iph ipl stat a Register function general purpose general purpose general purpose data memory offset MSB of the data memory index i0 LBS of the data memory index i0 MSB of the data memory index i1 LBS of the data memory index i1 MSB of the data memory index i2 LBS of the data memory index i2 MSB of the data memory index i3 LBS of the data memory index i3 MSB of the program memory index ip LBS of the program memory index ip status register accumulator Table 3-1. CPU internal register definition bit 7 6 5 4 name IE2 IE1 GIE IN2 3 IN1 2 IN0 1 EV1 0 EV0 function enables (when 1) the interrupt request of level 2 enables (when 1) the interrupt request of level 1 enables (when 1) all interrupt request levels interrupt request of level 2. The interrupts labelled "low" in the interrupt handler are routed to this interrupt level. This bit has to be cleared when the interrupt is served. interrupt request of level 1. The interrupts labelled "mid" in the interrupt handler are routed to this interrupt level. This bit has to be cleared when the interrupt is served. interrupt request of level 0. The interrupts labelled "hig" in the interrupt handler are routed to this interrupt level. This bit has to be cleared when the interrupt is served. event request of level 1. The events labelled "low" in the event handler are routed to this event level. This bit has to be cleared when the event is served. event request of level 1. The events labelled "hig" in the event handler are routed to this event level. This bit has to be cleared when the event is served. Table 3-2. Status register description The CPU also has a number of flags that can be used for conditional jumps. These flags are defined in Table 3-3. symbol Z C V name zero carry overflow function Z=1 when the accumulator a content is zero This flag is used in shift or arithmetic operations. For a shift operation, it has the value of the bit that was shifted out (LSB for shift right, MSB for shift left). For an arithmetic operation with unsigned numbers: it is 1 at occurrence of an overflow during an addition (or equivalent). it is 0 at occurrence of an underflow during a subtraction (or equivalent). This flag is used in shift or arithmetic operations. For arithmetic or shift operations with signed numbers, it is 1 if an overflow or underflow occurs. Table 3-3. Flag description (c) Semtech 2006 www.semtech.com 3-3 XE8805/05A 3.3 CPU instruction short reference Table 3-4 shows a short description of the different instructions available on the Coolrisc816. The notation cc in the conditional jump instruction refers to the condition description as given in Table 3-6. The notation reg, reg1, reg2, reg3 refers to one of the CPU internal registers of Table 3-1. The notation eaddr and DM(eaddr) refer to one of the extended address modes as defined in Table 3-5. The notation DM(xxx) refers to the data memory location with address xxx. Instruction Modification Operation Jump addr[15:0] Jump ip Jcc addr[15:0] Jcc ip Call addr[15:0] Call ip Calls addr[15:0] Calls ip -,-,-, -,-,-, -,-,-, -,-,-, -,-,-, -,-,-, -,-,-, -,-,-, - PC := addr[15:0] PC := ip if cc is true then PC := addr[15:0] if cc is true then PC := ip STn+1 := STn (n>1); ST1 := PC+1; PC := addr[15:0] STn+1 := STn (n>1); ST1 := PC+1; PC := ip ip := PC+1; PC := addr[15:0] ip := PC+1; PC := ip Ret Rets Reti Push Pop -,-,-, -,-,-, -,-,-, -,-,-, -,-,-, - PC := ST1; STn := STn+1 (n>1) PC := ip PC := ST1; STn := STn+1 (n>1); GIE :=1 PC := PC+1; STn+1 := STn (n>1); ST1 := ip PC := PC+1; ip := ST1; STn := STn+1 (n>1) Move reg,#data[7:0] Move reg1, reg2 Move reg, eaddr Move eaddr, reg Move addr[7:0],#data[7:0] -,-, Z, a -,-, Z, a -,-, Z, a -,-,-, -,-,-, - a := data[7:0]; reg := data[7:0] a := reg2; reg1 := reg2 a := DM(eaddr); reg := DM(eaddr) DM(eaddr) := reg DM(addr[7:0]) := data[7:0] Cmvd reg1, reg2 Cmvd reg, eaddr Cmvs reg1, reg2 Cmvs reg, eaddr -,-, Z, a -,-, Z, a -,-, Z, a -,-, Z, a a := reg2; if C=0 then reg1 := a; a := DM(eaddr); if C=0 then reg := a a := reg2; if C=1 then reg1 := a; a := DM(eaddr); if C=1 then reg := a Shl reg1, reg2 Shl reg Shl reg, eaddr Shlc reg1, reg2 Shlc reg Shlc reg, eaddr Shr reg1, reg2 Shr reg Shr reg, eaddr Shrc reg1, reg2 Shrc reg Shrc reg, eaddr Shra reg1, reg2 Shra reg Shra reg, eaddr C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a a := reg2<<1; a[0] := 0; C := reg2[7]; reg1 := a a := reg<<1; a[0] := 0; C := reg[7]; reg := a a := DM(eaddr)<<1; a[0] :=0; C := DM(eaddr)[7]; reg := a a := reg2<<1; a[0] := C; C := reg2[7]; reg1 := a a := reg<<1; a[0] := C; C := reg[7]; reg := a a := DM(eaddr)<<1; a[0] := C; C := DM(eaddr)[7]; reg := a a := reg2>>1; a[7] := 0; C := reg2[0]; reg1 :=a a := reg>>1; a[7] := 0; C := reg[0]; reg := a a := DM(eaddr)>>1; a[7] := 0; C := DM(eaddr)[0]; reg := a a := reg2>>1; a[7] := C; C := reg2[0]; reg1 := a a := reg>>1; a[7] := C; C := reg[0]; reg := a a := DM(eaddr)>>1; a[7] := C; C := DM(eaddr)[0]; reg := a a := reg2>>1; a[7] := reg2[7]; C := reg2[0]; reg1 := a a := reg>>1; a[7] := reg[7]; C := reg[0]; reg := a a := DM(eaddr)>>1; a[7] := DM(eaddr)[7]; C := DM(eaddr)[0]; reg := a Cpl1 reg1, reg2 Cpl1 reg Cpl1 reg, eaddr Cpl2 reg1, reg2 Cpl2 reg Cpl2 reg, eaddr Cpl2c reg1, reg2 Cpl2c reg Cpl2c reg, eaddr -,-, Z, a -,-, Z, a -,-, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a a := NOT(reg2); reg1 := a a := NOT(reg); reg := a a := NOT(DM(eaddr)); reg := a a := NOT(reg2)+1; if a=0 then C:=1 else C := 0; reg1 := a a := NOT(reg)+1; if a=0 then C:=1 else C := 0; reg := a a := NOT(DM(eaddr))+1; if a=0 then C:=1 else C := 0; reg := a a := NOT(reg2)+C; if a=0 and C=1 then C:=1 else C := 0; reg1 := a a := NOT(reg)+C; if a=0 and C=1 then C:=1 else C := 0; reg := a a := NOT(DM(eaddr))+C; if a=0 and C=1 then C:=1 else C := 0; reg := a Inc reg1, reg2 Inc reg Inc reg, eaddr Incc reg1, reg2 Incc reg Incc reg, eaddr Dec reg1, reg2 C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a a := reg2+1; if a=0 then C := 1 else C := 0; reg1 := a a := reg+1; if a=0 then C := 1 else C := 0; reg := a a := DM(eaadr)+1; if a=0 then C := 1 else C := 0; reg := a a := reg2+C; if a=0 and C=1 then C := 1 else C := 0; reg1 := a a := reg+C; if a=0 and C=1 then C := 1 else C := 0; reg := a a := DM(eaadr)+C; if a=0 and C=1 then C := 1 else C := 0; reg := a a := reg2-1; if a=hFF then C := 0 else C := 1; reg1 := a (c) Semtech 2006 www.semtech.com 3-4 XE8805/05A Dec reg Dec reg, eaddr Decc reg1, reg2 Decc reg Decc reg, eaddr C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a a := reg-1; if a=hFF then C := 0 else C := 1; reg := a a := DM(eaddr)-1; if a=hFF then C := 0 else C := 1; reg := a a := reg2-(1-C); if a=hFF and C=0 then C := 0 else C := 1; reg1 := a a := reg-(1-C); if a=hFF and C=0 then C := 0 else C := 1; reg := a a := DM(eaddr)-(1-C); if a=hFF and C=0 then C := 0 else C := 1; reg := a And reg,#data[7:0] And reg1, reg2, reg3 And reg1, reg2 And reg, eaddr Or reg,#data[7:0] Or reg1, reg2, reg3 Or reg1, reg2 Or reg, eaddr Xor reg,#data[7:0] Xor reg1, reg2, reg3 Xor reg1, reg2 Xor reg, eaddr -,-, Z, a -,-, Z, a -,-, Z, a -,-, Z, a -,-, Z, a -,-, Z, a -,-, Z, a -,-, Z, a -,-, Z, a -,-, Z, a -,-, Z, a -,-, Z, a a := reg and data[7:0]; reg := a a := reg2 and reg3; reg1 := a a := reg1 and reg2; reg1 := a a := reg and DM(eaddr); reg := a a := reg or data[7:0]; reg := a a := reg2 or reg3; reg1 := a a := reg1 or reg2; reg1 := a a := reg or DM(eaddr); reg := a a := reg xor data[7:0]; reg := a a := reg2 xor reg3; reg1 := a a := reg1 xor reg2; reg1 := a a := reg or DM(eaddr); reg := a Add reg,#data[7:0] Add reg1, reg2, reg3 Add reg1, reg2 Add reg, eaddr Addc reg,#data[7:0] Addc reg1, reg2, reg3 Addc reg1, reg2 Addc reg, eaddr Subd reg,#data[7:0] Subd reg1, reg2, reg3 Subd reg1, reg2 Subd reg, eaddr Subdc reg,#data[7:0] Subdc reg1, reg2, reg3 Subdc reg1, reg2 Subdc reg, eaddr Subs reg,#data[7:0] Subs reg1, reg2, reg3 Subs reg1, reg2 Subs reg, eaddr Subsc reg,#data[7:0] Subsc reg1, reg2, reg3 Subsc reg1, reg2 Subsc reg, eaddr C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a a := reg+data[7:0]; if overflow then C:=1 else C := 0; reg := a a := reg2+reg3; if overflow then C:=1 else C := 0; reg1 := a a := reg1+reg2; if overflow then C:=1 else C := 0; reg1 := a a := reg+DM(eaddr); if overflow then C:=1 else C := 0; reg := a a := reg+data[7:0]+C; if overflow then C:=1 else C := 0; reg := a a := reg2+reg3+C; if overflow then C:=1 else C := 0; reg1 := a a := reg1+reg2+C; if overflow then C:=1 else C := 0; reg1 := a a := reg+DM(eaddr)+C; if overflow then C:=1 else C := 0; reg := a a := data[7:0]-reg; if underflow then C := 0 else C := 1; reg := a a := reg2-reg3; if underflow then C := 0 else C := 1; reg1 := a a := reg2-reg1; if underflow then C := 0 else C := 1; reg1 := a a := DM(eaddr)-reg; if underflow then C := 0 else C := 1; reg := a a := data[7:0]-reg-(1-C); if underflow then C := 0 else C := 1; reg := a a := reg2-reg3-(1-C); if underflow then C := 0 else C := 1; reg1 := a a := reg2-reg1-(1-C); if underflow then C := 0 else C := 1; reg1 := a a := DM(eaddr)-reg-(1-C); if underflow then C := 0 else C := 1; reg := a a := reg-data[7:0]; if underflow then C := 0 else C := 1; reg := a a := reg3-reg2; if underflow then C := 0 else C := 1; reg1 := a a := reg1-reg2; if underflow then C := 0 else C := 1; reg1 := a a := reg-DM(eaddr); if underflow then C := 0 else C := 1; reg := a a := reg-data[7:0]-(1-C); if underflow then C := 0 else C := 1; reg := a a := reg3-reg2-(1-C); if underflow then C := 0 else C := 1; reg1 := a a := reg1-reg2-(1-C); if underflow then C := 0 else C := 1; reg1 := a a := reg-DM(eaddr)-(1-C); if underflow then C := 0 else C := 1; reg := a Mul reg,#data[7:0] Mul reg1, reg2, reg3 Mul reg1, reg2 Mul reg, eaddr Mula reg,#data[7:0] Mula reg1, reg2, reg3 Mula reg1, reg2 Mula reg, eaddr u, u, u, a u, u, u, a u, u, u, a u, u, u, a u, u, u, a u, u, u, a u, u, u, a u, u, u, a a := (data[7:0]*reg)[7:0]; reg := (data[7:0]*reg)[15:8] a := (reg2*reg3)[7:0]; reg1 := (reg2*reg3)[15:8] a := (reg2*reg1)[7:0]; reg1 := (reg2*reg1)[15:8] a := (DM(eaddr)*reg)[7:0]; reg := (DM(eaddr)*reg)[15:8] a := (data[7:0]*reg)[7:0]; reg := (data[7:0]*reg)[15:8] a := (reg2*reg3)[7:0]; reg1 := (reg2*reg3)[15:8] a := (reg2*reg1)[7:0]; reg1 := (reg2*reg1)[15:8] a := (DM(eaddr)*reg)[7:0]; reg := (DM(eaddr)*reg)[15:8] Mshl reg,#shift[2:0] Mshr reg,#shift[2:0] Mshra reg,#shift[2:0] u, u, u, a u, u, u, a u, u, u, a* a := (reg*2 )[7:0]; reg := (reg*2 )[15:8] (8-shift (8-shift a := (reg*2 )[7:0]; reg := (reg*2 )[15:8] (8-shift (8-shift a := (reg*2 )[7:0]; reg := (reg*2 )[15:8] Cmp reg,#data[7:0] Cmp reg1, reg2 Cmp reg, eaddr Cmpa reg,#data[7:0] Cmpa reg1, reg2 Cmpa reg, eaddr C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a C, V, Z, a a := data[7:0]-reg; if underflow then C :=0 else C:=1; V := C and (not Z) a := reg2-reg1; if underflow then C :=0 else C:=1; V := C and (not Z) a := DM(eaddr)-reg; if underflow then C :=0 else C:=1; V := C and (not Z) a := data[7:0]-reg; if underflow then C :=0 else C:=1; V := C and (not Z) a := reg2-reg1; if underflow then C :=0 else C:=1; V := C and (not Z) a := DM(eaddr)-reg; if underflow then C :=0 else C:=1; V := C and (not Z) Tstb reg,#bit[2:0] Setb reg,#bit[2:0] Clrb reg,#bit[2:0] Invb reg,#bit[2:0] -, -, Z, a -, -, Z, a -, -, Z, a -, -, Z, a a[bit] := reg[bit]; other bits in a are 0 reg[bit] := 1; other bits unchanged; a := reg reg[bit] := 0; other bits unchanged; a := reg reg[bit] := not reg[bit]; other bits unchanged; a := reg shift shift (c) Semtech 2006 www.semtech.com 3-5 XE8805/05A Sflag -,-,-, a a[7] := C; a[6] := C xor V; a[5] := ST full; a[4] := ST empty Rflag reg Rflag eaddr C, V, Z, a C, V, Z, a a := reg << 1; ; a[0] := 0; C := reg[7] a := DM(eaddr)<<1; a[0] :=0; C := DM(eaddr)[7] Freq divn -,-,-, - reduces the CPU frequency (divn=nodiv, div2, div4, div8, div16) Halt -,-,-, - halts the CPU Nop -,-,-, - no operation - = unchanged, u = undefined, *MSHR reg,# 1 doesn't shift by 1 Table 3-4. Instruction short reference The Coolrisc816 has 8 different addressing modes. These modes are described in Table 3-5. In this table, the notation ix refers to one of the data memory index registers i0, i1, i2 or i3. Using eaddr in an instruction of Table 3-4 will access the data memory at the address DM(eaddr) and will simultaneously execute the index operation. extended address eaddr addr[7:0] (ix) (ix, offset[7:0]) (ix,r3) (ix)+ (ix,offset[7:0])+ -(ix) -(ix,offset[7:0]) accessed data memory location DM(eaddr) DM(h00&addr[7:0]) DM(ix) DM(ix+offset) DM(ix+r3) DM(ix) DM(ix+offset) DM(ix-1) DM(ix-offset) index operation ix := ix+1 ix := ix+offset ix := ix-1 ix := ix -offset direct addressing indexed addressing indexed addressing with immediate offset indexed addressing with register offset indexed addressing with index post-increment indexed addressing with index post-increment by the offset indexed addressing with index pre-decrement indexed addressing with index pre-decrement by the offset Table 3-5. Extended address mode description Eleven different jump conditions are implemented as shown in Table 3-6. The contents of the column CC in this table should replace the CC notation in the instruction description of Table 3-4. CC condition CS CC ZS ZC VS VC EV C=1 C=0 Z=1 Z=0 V=1 V=0 (EV1 or EV0)=1 After CMP op1,op2 EQ NE GT GE LT LE op1=op2 op1op2 op1>op2 op1op2 op1<op2 op1op2 Table 3-6. Jump condition description (c) Semtech 2006 www.semtech.com 3-6 XE8805/05A 4 Memory mapping 4.1 Memory organisation 4-2 4.2 4.2.1 4.2.2 4.2.3 4.2.4 4.2.5 4.2.6 4.2.7 4.2.8 4.2.9 4.2.10 4.2.11 4.2.12 4.2.13 4.2.14 4.2.15 4.2.16 4.2.17 Quick reference data memory register map Low power data registers (h0000-h0007) System, clock configuration and reset configuration (h0010-h001F) Port A (h0020-h0027) Port B (h0028-h002F) Port C (h0030-h0033) Flash programming (h0038-003B) Event handler (h003C-h003F) Interrupt handler (h0040-h0047) USRT (h0048-h004F) UART (h0050-h0057) Counter/Timer/PWM registers (h0058-h005F) Acquisition chain registers (h0060-h0067) Signal D/A registers (h0074-h0077) Bias D/A registers (h0078-h0079) Voltage multiplier (h007C) Voltage Level Detector registers (h007E-h007F) RAM (h0080-h027F) 4-2 4-3 4-4 4-4 4-4 4-5 4-5 4-5 4-6 4-7 4-7 4-7 4-8 4-8 4-8 4-8 4-9 4-9 (c) Semtech 2006 www.semtech.com 4-1 XE8805/05A 4.1 Memory organisation The XE8805 CPU is built with a Harvard architecture. The Harvard architecture uses separate instruction and data memories. The instruction bus and data bus are also separated. The advantage of such a structure is that the CPU can get a new instruction and read/write data simultaneously. The circuit configuration is shown in Figure 4-1. The CPU has its 16 internal registers. The instruction memory has a capacity of 8192 22-bit instructions. The data memory space has 8 low power registers, the peripheral register space and 512 bytes of RAM. F0h1FFF F 0h027F CPU capacity: 512 bytes instruction bus r2 r3 i0l i1h i1l i2h i2l i3h i3l 0h007F Peripheral registers iph ipl 0h0008 stat 0h0000 0h0080 i0h Data memory r1 data bus capacity: 8k x 22bit r0 CPU internal registers Instruction memory RAM Low power RAM 0h0000 a Figure 4-1. Memory mapping The CPU internal registers are described in the CPU chapter. A short reference of the low power registers and peripheral registers is given in 4.2. 4.2 Quick reference data memory register map The data register map is given in the tables below. A more detailed description of the different registers is given in the detailed description of the different peripherals. The tables give the following information: 1. The register name and register address 2. The different bits in the register 3. The access mode of the different bits (see Table 4-1 for code description) 4. The reset source and reset value of the different bits The reset source coding is given in Table 4-2. To get a full description of the reset sources, please refer to the reset block chapter. (c) Semtech 2006 www.semtech.com 4-2 XE8805/05A code r w r0 r1 c c1 ca s access mode bit can be read bit can be written bit always reads 0 bit always reads 1 bit is cleared by writing any value bit is cleared by writing a 1 bit is cleared after reading special function, verify the detailed description in the respective peripherals Table 4-1. Access mode codes used in the register definitions code sys cold pconf sleep reset source resetsystem resetcold resetpconf resetsleep Table 4-2. Reset source coding used in the register definitions 4.2.1 Low power data registers (h0000-h0007) Name Address 7 6 Reg00 h0000 Reg01 h0001 Reg02 h0002 Reg03 h0003 Reg04 h0004 Reg05 h0005 Reg06 h0006 Reg07 h0007 5 4 3 Reg00[7:0] rw, xxxxxxxx,Reg01[7:0] rw,xxxxxxxx,Reg02[7:0] rw,xxxxxxxx,Reg03[7:0] rw,xxxxxxxx,Reg04[7:0] rw,xxxxxxxx,Reg05[7:0] rw,xxxxxxxx,Reg06[7:0] rw,xxxxxxxx,Reg07[/:0] rw,xxxxxxxx,- 2 1 0 Table 4-3. Low power data registers (c) Semtech 2006 www.semtech.com 4-3 XE8805/05A 4.2.2 System, clock configuration and reset configuration (h0010-h001F) Name Address 7 RegSysCtrl SleepEn h0010 rw,0,cold RegSysReset Sleep h0011 rw,0,sys RegSysClock CpuSel h0012 rw,0,sleep RegSysMisc r0 h0013 RegSysWd r0 h0014 RegSysPre0 r0 h0015 RegSysRcTrim1 r0 h001B RegSysRcTrim2 r0 h001C 6 EnResPConf rw,0,cold 5 EnBusError rw,0,cold r0 r0 r0 r0 r0 3 2 1 0 r0 r0 r0 ResPad ResPadSleep rc, 0, cold rc, 0, cold r0 rc,0,cold rc,0,cold EnExtClock ColdXtal ColdRC EnableXtal EnableRC rw,0,cold r,1,sleep r,1,sleep rw,0,sleep rw,1,sleep RCOnPA0 DebFast OutputCkXtal OutputCpuCk r0 r0 rw,0,sleep rw,0,sleep rw,0,sleep rw,0,sleep WatchDog[3:0] r0 r0 s,0000,cold ResPre r0 r0 r0 r0 r0 c1r0,0,Reserved RcFreqRange RcFreqCoarse[3:0] rw,0,cold rw,0000,cold rw,0,cold RcFreqFine[5:0] rw,10000,cold ResetBusError ExtClk r,0,cold 4 EnResWD rw,0,cold ResetWD rc, 0, cold BiasRC rw,1,cold r0 ResetfromportA Table 4-4. Reset block and clock block registers 4.2.3 Port A (h0020-h0027) Name Address 7 6 5 RegPAIn h0020 RegPADebounce h0021 RegPAEdge h0022 RegPAPullup h0023 RegPARes0 h0024 RegPARes1 h0025 4 3 PAIn[7:0] r PADebounce[7:0] rw,00000000,pconf PAEdge[7:0] rw,00000000,sys PAPullup[7:0] rw,00000000,pconf PARes0[7:0] rw, 00000000, sys PARes1[7:0] rw,00000000,sys 2 1 0 4 2 1 0 r0 PBAna[3:0] rw,0000,pconf Table 4-5. Port A registers 4.2.4 Port B (h0028-h002F) Name Address 7 6 5 r0 r0 r0 RegPBOut h0028 RegPBIn h0029 RegPBDir h002A RegPBOpen h002B RegPBPullup h002C RegPBAna h002D 3 PBOut[7:0] rw,00000000,pconf PBIn[7:0] r PBDir[7:0] rw,00000000,pconf PBOpen[7:0] rw,00000000,pconf PBPullup[7:0] rw,00000000,pconf Table 4-6. Port B registers (c) Semtech 2006 www.semtech.com 4-4 XE8805/05A 4.2.5 Port C (h0030-h0033) Name Address 7 6 5 4 3 PCOut[7:0] rw,00000000,pconf PCIn[7:0] r,-,PD1Dir[7:0] rw,00000000,pconf RegPCOut h0030 RegPCIn h0031 RegPCDir h0032 2 1 0 Table 4-7. Port C registers 4.2.6 Flash programming (h0038-003B) These four registers are used during flash programming only. Refer to the flash programming algorithm documentation for more details. 4.2.7 Event handler (h003C-h003F) Name Address RegEvn h003C RegEvnEn h003D RegEvnPriority h003E RegEvnEvn h003F 7 CntIrqA rc1,0,sys 6 CntIrqC rc1,0,sys 5 128Hz rc1,0,sys r0 r0 r0 4 3 PAEvn[1] CntIrqB rc1,0,sys rc1,0,sys EvnEn[7:0] rw,00000000,sys EvnPriority[7:0] r,11111111,sys r0 r0 2 CntIrqD rc1,0,sys 1 1Hz rc1,0,sys 0 PAEvn[0] rc1,0,sys r0 EvnHigh r,0,sys EvnLow r,0,sys Table 4-8. Event handler registers The origin of the different events is summarised in the table below. Event CntIrqA CntIrqB CntIrqC CntIrqD 128Hz 1Hz PAEvn[1:0] Event source Counter/Timer A (counter block) Counter/Timer B (counter block) Counter/Timer C (counter block) Counter/Timer D (counter block) Low prescaler (clock block) Low prescaler (clock block) Port A Table 4-9. Event source description (c) Semtech 2006 www.semtech.com 4-5 XE8805/05A 4.2.8 Interrupt handler (h0040-h0047) Name Address 7 IrqAC h0040 rc1,0,sys RegIrqMid UsrtCond1 h0041 rc1,0,sys RegIrqLow PAIrq[7] h0042 rc1,0,sys RegIrqEnHig h0043 RegIrqEnMid h0044 RegIrqEnLow h0045 RegIrqPriority h0046 RegIrqIrq r0 h0047 RegIrqHig 6 128Hz rc1,0,sys UrstCond2 rc1,0,sys PAIrq[6] rc1,0,sys 5 r0 PAIrq[5] rc1,0,sys CntIrqB rc1,0,sys r0 r0 4 3 CntIrqA CntIrqC rc1,0,sys rc1,0,sys PAIrq[4] 1Hz rc1,0,sys rc1,0,sys CntIrqD PAIrq[3] rc1,0,sys rc1,0,sys IrqEnHig[7:0] rw,0000000,sys IrqEnMid[7:0] rw,0000000,sys IrqEnLow[7:0] rw,0000000,sys IrqPriority[7:0] r,11111111,sys r0 r0 2 r0 VldIrq rc1,0,sys PAIrq[2] rc1,0,sys IrqHig r,0,sys 1 UartIrqTx rc1,0,sys PAIrq[1] rc1,0,sys 0 UartIrqRx rc1,0,sys PAIrq[0] rc1,0,sys r0 r0 IrqMid r,0,sys IrqLow r,0,sys Table 4-10. Interrupt handler registers The origin of the different interrupts is summarised in the table below. Event CntIrqA CntIrqB CntIrqC CntIrqD 128Hz 1Hz PAIrq[7:0] UartIrqRx UartIrqTx UrstCond1 UsrtCond2 VldIrq IrqAC Event source Counter/Timer A (counter block) Counter/Timer B (counter block) Counter/Timer C (counter block) Counter/Timer D (counter block) Low prescaler (clock block) Low prescaler (clock block) Port A UART reception UART transmission USRT condition 1 USRT condition 2 Voltage level detector Acquisition chain end of conversion interrupt Table 4-11. Interrupt source description (c) Semtech 2006 www.semtech.com 4-6 XE8805/05A 4.2.9 USRT (h0048-h004F) Name Address RegUsrtS1 h0048 RegUsrtS0 h0049 RegUsrtCond1 h004A RegUsrtCond2 h004B RegUsrtCtrl h004C RegUsrtBufferS1 h004D RegUsrtEdgeS0 h004E 7 6 5 4 3 2 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 UsrtEnWaitCond1 r0 r0 r0 r0 r0 UsrtWaitS0 r,0,sys r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 r0 rw,0,sys 1 0 UsrtS1 r0 s,1,sys UsrtS0 r0 s,1,sys UsrtCond1 r0 rc,0,sys UsrtCond2 r0 rc,0,sys UsrtEnWaitS0 UsrtEnable rw,0,sys rw,0,sys UsrtBufferS1 r0 r,0,sys UsrtEdgeS0 r0 r,0,sys Table 4-12. USRT register description 4.2.10 UART (h0050-h0057) Name Address RegUartCtrl h0050 RegUartCmd h0051 RegUartTx h0052 RegUartTxSta h0053 RegUartRx h0054 RegUartRxSta h0055 7 UartEcho rw,0,sys SelXtal rw,0,sys 6 UartEnRx1 rw,0,sys UartEnRx2 rw,0,sys r0 r0 r0 r0 5 UartEnTx rw,0,sys 4 3 UartXRx UartXTx rw,0,sys rw,0,sys UartRcSel[2:0] rw,000,sys UartTx[7:0] rw,0000000,sys 2 UartPM rw,0,sys 1 UartBR[2:0] rw,101,sys UartPE rw,0,sys 0 UartWL rw,1,sys UartTxBusy UartTxFull r0 r0 r0 r,0,sys r,0,sys UartRx[7:0] r,00000000,sys UartRxSErr UartRxPErr UartRxFErr UartRxOerr UartRxBusy UartRxFull r,0,sys r,0,sys r,0,sys rc,0,sys r,0,sys r,0,sys r0 Table 4-13. UART register description 4.2.11 Counter/Timer/PWM registers (h0058-h005F) Name Address 7 RegCntA h0058 RegCntB h0059 RegCntC h005A RegCntD h005B RegCntCtrlCk h005C RegCntConfig1 h005D RegCntConfig2 h005E RegCntOn h005F 6 5 4 3 2 1 0 CounterA[7:0] s,xxxxxxxx,CounterB[7:0] s,xxxxxxxx,CounterC[7:0] s,xxxxxxxx,CounterD[7:0] s,xxxxxxxx,CntDCkSel[1:0] CntCCkSel[1:0] CntBCkSel[1:0] CntACkSel[1:0] rw,xx,rw,xx,rw,xx,rw,xx,CntDDownUp CntCDownUp CntBDownUp CntADownUp CascadeCD CascadeAB CntPWM1 CntPWM0 rw,x,rw,x,rw,0,sys rw,0,sys rw,x,rw,x,rw,x,rw,x,CapSel[1:0] CapFunc[1:0] Pwm1Size[1:0] Pwm0Size[1:0] rw,00,sys rw,00,sys rw,xx,rw,xx,CntDEnable CntCEnable CntBEnable CntAEnable r0 r0 r0 rw,0,sys rw,0,sys rw,0,sys rw,0,sys r0 Table 4-14. Counter/timer/PWM register description. (c) Semtech 2006 www.semtech.com 4-7 XE8805/05A 4.2.12 Acquisition chain registers (h0060-h0067) Name Address 7 6 5 4 3 2 1 RegAcOutLsb OUT[7:0] h0060 r,0,sys RegAcOutMsb OUT[15:8] h0061 r,0,sys RegAcCfg0 START SET_NELCONV[1:0] SET_OSR[2:0] CONT rw,010,sys h0062 w r0,0,sys rw,01,sys rw,0,sys RegAcCfg1 IB_AMP_ADC[1:0] IB_AMP_PGA[1:0] ENABLE[3:0] h0063 rw,11,sys rw,11,sys rw,0000,sys RegAcCfg2 FIN PGA2_GAIN[1:0] PGA2_OFFSET[3:0] h0064 rw,00,sys rw,00,sys rw,0000,sys RegAcCfg3 PGA1_GAIN PGA3_GAIN[6:0] h0065 rw,0000000,sys Rw,0,sys PGA3_OFFSET RegAcCfg4 r0 rw,0000000,sys h0066 RegAcCfg5 BUSY DEF AMUX[4:0] h0067 r,0,sys w r0 rw,00000,sys 0 r0 VMUX rw,0,sys Table 4-15. Acquisition chain register description. 4.2.13 Signal D/A registers (h0074-h0077) Name Address RegDasInLsb h0074 RegDasInMsb h0075 RegDasCfg0 h0076 RegDasCfg1 h0077 7 6 5 NsOrder(1:0) rw,00,sys r0 r0 r0 4 3 DasInLsb(7:0) rw,00000000,sys DasInMsb(7:0) rw,00000000,sys CodeLmax(2:0) rw,000,sys r0 r0 2 1 0 Enable(1:0) rw,00,sys BW r0 rw,0,sys Fin rw,0,sys Inv rw,0,sys Table 4-16. Signal D/A register description 4.2.14 Bias D/A registers (h0078-h0079) Name Address 7 6 5 r0 r0 r0 r0 r0 r0 5 4 3 2 Enable rw,0,sys RegDabIn h0078 4 3 DabIn(7:0) rw,00000000,sys 2 1 Enable(1:0) rw,00,sys RegDabCfg h0079 0 Table 4-17. Bias D/A register description 4.2.15 Voltage multiplier (h007C) Name Address RegVmultCfg0 h007C 7 r0 6 r0 r0 r0 r0 1 0 Fin[1:0] rw,00,sys Table 4-18. VMULT register. (c) Semtech 2006 www.semtech.com 4-8 XE8805/05A 4.2.16 Voltage Level Detector registers (h007E-h007F) Name Address 7 6 5 4 r0 r0 r0 r0 3 VldRange rw,0,sys r0 r0 r0 r0 r0 RegVldCtrl h007E RegVldStat h007F 2 1 VldTune[2:0] rw,000,sys VldResult VldValid r,0,sys r,0,sys 0 VldEn rw,0,sys Table 4-19. Voltage level detector register description 4.2.17 RAM (h0080-h027F) The 512 RAM bytes can be accessed for read and write operations. The RAM has no reset function. Variables stored in the RAM should be initialised before use since they can have any value at circuit start up. (c) Semtech 2006 www.semtech.com 4-9 XE8805/05A 5 System Block 5.1 Overview 5-2 5.2 Operating mode 5-2 (c) Semtech 2006 www.semtech.com 5-1 XE8805/05A 5.1 Overview The XE8000 chips have three operating modes. There is the normal, the low current and the very low current modes (see Figure 5-1). The different modes are controlled by the reset and clock blocks (see the documentation of the respective blocks). 5.2 Operating mode Start-up All bits are reset in the design when a POR (power-on-reset) is active. RC is enabled, Xtal is disabled and the CPU is reset (pmaddr = 0000). If Port A is used to return from the sleep mode, all bits with resetcold don't change (see sleep mode). Reset All bits with resetsystem and resetpconf (if enabled) are reset. Clock configuration doesn't change except cpuck. The CPU is reset. Active mode This is the mode where the CPU and all peripherals can work and execute the embedded software. Standby mode Executing a HALT instruction moves the XE8805 into the Standby mode. The CPU is stopped, but the clocks remain active. Therefore, the enabled peripherals remain active e.g. for time keeping. A reset or an interrupt/event request (if enabled) cancels the standby mode. Sleep mode This is a very low-power mode because all circuit clocks and all peripherals are stopped. Only some service blocks remain active. No time-keeping is possible. Two instructions are necessary to move into sleep mode. First, the SleepEn (sleep enable) bit in RegSysCtrl has to be set to 1. The sleep mode can then be activated by setting the Sleep bit in RegSysReset to 1. There are three possibe ways to wake-up from the sleep mode: 1. The por (power-on-reset caused by a power-down followed by power-on). The RAM information is lost. 2. The padreset 3. The Port A reset combination (if the Port A is present in the product). See Port A documentation for more details. Note: If the Port A is used to return from the sleep mode, all bits with resetcold don't change (RegSysCtrl, RegSysReset (except bit sleep), EnExtClock and BiasRc in RegSysClock, RegSysRcTrim1 and RegSysRcTrim2). The SleepFlag bit in RegSysReset, reads back a 1 if the circuit was in sleep mode since the flag was last cleared (see reset block for more details). Note: For a lower power consumption, disable the BiasRc bit in RegSysClock before to going to sleep mode. The start-up time of the oscillator will then be longer however. Note: It is recommended to insert a NOP instruction after the instruction that sets the circuit in sleep mode because this instruction can be executed when the sleep mode is left using the resetfromportA. (c) Semtech 2006 www.semtech.com 5-2 XE8805/05A START-UP without condition por por RESET por padreset portA reset padreset portA reset watchdog reset por without condition padreset portA reset watchdog reset buserror reset Halt instruction ACTIVE Interrupt/event STAND-BY SLEEP set bit sleep normal mode low current very low current Figure 5-1. XE8805 operating modes. (c) Semtech 2006 www.semtech.com 5-3 XE8805/05A 6 Reset Block 6.1 Features 6-2 6.2 Overview 6-2 6.3 Register map 6-2 6.4 Reset handling capabilities 6-3 6.5 6.5.1 6.5.2 6.5.3 6.5.4 6.5.5 6.5.6 Reset source description Power On Reset RESET pin Programmable Port A input combination Watchdog reset BusError reset Sleep mode 6-4 6-4 6-4 6-4 6-4 6-4 6-4 6.6 Control register description and operation 6-4 6.7 Watchdog 6-5 6.8 Start-up and watchdog specifications 6-5 (c) Semtech 2006 www.semtech.com 6-1 XE8805/05A 6.1 * * * * * * Features Power On Reset (POR) External reset from the RESET pin Programmable Watchdog timer reset Programmable BusError reset Sleep mode management Programmable Port A input combination reset 6.2 Overview The reset block is the reset manager. It handles the different reset sources and distributes them through the system. It also controls the sleep mode of the circuit. 6.3 Register map Pos. 7 RegSysCtrl SleepEn Rw rw Reset 0 resetcold 6 EnResPConf rw 0 resetcold 5 EnBusError rw 0 resetcold 4 EnResWD rw 0 resetcold - r 0000 3-0 Function enables Sleep mode 0: sleep mode is disabled 1: sleep mode is enabled enables the resetpconf signal when the resetglobal is active 0: resetpconf is disabled 1: resetpconf is enabled enables reset from BusError 0: BusError reset source is disabled 1: BusError reset source is enabled enables reset from Watchdog 0: Watchdog reset source is disabled 1: Watchdog reset source is enabled this bit can not be set to 0 by SW unused Table 6-1. RegSysCtrl register. Pos. 7 6 5 4 3 2 1 0 RegSysReset Sleep ResetBusError ResetWD ResetfromportA ResPad ResPadSleep - Rw rw r rc rc rc rc rc r Reset 0 resetsystem 0 0 resetcold 0 resetcold 0 resetcold 0 resetcold 0 resetcold 0 Function Sleep mode control (reads always 0) unused reset source was BusError reset source was Watchdog reset source was Port A combination reset source was reset pad reset source was reset pad in sleep mode unused Table 6-2. RegSysReset register (c) Semtech 2006 www.semtech.com 6-2 XE8805/05A Pos. 7-4 3 2 1 0 RegSysWD WDKey[3] WDCounter[3] WDKey[2] WDCounter[2] WDKey[1] WDCounter[1] WDKey[0] WDCounter[0] Rw r w r w r w r w r Reset 0000 0 resetcold 0 resetcold 0 resetcold 0 resetcold Function unused Watchdog Key bit 3 Watchdog counter bit 3 Watchdog Key bit 2 Watchdog counter bit 2 Watchdog Key bit 1 Watchdog counter bit 1 Watchdog Key bit 0 Watchdog counter bit 0 Table 6-3. RegSysWD register 6.4 Reset handling capabilities There are 5 reset sources: * Power On Reset (POR) * External reset from the RESET pin * Programmable Port A input combination * Programmable watchdog timer reset * Programmable BusError reset on processor access outside the allocated memory map Another reset source is the bit Sleep in the RegSysReset register. This source is fully controlled by software and is only used during the sleep mode. Four internal reset signals are generated from these sources and distributed through the system: * resetcold: is asserted on POR * resetsystem: is asserted when resetcold or any other enabled reset source is active * resetpconf: is asserted when resetsystem is active and if the EnResPConf bit in the RegSysCtrl register is set. This reset is generally used in the different ports. It allows to maintain the port configuration unchanged while the rest of the circuit is reset. * resetsleep: is asserted when the circuit is in sleep mode For the circuits XE8801A and XE8805A (2) For the circuits XE8801 and XE8805 Table 6-4 shows a summary of the dependency of the internal reset signals on the various reset sources. In all the tables describing the different registers, the reset source is indicated. Internal reset signals Asserted reset source resetsystem POR RESET pin (1) RESET pin (2) PortA input Watchdog BusError Sleep Asserted Asserted Asserted Asserted Asserted Asserted - (1) resetpconf when when EnResPConf=0 EnResPConf=1 Asserted Asserted Asserted Asserted Asserted Asserted Asserted Asserted - resetsleep resetcold Asserted Asserted Asserted Asserted Asserted - For the circuits XE8801A and XE8805A (2) For the circuits XE8801 and XE8805 Table 6-4. Internal reset assertion as a function of the reset source. (c) Semtech 2006 www.semtech.com 6-3 XE8805/05A 6.5 6.5.1 Reset source description Power On Reset The power on reset (POR) monitors the external supply voltage. It activates a reset on a rising edge of this supply voltage. The reset is inactivated only if the internal voltage regulator has started up. No precise voltage level detection is performed by the POR block. 6.5.2 RESET pin The reset can be activated by applying a high input state on the RESET pin. 6.5.3 Programmable Port A input combination A reset signal can be generated by Port A. See the description of the Port A for further information. 6.5.4 Watchdog reset The Watchdog will generate a reset if the EnResetWD bit in the RegSysCtrl register has been set and if the watchdog is not cleared in time by the processor. See chapter 6.7 describing the watchdog for further information. 6.5.5 BusError reset The address space is assigned as shown in the register map of the product. If the EnBusError bit in the RegSysCtrl register is set and a non-existant address is accessed by the software, a reset is generated. 6.5.6 Sleep mode Entering the sleep mode will reset a part of the circuit. The reset is used to configure the circuit for correct wake-up after the sleep mode. If the SleepEn bit in the RegSysCtrl register has been set, the sleep mode can be entered by setting the bit Sleep in RegSysReset. During the sleep mode, the resetsleep signal is active. For detailed information on the sleep mode, see the system documentation. 6.6 Control register description and operation Two registers are dedicated for reset status and control, RegSysReset and RegSysCtrl. The bits Sleep, and SleepEn are also located in those registers and are described in the chapter dedicated to the different operating modes of the circuit (system block). The RegSysReset register gives information on the source which generated the last reset. It can be read at the beginning of the application program to detect if the circuit is recovering from an error or exception condition, or if the circuit is starting up normally. * when ResBusError is 1, a forbidden address access generated the reset. * when ResWD is 1, the watchdog generated the reset. * when ResPortA is 1, a PortA combination generated the reset. * when ResPad is 1, a reset pin generated the reset. * when ResPadSleep is 1, a reset pin in sleep mode generated the reset. Note: If no bit is set to 1, the reset source was the internal POR. Note: Several bits might be set or not, if the register was not cleared in between 2 reset occurrences. Write any value in RegSysReset to clear it. Note: When a reset pin wakes up the chip from the sleep mode, ResPad and ResPadSleep bits are equal at 1. The last bit concerns the sleep mode control (see system documentation for the sleep mode description). (c) Semtech 2006 www.semtech.com 6-4 XE8805/05A * when Sleep is set to 1, and SleepEn is 1, the sleep mode is entered. The bit always reads back a 0. The RegSysCtrl register enables the different available reset sources and the sleep mode. * EnResWD enables the reset due to the watchdog (can not be disabled once enabled). * EnBusError enables the reset due to a bus error condition. * EnResPConf enables the reset of the port configurations when reset by Port A, a Bus Error or the watchdog. * SleepEn unlocks the Sleep bit. As long as SleepEn is 0, the Sleep bit has no effect. 6.7 Watchdog The watchdog is a timer which has to be cleared at least every 2 seconds by the software to prevent a reset being generated by the timeout condition. The watchdog can be enabled by software by setting the EnResWD bit in the RegSysCtrl register to 1. It can then only be disabled by a power on reset. The watchdog timer can be cleared by writing consecutively the values Hx0A and Hx03 to the RegSysWD register. The sequence must strictly be respected to clear the watchdog. In assembler code, the sequence to clear the watchdog is: move AddrRegSysWD, #0x0A move AddrRegSysWD, #0x03 Only writing Hx0A followed by Hx03 resets the WD. If some other write instruction is done to the RegSysWD between the writing of the Hx0A and Hx03 values, the watchdog timer will not be cleared. It is possible to read the status of the watchdog in the RegSysWD register. The watchdog is a 4 bit counter with a count range between 0 and 7. The system reset is generated when the counter is reaching the value 8. 6.8 Start-up and watchdog specifications At start-up of the circuit, the POR (power-on-reset) block generates a reset signal during tPOR. The circuit starts software execution after this period (see system chapter). The POR is intended to force the circuit in a correct state at start-up. For precise monitoring of the supply voltage, the voltage level detector (VLD) has to be used. (c) Semtech 2006 www.semtech.com 6-5 XE8805/05A Symbol Parameter Min Typ Max Unit Comments TPOR POR reset duration 5 20 ms TRESET RESET pin reset duration 20 200 s 3 TRESET RESET pin reset duration 5 20 ms 4 Vbat_sl_M Supply ramp up of MTP version 20 V/ms 1 Vbat_sl_R Supply ramp up of ROM version 0.25 V/ms 1 2 s 2 WDtime Watchdog timeout period Table 6-5. Electrical and timing specifications Note: 1) The Vbat_sl defines the minimum slope required on VBAT. Correct start-up of the circuit is not guaranteed if this slope is too slow. In such a case, a delay has to be built using the RESET pin. Note: 2) The minimal watchdog timeout period is guaranteed when the internal oscillators are used. The watchdog takes its clock from the low prescaler. In case an external clock source is used, the RC oscillator must be enabled also (EnRC=1 in RegSysClock). Otherwise, the watchdog is stopped (see the clock block documentation). Note: 3) For the circuit versions XE8801 and XE8805. Gives the time the reset is active after the falling edge of the RESET pin. Note: 4) For the circuit versions XE8801A and XE8805A. Gives the time the reset is active after the falling edge of the RESET pin. (c) Semtech 2006 www.semtech.com 6-6 XE8805/05A 7 Clock Generator 7.1 Features 7-2 7.2 Overview 7-2 7.3 Register map 7-2 7.4 Interrupts and events map 7-3 7.5 7.5.1 7.5.2 7.5.3 Clock sources RC oscillator Xtal oscillator External clock 7-4 7-4 7-6 7-7 7.6 Clock source selection 7-8 7.7 RegSysMisc Description 7-8 7.8 Prescalers 7-9 7.9 32 kHz frequency selector 7-9 (c) Semtech 2006 www.semtech.com 7-1 XE8805/05A 7.1 Features * * * 7.2 3 available clock sources (RC oscillator, quartz oscillator and external clock). 2 divider chains: high-prescaler (8 bits) and low-prescaler (15 bits). CPU clock disabling in halt mode. Overview The XE8805 chips can work on different clock sources (RC oscillator, quartz oscillator and external clock). The clock generator block is in charge of distributing the necessary clock frequencies to the circuit. Figure 7-1 represents the functionality of the clock block. The internal RC oscillator drives the high prescaler. This prescaler generates frequency divisions down to 1/256 of its input frequency. A 32kHz clock is generated by enabling the quartz oscillator (if present in the product) or by selecting the appropriate tap on the high prescaler. The low prescaler generates clock signals from 32kHz down to 1Hz. The clock source for the CPU can be selected from the RC oscillator, the external clock or the 32kHz clock. 7.3 Register map pos. 7 RegSysClock CpuSel rw rw Reset 0 resetsleep 6 5 4 3 2 1 0 Extclk EnExtClock BiasRc ColdXtal ColdRC EnableXtal EnableRc r rw rw r r rw rw 0 resetcold 0 resetcold 1 resetcold 1 resetsleep 1 resetsleep 0 resetsleep 1 resetsleep function Select speed for cpuck, 0=RC, 1=xtal or external clock External clock detected, 1=available Enable for external clock, 1=enabled Enable Rcbias (reduces start-up time of RC). Xtal in start phase RC in start phase Enable Xtal oscillator, 0=disabled, 1=enabled Enable RC oscillator, 0=disabled, 1=enabled Table 7-1: RegSysClock register pos. 7-4 3 2 1 RegSysMisc -RCOnPA0 DebFast OutputCkXtal rw r rw rw rw Reset 0000 0 resetsleep 0 resetsleep 0 resetsleep 0 OutputCpuCk rw 0 resetsleep Function Unused Start RC on PA[0], 0=disabled, 1=enabled Debouncer clock speed, 0=256Hz, 1=8kHz Output Xtal Clock on PB[3], 0=disabled, 1=enabled if EnXtal=1 else PB[3]=0 Output CPU clock on PB[2], 0=disabled, 1=enabled Table 7-2: RegSysMisc register pos. 7-1 0 RegSysPre0 -ResPre rw r w1 r0 reset 0000000 0 Function Unused Write 1 to reset low prescaler, but always reads 0 Table 7-3: RegSysPre0 register (c) Semtech 2006 www.semtech.com 7-2 XE8805/05A pos. 7-4 5 4 3 2 1 0 RegSysRcTrim1 -Reserved RcFreqRange RcFreqCoarse[3] RcFreqCoarse[2] RcFreqCoarse[1] RcFreqCoarse[0] rw r rw rw rw rw rw rw reset 00 0 resetcold 0 resetcold 0 resetcold 0 resetcold 0 resetcold 0 resetcold Function Unused Reserved Low/high freq. range (low=0) RC coarse trim bit 3 RC coarse trim bit 2 RC coarse trim bit 1 RC coarse trim bit 0 Table 7-4: RegSysRCTrim1 register pos. 7-6 5 4 3 2 1 0 RegSysRcTrim2 -RcFreqFine[5] RcFreqFine[4] RcFreqFine[3] RcFreqFine[2] RcFreqFine[1] RcFreqFine[0] Rw r rw rw rw rw rw rw reset 00 1 resetcold 0 resetcold 0 resetcold 0 resetcold 0 resetcold 0 resetcold function Unused RC fine trim bit 5 RC fine trim bit 4 RC fine trim bit 3 RC fine trim bit 2 RC fine trim bit 1 RC fine trim bit 0 Table 7-5: RegSysRCTrim2 register RegSysRcTrim1 RegSysRcTrim2 CkRc to CkRc/256 CkRc High prescaler RC External Clock CkXtal OSCIN Xtal 0 0 1 1 CpuSel 0 Low prescaler 32kHz to 1Hz 1 EnXtal and not(ExtClk or EnExtClk) CpuCk Figure 7-1. Clock block structure 7.4 Interrupts and events map Interrupt IrqPre1 Interrupt source Ck128Hz IrqPre2 Ck1Hz Mapping in the interrupt manager RegIrqHig(6) Mapping in the event manager RegEvn(5) RegIrqMid(3) RegEvn(1) Table 7-6: Interrupts and events map (c) Semtech 2006 www.semtech.com 7-3 XE8805/05A 7.5 Clock sources 7.5.1 RC oscillator 7.5.1.1 Configuration The RC oscillator is always turned on and selected for CPU and system operation at power-on reset and when exiting sleep mode. It can be turned off after the Xtal (quartz oscillator) has been started, after selection of the external clock or by entering sleep mode. The RC oscillator has two frequency ranges: sub-MHz (100 kHz to 1 MHz) and above-MHz (1 MHz to 10 MHz). Inside a range, the frequency can be tuned by software for coarse and fine adjustment. See registers RegSysRcTrim1 and RegSysRcTrim2. Bit EnableRC in register RegSysClock controls the propagation of the RC clock signal and the operation of the oscillator. The user can stop the RC oscillator by resetting the bit EnableRC. Entering the sleep mode disables the RC oscillator. Note: Before turning off the RC oscillator, the cpusel bit in RegSysClock must be set to one. Note: The RC oscillator bias can be maintained while the oscillator is switched off by setting the bit BiasRc in RegSysClock. This allows a faster restart of the RC oscillator at the cost of increased power consumption when the oscillator is disabled (see section 7.5.1.3). 7.5.1.2 RC oscillator frequency tuning The RC oscillator frequency can be set using the bits in the RegSysRcTrim1 and RegSysRcTrim2 registers. Figure 7-2 shows the nominal frequency of the RC oscillator as a function of these bits. The absolute value of the frequency for a given register content may change by 50% from chip to chip due to the tolerances on the integrated capacitors and resistors. However, the modification of the frequency as a function of a modification of the register content is fairly precise for frequencies below 2MHz. This means that the curves in Figure 7-2 can shift up and down but that the slope remains unchanged. The bit RcFreqRange modifies the oscillator frequency by a factor of 10. The upper curve in the figure corresponds to RcFreqRange=1. The RcFreqCoarse modifies the frequency of the oscillator by a factor (RcFreqCoarse+1). The figure represents the frequency for 5 different values of the bits RcFreqCoarse: for each value the frequency is multiplied by 2. Incrementing the RcFreqFine code increases the frequency by about 1.4%. The frequency of the oscillator is therefor given by: fRC=fRcmin(1+9RcFreqRange)(1+RcFreqCoarse)(1.014)RcFreqFine with fRcmin the RC oscillator frequency if the registers are all 0. At higher frequencies, the frequency may deviate from the value predicted by the equation. (c) Semtech 2006 www.semtech.com 7-4 XE8805/05A 1E+08 11 11 11 00 00 11 00 00 01 11 11 11 00 00 11 00 00 10 00 00 01 00 00 11 00 00 01 11 11 11 00 00 10 00 00 10 00 00 11 00 00 01 11 11 11 00 00 10 00 00 00 00 00 01 00 00 11 00 00 10 00 00 01 11 11 11 00 00 11 00 00 01 11 11 11 00 00 11 00 00 10 11 11 11 00 00 10 00 00 01 11 11 11 00 00 11 00 00 10 11 11 11 00 00 11 00 00 10 00 00 01 1E+05 00 00 01 1E+06 11 11 11 00 00 11 RcFreqFine(5:0) 00 00 10 1E+07 00 00 00 Nominal RC oscillator frequency [Hz] RcFreqRange='1' RcFreqRange='0' RcFreqFine(5:0) RcFreqCoarse(3:0) 1E+04 0000 0001 0011 0111 1111 Figure 7-2. RC oscillator nominal frequency tuning. 7.5.1.3 RC oscillator specifications symbol fRCmin RcFreqFine RC_su description Lowest RC frequency fine tuning step startup time PSRR @ DC Supply voltage dependence Temperature dependence f/T min 40 typ 80 1.4 30 3 TBD TBD 0.1 max 120 2.0 50 5 unit kHz % us us %/V %/V %/C Comments Note 1 BiasRc=0 BiasRc=1 Note 2 Note 3 Table 7-7. RC oscillator specifications Note 1: this is the frequency tolerance when all trimming codes are 0. Note 2: frequency shift as a function of VBAT with normal regulator function. Note 3: frequency shift as a function of VBAT while the regulator is short-circuited to VBAT. The tolerances on the minimal frequency and the drift with supply or temperature can be cancelled using the software DFLL (digital frequency locked loop) which uses the crystal oscillator as a reference frequency. (c) Semtech 2006 www.semtech.com 7-5 XE8805/05A 7.5.2 Xtal oscillator 7.5.2.1 Xtal configuration The Xtal operates with an external crystal of 32'768 Hz. During Xtal oscillator start-up, the first 32768 cycles are masked. The two bits EnableXtal and ColdXtal in register RegSysClock control the oscillator. At power-on reset or during sleep mode, EnableXtal is reset and ColdXtal is set (Xtal oscillator is not selected at start-up). The user can start Xtal oscillator by setting EnableXtal. When the Xtal oscillator starts, bit ColdXtal is reset after 32768 cycles. Before ColdXtal is reset by the system, the Xtal frequency precision is not guaranteed. The Xtal oscillator can be stopped by the user by resetting bit EnableXtal. When the user enters into sleep mode, the Xtal is stopped. When an external clock is detected (ExtClk = 1) or the EnExtClock is set 1, the EnableXtal bit can not be set to 1. 7.5.2.2 Xtal oscillator specifications The crystal oscillator has been designed for a crystal with the specifications given in Table 7-8. The oscillator precision can only be guaranteed for this crystal. Symbol Fs CL Rm Cm C0 Rmp Q Description Resonance frequency CL for nominal frequency Motional resistance Motional capacitance Shunt capacitance Motional resistance of 6th overtone (parasitic) Quality factor Min Typ 32768 8.2 Max 1.8 0.7 4 40 2.5 1.1 8 100 3.2 2.0 k fF pF k 30k 50k 400k - 15 Unit Hz pF Comments Table 7-8. Crystal specifications. For safe operation, low power consumption and to meet the specified precision, careful board layout is required: Keep lines OSCIN and OSCOUT short and insert a VSS line in between them. Connect the crystal package to VSS. No noisy or digital lines near OSCIN or OSCOUT. Insert guards where needed. Respect the board specifications of Table 7-9. Symbol Rh_oscin Rh_oscout Rh_oscin_oscout Cp_oscin Cp_oscout Cp_oscin_oscout Description Resistance OSCIN-VSS Resistance OSCOUT-VSS Resistance OSCIN-OSCOUT Capacitance OSCIN-VSS Capacitance OSCOUT-VSS Capacitance OSCIN-OSCOUT Min 10 10 50 0.5 0.5 0.2 Typ Max 3.0 3.0 1.0 Unit M M M pF pF pF Comments Table 7-9. Board layout specifications. The oscillator characteristics are given in Table 7-10. The characteristics are valid only if the crystal and board layout meet the specifications above. (c) Semtech 2006 www.semtech.com 7-6 XE8805/05A Symbol fXtal St_xtal Fstab Description Nominal frequency Start-up time Frequency deviation Min Typ 32768 1 -100 Max 2 300 Unit Hz s ppm Comments Note 1 Table 7-10. Crystal oscillator characteristics. Note 1. This gives the relative frequency deviation from nominal for a crystal with CL=8.2pF and within the temperature range -40C to 85C. The crystal tolerance, crystal aging and crystal temperature drift are not included in this figure. 7.5.3 External clock 7.5.3.1 External clock configuration The user can provide an external clock instead of the internal oscillators. Only the CPU can use the external clock. The external clock input pin is OSCIN. The system is configured for external clock by bit EnExtClock in register RegSysClock. When EnExtClock is set to 1, the external clock is detected after 4 pulses on pin OSCIN. The ExtClk bit shows when the external clock is available. Note: when using the external clock, the Xtal is not available. 7.5.3.2 External clock specification The external clock has to satisfy the specifications in the table below. Correct behavior of the circuit can not be guaranteed if the external clock signal does not respect the specifications below. Symbol FEXT PW_1 PW_0 Description External frequency Pulse 1 width Pulse 0 width Min Typ clock Max 2 Unit MHz Comments s s 0.2 0.2 Table 7-11. External clock specifications. (c) Semtech 2006 www.semtech.com 7-7 XE8805/05A 7.6 Clock source selection There are three possible clock sources available for the CPU clock. The RC clock is always selected after powerup or after Sleep mode. The CPU clock selection is done with the bit CpuSel in RegSysClock (0= RC clock, 1= 32 kHz from Xtal if EnableXtal =1, ExtClk = 0 and EnExtClk = 0 else external clock). Switching from one clock source to another is glitch free. The next table summarizes the different clock configurations of the circuit: Clock Sources Clock targets EnableRc EnableXtal Cpuck Note 1 EnExtClk Mode name Sleep Xtal RC 0 0 0 0 0 1 0 1 0 Off Off RC Off Xtal RC Off Off RC Off Xtal High presc. RC + Xtal 0 1 1 RC Xtal RC Xtal External RC + External 1 0 X Off External Off Off 1 1 X RC External RC High presc. CpuSel=0 CpuSel=1 High Prescaler Clock input Low Prescaler Clock input Table 7-12: Table of clocking modes. Note 1: The CPU clock can be divided by using the freq instruction (see coolrisc instruction set) Switching from one clock source to another and stopping the unused clock source must be performed using 3 MOVE instructions to RegSysClock. First select the new clock source, secondly change the CpuSel bit and finally stop the unused one. 7.7 RegSysMisc Description The RCOnPA0 bit in RegSysMisc can be used to enable the RC oscillator on an event external to the circuit. If RCOnPA0 is 1, the RC oscillator is enabled (EnableRC bit is set to 1) as soon as the value on port A pin PA[0] is equal to 1. The port A pin can be debounced (see port A documentation). Bit DebFast in the RegSysMisc register allows to chose the debouncer clock between 256Hz and 8kHz (DebFast = 0 and DebFast = 1 respectively). The Debouncer clock is used to debounce PA inputs (see port A documentation). Bit OutputCkXtal allows to show the Xtal clock on PB[3]. The EnableXtal bit must be set to 1 else PB[3] equals 0 (see port B documentation to set up the port B). Bit OutputCpuCk allows to show the CpuClock on PB[2] (see port B documentation). (c) Semtech 2006 www.semtech.com 7-8 XE8805/05A 7.8 Prescalers The clock generator block embeds two divider chains: the high prescaler and the low prescaler. The high prescaler is made of an 8 stage dividing chain and the low prescaler of a 15 stage dividing chain. Features: * High prescaler can only be driven with RC clock (bit EnableRc have to be set, see Table 7-12). * Low prescaler can be driven from the high prescaler or directly with the Xtal clock when bit EnableXtal is set to 1, bit EnExtClock is set to 0 and ExtClk is equal at 0. * Bit ResPre in the RegSysPre0 register allows to reset synchronously the low prescaler, the low prescaler is also automatically cleared when bit EnableXtal is set. Both dividing chains are reset asynchronously by the resetsleep signal. * Bit ColdXtal=1 indicates the Xtal is in its start up phase. It is active for 37268 Xtal cycles after setting EnableXtal. 7.9 32 kHz frequency selector A decoder is used to select from the high prescaler the frequency tap that is the closest to 32 kHz to operate the low prescaler when the Xtal is not running. In this case, the RC oscillator frequency of 50% will also be valid for the low prescaler frequency outputs. (c) Semtech 2006 www.semtech.com 7-9 XE8805/05A 8 IRQ - Interrupt handler 8.1 Features 8-2 8.2 Overview 8-2 8.3 Register map 8-2 (c) Semtech 2006 www.semtech.com 8-1 XE8805/05A 8.1 Features The XE8000 chips support 24 interrupt sources, divided into 3 levels of priority. 8.2 Overview A CPU interruption is generated and memorized when an interrupt becomes active. The 24 interrupt sources are divided into 3 levels of priority: High (8 interrupt sources), Mid (8 interrupt sources), and Low (8 interrupt sources). Those 3 levels of priority are directly mapped to those supported by the CoolRisc(R) (IN0, IN1 and IN2; see CoolRisc(R) documentation for more information). RegIrqHig, RegIrqMid, and RegIrqLow are 8-bit registers containing flags for the interrupt sources. Those flags are set when the interrupt is enabled (i.e. if the corresponding bit in the registers RegIrqEnHig, RegIrqEnMid or RegIrqEnLow is set) and a rising edge is detected on the corresponding interrupt source. Once memorized, an interrupt flag can be cleared by writing a `1' in the corresponding bit of RegIrqHig, RegIrqMid or RegIrqLow. Writing a `0' does not modify the flag. To definitively clear the interrupt, one has to clear the CoolRisc interrupt in the CoolRisc status register. All interrupts are automatically cleared after a reset. Two registers are provided to facilitate the writing of interrupt service software. RegIrqPriority contains the number of the highest priority interrupt set (its value is 0xFF when no interrupt is set). RegIrqIrq indicates the priority level of the current interrupts. RegIrqIrq and RegIrqPriority `s values are dependent upon the memorized state of the interrupts (as reflected in flags in RegIrqHig, RegIrqMid and RegIrqLow). 8.3 Register map pos. 7 RegIrqHig RegIrqHig[7] 6 RegIrqHig[6] 5 RegIrqHig[5] 4 RegIrqHig[4] 3 RegIrqHig[3] 2 RegIrqHig[2] 1 RegIrqHig[1] 0 RegIrqHig[0] rw r c1 r c1 r c1 r c1 r c1 r c1 r c1 r c1 reset 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem function interrupt #23 (high priority) clear interrupt #23 when 1 is written interrupt #22 (high priority) clear interrupt #22 when 1 is written interrupt #21 (high priority) clear interrupt #21 when 1 is written interrupt #20 (high priority) clear interrupt #20 when 1 is written interrupt #19 (high priority) clear interrupt #19 when 1 is written interrupt #18 (high priority) clear interrupt #18 when 1 is written interrupt #17 (high priority) clear interrupt #17 when 1 is written interrupt #16 (high priority) clear interrupt #16 when 1 is written Table 8-1: RegIrqHig (c) Semtech 2006 www.semtech.com 8-2 XE8805/05A pos. 7 RegIrqMid RegIrqMid[7] 6 RegIrqMid[6] 5 RegIrqMid[5] 4 RegIrqMid[4] 3 RegIrqMid[3] 2 RegIrqMid[2] 1 RegIrqMid[1] 0 RegIrqMid[0] rw r c1 r c1 r c1 r c1 r c1 r c1 r c1 r c1 reset 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem function interrupt #15 (mid priority) clear interrupt #15 when 1 is written interrupt #14 (mid priority) clear interrupt #14 when 1 is written interrupt #13 (mid priority) clear interrupt #13 when 1 is written interrupt #12 (mid priority) clear interrupt #12 when 1 is written interrupt #11 (mid priority) clear interrupt #11 when 1 is written interrupt #10 (mid priority) clear interrupt #10 when 1 is written interrupt #9 (mid priority) clear interrupt #9 when 1 is written interrupt #8 (mid priority) clear interrupt #8 when 1 is written Table 8-2: RegIrqMid pos. 7 RegIrqLow RegIrqLow[7] 6 RegIrqLow[6] 5 RegIrqLow[5] 4 RegIrqLow[4] 3 RegIrqLow[3] 2 RegIrqLow[2] 1 RegIrqLow[1] 0 RegIrqLow[0] rw r c1 r c1 r c1 r c1 r c1 r c1 r c1 r c1 reset 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem function interrupt #7 (low priority) clear interrupt #7 when 1 is written interrupt #6 (low priority) clear interrupt #6 when 1 is written interrupt #5 (low priority) clear interrupt #5 when 1 is written interrupt #4 (low priority) clear interrupt #4 when 1 is written interrupt #3 (low priority) clear interrupt #3 when 1 is written interrupt #2 (low priority) clear interrupt #2 when 1 is written interrupt #1 (low priority) clear interrupt #1 when 1 is written interrupt #0 (low priority) clear interrupt #0 when 1 is written Table 8-3: RegIrqLow pos. 7 6 5 4 3 2 1 0 RegIrqEnHig RegIrqEnHig[7] RegIrqEnHig[6] RegIrqEnHig[5] RegIrqEnHig[4] RegIrqEnHig[3] RegIrqEnHig[2] RegIrqEnHig[1] RegIrqEnHig[0] rw rw rw rw rw rw rw rw rw reset 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem function 1= enable interrupt #23 1= enable interrupt #22 1= enable interrupt #21 1= enable interrupt #20 1= enable interrupt #19 1= enable interrupt #18 1= enable interrupt #17 1= enable interrupt #16 Table 8-4: RegIrqEnHig (c) Semtech 2006 www.semtech.com 8-3 XE8805/05A pos. 7 6 5 4 3 2 1 0 RegIrqEnMid RegIrqEnMid[7] RegIrqEnMid[6] RegIrqEnMid[5] RegIrqEnMid[4] RegIrqEnMid[3] RegIrqEnMid[2] RegIrqEnMid[1] RegIrqEnMid[0] rw rw rw rw rw rw rw rw rw reset 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem function 1= enable interrupt #15 1= enable interrupt #14 1= enable interrupt #13 1= enable interrupt #12 1= enable interrupt #11 1= enable interrupt #10 1= enable interrupt #9 1= enable interrupt #8 Table 8-5: RegIrqEnMid pos. 7 6 5 4 3 2 1 0 RegIrqEnLow RegIrqEnLow[7] RegIrqEnLow[6] RegIrqEnLow[5] RegIrqEnLow[4] RegIrqEnLow[3] RegIrqEnLow[2] RegIrqEnLow[1] RegIrqEnLow[0] rw rw rw rw rw rw rw rw rw reset 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem function 1= enable interrupt #7 1= enable interrupt #6 1= enable interrupt #5 1= enable interrupt #4 1= enable interrupt #3 1= enable interrupt #2 1= enable interrupt #1 1= enable interrupt #0 Table 8-6: RegIrqEnLow pos. 7-0 RegIrqPriority RegIrqPriority rw r reset 11111111 resetsystem function code of highest priority set Table 8-7: RegIrqPriority pos. 7-3 2 RegIrqIrq IrqHig rw r r Reset 00000 0 resetsystem 1 IrqMid r 0 resetsystem 0 IrqLow r 0 resetsystem function unused one or more high priority interrupt is set one or more mid priority interrupt is set one or more low priority interrupt is set Table 8-8: RegIrqIrq (c) Semtech 2006 www.semtech.com 8-4 XE8805/05A 9 Event handler 9.1 Features 9-2 9.2 Overview 9-2 9.3 Register map 9-2 (c) Semtech 2006 www.semtech.com 9-1 XE8805/05A 9.1 Features The XE8000 chips support 8 event sources, divided into 2 levels of priority. 9.2 Overview A CPU event is generated and memorized when an event source becomes active. The 8 event sources are divided into 2 levels of priority: High (4 event sources) and Low (4 event sources). Those 2 levels of priority are directly mapped to those supported by the CoolRisc (EV0 and EV1; see CoolRisc documentation for more information). RegEvn is an 8-bit register containing flags for the event sources. Those flags are set when the event is enabled (i.e. if the corresponding bit in the registers RegEvnEn is set) and a rising edge is detected on the corresponding event source. Once memorized, writing a `1' in the corresponding bit of RegEvn clears an event flag. Writing a `0' does not modify the flag. All interrupts are automatically cleared after a reset. Two registers are provided to facilitate the writing of event service software. RegEvnPriority contains the number of the highest priority event set (its value is 0xFF when no event is set). RegEvnEvn indicates the priority level of the current interrupts. RegEvnEvn and RegEvnPriority `s values are dependent upon the memorized state of the events (as reflected in flags in RegEvn). 9.3 Register map pos. 7 RegEvn RegEvn[7] 6 RegEvn[6] 5 RegEvn[5] 4 RegEvn[4] 3 RegEvn[3] 2 RegEvn[2] 1 RegEvn[1] 0 RegEvn[0] rw r c1 r c1 r c1 r c1 r c1 r c1 r c1 r c1 reset 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem function event #7 (high priority) clear event #7 when written 1 event #6 (high priority) clear event #6 when written 1 event #5 (high priority) clear event #5 when written 1 event #4 (high priority) clear event #4 when written 1 event #3 (low priority) clear event #3 when written 1 event #2 (low priority) clear event #2 when written 1 event #1 (low priority) clear event #1 when written 1 event #0 (low priority) clear event #0 when written 1 Table 9-1: RegEvn (c) Semtech 2006 www.semtech.com 9-2 XE8805/05A pos. 7 6 5 4 3 2 1 0 RegEvnEn RegEvnEn[7] RegEvnEn[6] RegEvnEn[5] RegEvnEn[4] RegEvnEn[3] RegEvnEn[2] RegEvnEn[1] RegEvnEn[0] rw rw rw rw rw rw rw rw rw reset 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem function 1= enable event #7 1= enable event #6 1= enable event #5 1= enable event #4 1= enable event #3 1= enable event #2 1= enable event #1 1= enable event #0 Table 9-2: RegEvnEn pos. 7-0 RegEvnPriority RegEvnPriority rw r reset 11111111 resetsystem function code of highest event set Table 9-3: RegEvnPriority pos. 7-2 1 0 RegEvnEvn EvnHig EvnLow rw r r r reset 00000 0 resetsystem 0 resetsystem function unused one or more high priority event is set one or more low priority event is set Table 9-4: RegEvnEvn (c) Semtech 2006 www.semtech.com 9-3 XE8805/05A 10 Low power RAM 10.1 Features 10-2 10.2 Overview 10-2 10.3 Register map 10-2 (c) Semtech 2006 www.semtech.com 10-1 XE8805/05A 10.1 Features * Low power RAM locations. 10.2 Overview In order to save power consumption, 8 8-bit registers are provided in page 0. These memory locations should be reserved for often-updated variables. Accessing these register locations requires much less power than the other general purpose RAM locations. 10.3 Register map pos. 7-0 Reg00 Reg00 rw rw reset XXXXXXXX function low-power data memory Table 10-1: Reg00 pos. 7-0 Reg01 Reg01 rw rw reset XXXXXXXX function low-power data memory Table 10-2: Reg01 pos. 7-0 Reg02 Reg02 rw rw reset XXXXXXXX function low-power data memory Table 10-3: Reg02 pos. 7-0 Reg03 Reg03 rw rw reset XXXXXXXX function low-power data memory Table 10-4: Reg03 pos. 7-0 Reg04 Reg04 rw rw reset XXXXXXXX function low-power data memory Table 10-5: Reg04 pos. 7-0 Reg05 Reg05 rw rw reset XXXXXXXX function low-power data memory Table 10-6: Reg05 pos. 7-0 Reg06 Reg06 rw rw reset XXXXXXXX function low-power data memory Table 10-7: Reg06 pos. 7-0 Reg07 Reg07 rw rw reset XXXXXXXX function low-power data memory Table 10-8: Reg07 (c) Semtech 2006 www.semtech.com 10-2 XE8805/05A 11 Port A 11.1 Features 11-2 11.2 Overview 11-2 11.3 Register map 11-3 11.4 Interrupts and events map 11-3 11.5 Port A (PA) Operation 11-4 11.6 Port A electrical specification 11-5 (c) Semtech 2006 www.semtech.com 11-1 XE8805/05A 11.1 Features * * * * * * * * Input port, 8 bits wide Each bit can be set individually for debounced or direct input Each bit can be set individually for pull-up or not Each bit is an interrupt request source on the rising or falling edge A system reset can be generated on an input pattern PA[0] and PA[1] can generate two events for the CPU, individually maskable PA[0] to PA[3] can be used as clock inputs for the counters/timers/PWM (product dependent) PA[0] can be used to enable the RC oscillator 11.2 Overview Port A is a general purpose 8 bit wide digital input port, with interrupt capability. Figure 11-1 shows its structure. Port A VBat 8 RegPAPullup 8 RegPADebounce 8x debounce 8 0 RegPACtrl 8 RegPAIn 8 1 RegPAEdge 1 0 RC 1 DebFast (RegSysMisc(2)) 8 interrupts 0 events 8x cntclocks 256 Hz 8 kHz 8 8 RegPARes1 RegPARes0 1 11 PAReset[x] 10 01 0 resetfromporta 00 8x Figure 11-1:structure of Port A (c) Semtech 2006 www.semtech.com 11-2 XE8805/05A 11.3 Register map There are six registers in the Port A (PA), namely RegPAIn, RegPADebounce, RegPAEdge, RegPAPullup, RegPARes0 and RegPARes1. Table 11-1 to Table 11-6 show the mapping of control bits and functionality. pos. 7:0 RegPAIn PAIn[7:0] rw r reset description pad PA[7] to PA[0] input value Table 11-1: RegPAIn pos. 7:0 RegPADebounce rw PADebounce[7:0] rw reset description PA[7] to PA[0] 1: debounce enabled 0: debounce disabled 00000000 resetpconf Table 11-2: RegPADebounce pos. 7:0 RegPAEdge PAEdge[7:0] rw rw reset description PA[7] to PA[0] edge configuration 0: positive edge 1: negative edge 00000000 resetsystem Table 11-3: RegPAEdge pos. 7:0 RegPAPullup PAPullup[7:0] rw rw reset description PA[7] to PA[0] pullup enable 0: pullup disabled 1: pullup enabled 00000000 resetpconf Table 11-4: RegPAPullup pos. RegPARes0 7:0 PARes0[7:0] rw rw Reset 00000000 resetsystem description PA[7] to PA[0] reset configuration Table 11-5: RegPARes0 pos. 7:0 RegPARes1 PARes1[7:0] rw rw reset 00000000 resetsystem Description PA[7] to PA[0] reset configuration Table 11-6: RegPARes 11.4 Interrupts and events map Interrupt source pa_irqbus[5] pa_irqbus[4] pa_irqbus[1] pa_irqbus[0] pa_irqbus[7] pa_irqbus[6] pa_irqbus[3] pa_irqbus[2] Default mapping in the interrupt manager RegIrqMid[5] RegIrqMid[4] RegIrqMid[1] RegIrqMid[0] RegIrqLow[7] RegIrqLow[6] RegIrqLow[3] RegIrqLow[2] Default mapping in the event manager RegEvn[4] RegEvn[0] (c) Semtech 2006 www.semtech.com 11-3 XE8805/05A 11.5 Port A (PA) Operation The Port A input status (debounced or not) can be read from RegPAin. Debounce mode: Each bit in Port A can be individually debounced by setting the corresponding bit in RegPADebounce. After reset, the debounce function is disabled. After enabling the debouncer, the change of the input value is accepted only if the input value is identical at two consecutive sampling on the rising edge of the selected clock. Selection of the clock is done by the bit DebFast in Register RegSysMisc (see clock block documentation for more precision on the frequency). DebFast Clock filter 0 256 Hz 1 8 kHz Table 11-7: debounce frequency selection Input CkDebounce 1 1 2 1 1 2 Debounced Figure 11-2: digital debouncer Pull-ups: When the corresponding bit in RegPAPullup is set to 0, the inputs are floating (pull-up resistors are disconnected). When the corresponding bit in RegPAPullup is set to 1, a pull-up resistor is connected to the input pin. Port A starts up with the pull-up resistors disconnected. Port A as an interrupt source: Each Port A input is an interrupt request source and can be set on rising or falling edge with the corresponding bit in RegPAEdge. After reset, the rising edge is selected for interrupt generation by default. The interrupt source can be debounced by setting register RegPADebounce. Note: care must be taken when modifying RegPAEdge because this register performs an edge selection. The change of this register may result in a transition which may be interpreted as a valid interruption. Port A as an event source: The interrupt signals of the pins PA[0] and PA[1] are also available as events on the event controller. Port A as a clock source (product dependent): Images of the PA[0] to PA[3] input ports (debounced or not) are available as clock sources for the counter/timer/PWM peripheral (see the counter block documentation for more information). Port A as a reset source: Port A can be used to generate a system reset by placing a predetermined word on Port A externally. The reset is built using a logical and of the 8 PARes[x] signals: resetfromportA = PAReset[7] AND PAReset[6] AND PAReset[5] AND ... AND PAReset[0] PAReset[x] is itself a logical function of the corresponding pin PA[x]. One of four logical functions can be selected for each pin by writing into two registers RegPARes0 and RegPARes1 as shown in Table 11-8. (c) Semtech 2006 www.semtech.com 11-4 XE8805/05A PARes1[x] 0 0 1 1 PARes0[x] 0 1 0 1 PAReset[x] 0 PA[x] not(PA[x]) 1 Table 11-8: Selection bits for reset signal A reset from Port A can be inhibited by placing a 0 on both PARes1[x] and PARes0[x] for at least 1 pin. Setting both PARes1[x] and PARes0[x] to 1, makes the reset independent of the value on the corresponding pin. Setting both registers to hFF, will reset the circuit independent from the Port A input value. This makes it possible to do a software reset. Note: depending of the value of PA[0] to PA[7], the change of RegPARes0 and RegPARes1 can cause a reset. Therefore it is safe to have always one (RegPARes0[x], RegPARes1[x]) equal to `00' during the setting operations. Port A as a RC enable: PA[0] can be used to enable the RC oscillator. When RCOnPA0 bit in RegSysMisc is set to 1 and the value of PA[0] (debounced or not) is equal to 1, the EnRc bit in RegSysClock is automatically set to 1. 11.6 Port A electrical specification Sym VINH VINL RPU Cin description Input high voltage Input low voltage Pull-up resistance Input capacitance min 0.7*VBAT VSS 20 typ 50 3.5 max VBAT 0.2*VBAT 80 unit Comments V VBAT2.4V V VBAT2.4V k pF Note 1 Note 1: this value is indicative only since it depends on the package. Table 11-9. Port A electrical specification. (c) Semtech 2006 www.semtech.com 11-5 XE8805/05A 12 Port B 12.1 Features 12-2 12.2 Overview 12-2 12.3 Register map 12-2 12.4 Port B capabilities 12-3 12.5 12.5.1 12.5.2 Port B analog capability Port B analog configuration Port B analog function specification 12-3 12-3 12-4 12.6 Port B function capability 12-4 12.7 12.7.1 12.7.2 Port B digital capabilities Port B digital configuration Port B digital function specification 12-5 12-5 12-6 (c) Semtech 2006 www.semtech.com 12-1 XE8805/05A 12.1 Features * Input / output / analog port, 8 bits wide * Each bit can be set individually for input or output * Each bit can be set individually for open-drain or push-pull * Each bit can be set individually for pull-up or not (for input or open-drain mode) * In open-drain mode, pull-up is not active when corresponding pad is set to zero * The 8 pads can be connected by pairs to four internal analog lines (4 line analog bus) * Two internal freq. (cpuck and 32 kHz) can be output on PB[2] and PB[3] Product dependant: * Two PWM signal can be output on pads PB[0] and PB[1] * The synchronous serial interface (USRT) uses pads PB[5], PB[4] * The UART interface uses pads PB[6] and PB[7] for Tx and Rx 12.2 Overview Port B is a multi-purpose 8 bit Input/output port. In addition to digital functions, all pins can be used for analog signals. All port terminals can be selected by pairs as digital input or output or as analog to share one of four possible analog lines. 12.3 Register map Pos. 7-0 RegPBOut PBOut[7-0] rw rw reset description in digital mode 0 resetpconf Pad PB[7-0] output value description in analog mode Analog bus selection for pad PB[7-0] Table 12-1: RegPBOut RegPBIn Pos. 7-0 PBIn[7-0] rw reset description in digital mode rw Pad PB[7-0] input status description in analog mode Unused Table 12-2: RegPBIn RegPBDir Pos. 7-0 PBDir [7-0] rw rw reset description in digital mode 0 resetpconf Pad PB[7-0] direction (0=input) description in analog mode Analog bus selection for pad PB[7-0] Table 12-3: RegPBDir Pos. 7-0 RegPBOpen PBOpen[7-0] rw rw reset description in digital mode 0 resetpconf Pad PB[7-0] open drain (1 = open drain) description in analog mode Unused Table 12-4: RegPBOpen Pos. 7 -0 RegPBPullup PBPullup[7] rw rw reset description in digital mode 0 resetpconf Pull-up for pad PB[7-0] (1=active) description in analog mode Connect pad PB[7-0] on selected ana bus Table 12-5: RegPBPullup Pos. 7-4 3 2 1 0 RegPBAna -PBAna [3] PBAna [2] PBAna [1] PBAna [0] rw r rw rw rw rw reset 0000 0 resetpconf 0 resetpconf 0 resetpconf 0 resetpconf description in digital mode Unused Set PB[7:6] in analog mode Set PB[5:4] in analog mode Set PB[3:2] in analog mode Set PB[1:0] in analog mode description in analog mode Unused Set PB[7:6] in analog mode Set PB[5:4] in analog mode Set PB[3:2] in analog mode Set PB[1:0] in analog mode Table 12-6: RegPBAna (c) Semtech 2006 www.semtech.com 12-2 XE8805/05A Note: Depending on the status of the EnResPConf bit in RegSysCtrl, the reset conditions of the registers are different. See the reset block documentation for more details on the resetpconf signal. 12.4 Port B capabilities Port B name PB[7] PB[6] PB[5] PB[4] PB[3] PB[2] PB[1] PB[0] usage (priority) functions (medium) uart Rx uart Tx usrt S1 usrt S0 32 kHz clock CPU PWM1 Counter C (C+D) PWM0 Counter A (A+B) analog (high) analog analog analog analog digital (low) (default) I/O I/O I/O I/O I/O I/O I/O I/O Table 12-7: Different Port B functionality Table 12-7 shows the different usage that can be made of the port B with the order of priority. If a pair of pins is selected to be analog, it overwrites the function and digital set-up. If the pin is not selected as analog, but a function is enabled, it overwrites the digital set-up. If neither the analog nor function are selected for a pin, it is used as an ordinary digital I/O. This is the default configuration at start-up. 12.5 Port B analog capability 12.5.1 Port B analog configuration Port B terminals can be attached to a 4 line analog bus by setting the PBAna[x] bits to 1 in the RegPBAna register. The other registers then define the connection of these 4 analog lines to the different pads of Port B. This can be used to implement a simple LCD driver or A/D converter. Analog switching is available only when the circuit is powered with sufficient voltage (see specification below). Below the specified supply voltage, only voltages that are close to VSS or VBAT can be switched. When PBAna[x] is set to 1, a pair of Port B terminals is switched from digital I/O mode to analog mode. The usage of the registers RegPBPullup, RegPBOut and RegPBDir define the analog configuration (see Table 12-8). When PBAna[x] = 1, then PBPullup[x] connects the pin to the analog bus. PBDir[x] and PBPOut[x] select which of the 4 analog lines is used. For odd values of x, the selection bits are in the register RegPBOut (see Table 12-8). For even values of x, the selection bits are in the register RegPBDir (see Table 12-9). if x is odd, PBOut[x, x-1] PBPullup[x] PB[x] selection on 00 01 10 11 XX 1 1 1 1 0 analog line 0 analog line 1 analog line 2 analog line 3 High impedance Table 12-8: Selection of the analog lines for PB[x] when x is odd and PBAna[x] = 1 (c) Semtech 2006 www.semtech.com 12-3 XE8805/05A if x is even, PBDir[x+1, x] PBPullup[x] PB[x] selection on 00 01 10 11 XX 1 1 1 1 0 analog line 0 analog line 1 analog line 2 analog line 3 High impedance Table 12-9: Selection of the analog lines for PB[x] when x is even and PBAna[x] = 1 Example: Set the pads PB[2] and PB[3] on the analog line 3. (the values X depend on the configuration of others pads) - apply high impedance in the analog mode (move RegPBPullup,#0bXXXX00XX) - go to analog mode (move RegPBAna,#0bXXXXXX1X) - select the analog line3 (move RegPBDir,#0bXXXX11XX and move RegPBOut,#0bXXXX11XX) - connect the analog line to the pins (move RegPBPullup,#0bXXXX11XX) 12.5.2 Port B analog function specification The table below defines the on-resistance of the switches between the pin and the analog bus for different conditions. The series resistance between 2 pins of Port B connected to the same analog line is twice the resistance given in the table. sym Ron Ron Cin Cin description switch resistance switch resistance input capacitance (off) input capacitance (on) min typ 3.5 4.5 max 11 15 unit k k pF pF Comments Note 1 Note 2 Note 3 Note 4 Table 12-10. Analog input specifications. Note 1: This is the series resistance between the pad and the analog line in 2 cases 1. VBAT 2.4V and the VMULT peripheral is present on the circuit and enabled. 2. VBAT 3.0V and the VMULT peripheral is not present on the circuit. Note 2: This is the series resistance in case VBAT 2.8V and the peripheral VMULT is not present on the circuit. Note 3: This is the input capacitance seen on the pin when the pin is not connected to an analog line. This value is indicative only since it is product and package dependent. Note 4: This is the input capacitance seen on the pin when the pin is connected to an analog line and no other pin is connected to the same analog line. This value is indicative only since it is product and package dependent. 12.6 Port B function capability The Port B can be used for different functions implemented by other peripherals. The description below is applicable only in so far the circuit contains these peripherals. When the counters are used to implement a PWM function (see the documentation of the counters), the PB[0] and PB[1] terminals are used as outputs (PB[0] is used if CntPWM0 in RegCntConfig1 is set to 1, PB[1] is used if CntPWM1 in RegCntConfig1 is set to 1) and the PWM generated values overwrite the values written in RegPBout. However, PBDir(0) and PBDir(1) are not automatically overwritten and have to be set to 1. If OutputCkXtal is set in RegSysMisc, the Xtal clock is output on PB[3] (EnableXtal in RegSysClock must be set to 1). This overrides the value contained in PBOut(3). However, PBDir(3) must be set to 1. The duty cycle of the clock signal is about 50%. Similarly, if OutputCkCpu is set in RegSysMisc, the CPU frequency is output on PB[2]. This overrides the value contained in PBOut(2). However, PBDir(2) must be set to 1. (c) Semtech 2006 www.semtech.com 12-4 XE8805/05A The frequency of the CPU clock depends on the selection of the CpuSel bit in the RegSysClock register (see clock_gen_ff). Pins PB[5] and PB[4] can be used for S1 and S0 of the USRT (see USRT documentation) when the UsrtEnable bit is set in RegUsrtCtrl. The PB[5] and PB[4] then become open-drain. This overrides the values contained in PBOpen(5:4), PBOut(5:4) and PBDir(5:4). If there is no external pull-up resistor on these pins, internal pull-ups should be selected by setting PBPullup(5:4). When S0 is an output, the pin PB[4] takes the value of UsrtS0 in RegUrstS0. When S1 is an output, the pin PB[5] takes the value of UsrtS1 in RegUrstS1. Pins PB[6] and PB[7] can be used by the UART (see UART documentation). When UartEnTx in RegUartCtrl is set to 1, PB[6] is used as output signal Tx. When UartEnRx in RegUartCtrl is set to 1, PB[7] is used as input signal Rx. This overrides the values contained in PBOut(7:6) and PBDir(7:6). 12.7 Port B digital capabilities 12.7.1 Port B digital configuration The direction of each bit within Port B (input only or input/output) can be individually set using the RegPBDir register. If PBDir[x] = 1, both the input and output buffer are active on the corresponding Port B. If PBDir[x] = 0, the corresponding Port B pin is an input only and the output buffer is in high impedance. After reset (resetpconf) Port B is in input only mode (PBDir[x] are reset to 0). The input values of Port B are available in RegPBIn (read only). Reading is always direct - there is no debounce function in Port B. In case of possible noise on input signals, a software debouncer with polling or an external hardware filter have to be realized. The input buffer is also active when the port is defined as output and allows to read back the effective value on the pin. Data stored in RegPBOut are output at Port B if PBDir[x] is 1. The default value after reset is low (0). When a pin is in output mode (PBDir[x] is set to 1), the output can be a conventional CMOS (Push-Pull) or a Nchannel Open-drain, driving the output only low. By default, after reset (resetpconf) the PBOpen[x] in RegPBOpen is cleared to 0 (push-pull). If PBOpen[x] in RegPBOpen is set to 1 then the internal P transistor in the output buffer is electrically removed and the output can only be driven low (PBOut[x]=0). When PBOut[x]=1, the pin is high Impedance. The internal pull-up or an external pull-up resistor can be used to drive the pin high. Note: Because the P transistor actually exists (this is not a real Open-drain output) the pull-up range is limited to VDD + 0.2V (avoid forward bias the P transistor / diode). Each bit can be set individually for pull-up or not using register RegPBPullup. Input is pulled up when its corresponding bit in this register is set to 1. Default status after (resetpconf) is 0, which means without pull up. To limit power consumption, pull-up resistors are only enabled when the associated pin is either a digital input or an Nchannel open-drain output with the pad set to 1. In the other cases (push-pull output or open-drain output driven low), the pull up resistors are disabled independent of the value in RegPBPullup. After power-on reset, the Port B is configured as an input port without pull-up. The input buffer is always active, except in analog mode. This means that the Port B input should be a valid digital value at all times unless the pin is set in analog mode. Violating this rule may lead to higher power consumption. (c) Semtech 2006 www.semtech.com 12-5 XE8805/05A 12.7.2 Sym VINH VINL VOH Port B digital function specification description Input high voltage Input low voltage Output high voltage min 0.7*VBAT VSS VBAT-0.4 VOL Output low voltage VSS RPU Cin Pull-up resistance Input capacitance 20 typ max VBAT 0.2*VBAT VBAT VSS+0.4 50 3.5 80 unit Comments V VBAT2.4V V VBAT2.4V V VBAT=1.2V, IOH =0.3mA VBAT=2.4V, IOH =5.0mA VBAT=4.5V, IOH =8.0mA V VBAT=1.2V, IOL =0.3mA VBAT=2.4V, IOL =12.0mA VBAT=4.5V, IOL =15.0mA k pF Note 1 Note 1: this value is indicative only since it depends on the package. (c) Semtech 2006 www.semtech.com 12-6 XE8805/05A 13 Port C 13.1 Features 13-2 13.2 Overview 13-2 13.3 Port C (PC) Operation 13-2 13.4 Register map 13-2 13.5 Port C electrical specification 13-3 (c) Semtech 2006 www.semtech.com 13-1 XE8805/05A 13.1 Features * * Input / output port, 8 bits wide Each bit can be set individually for input or output 13.2 Overview Port C (PC) is a general purpose 8 bit input/output digital port. Figure 13-1 shows its structure. Port C 8 RegPCOut 8 RegPCDir 8 RegPCIn Figure 13-1: structure of Port C 13.3 Port C (PC) Operation The direction of each bit within Port C (input or output) can be individually set by using the RegPCDir register. If PCDir[x] = 1, the corresponding Port C pin becomes an output. After reset, Port C is in input mode (PCDir[x] are reset to 0). Output mode: Data is stored in RegPCOut prior to output at Port C. Input mode: The status of Port C is available in RegPCIn (read only). Reading is always direct - there is no digital debounce function associated with Port C. In case of possible noise on input signals, a software debouncer or an external filter must be realized. By default after reset, Port C is configured as an input port. 13.4 Register map There are three registers in the Port C (PC), namely RegPCIn, RegPCOut and RegPCDir. Table 13-1 to Table 13-3 show the mapping of control bits and functionality of these registers. Pos. 7-0 RegPCIn PCIn Rw r Reset - Description pad PC input value Table 13-1. RegPCIn (c) Semtech 2006 www.semtech.com 13-2 XE8805/05A Pos. 7-0 RegPCOut PCOut Rw rw Reset 0 resetpconf Description pad PC output value Rw rw Reset 0 resetpconf Description pad PC direction (0=input) Table 13-2. RegPCOut Pos. 7-0 RegPCDir PCDir Table 13-3. RegPCDir 13.5 Port C electrical specification Sym VINH VINL VOH description Input high voltage Input low voltage Output high voltage min 0.7*VBAT VSS VBAT-0.4 VOL Output low voltage VSS Cin Input capacitance typ max VBAT 0.2*VBAT VBAT VSS+0.4 3.0 unit Comments V VBAT2.4V V VBAT2.4V V VBAT=1.2V, IOH =0.3mA VBAT=2.4V, IOH =5.0mA VBAT=4.5V, IOH =8.0mA V VBAT=1.2V, IOL =0.3mA VBAT=2.4V, IOL =12.0mA VBAT=4.5V, IOL =15.0mA pF Note 1 Note 1: this value is indicative only since it depends on the package. Table 13-4. Port C electrical specification (c) Semtech 2006 www.semtech.com 13-3 XE8805/05A 14 UART 14.1 Features 14-2 14.2 Overview 14-2 14.3 Registers map 14-2 14.4 Interrupts map 14-3 14.5 14.5.1 14.5.2 Uart baud rate selection Uart on the RC oscillator Uart on the crystal oscillator 14-3 14-3 14-4 14.6 14.6.1 14.6.2 14.6.3 Function description Configuration bits Transmission Reception 14-4 14-4 14-5 14-5 14.7 Interrupt or polling 14-6 14.8 Software hints 14-7 (c) Semtech 2006 www.semtech.com 14-1 XE8805/05A 14.1 Features * * * * * * * * * * Full duplex operation with buffered receiver and transmitter. Internal baud rate generator with 12 programmable baud rates (300 - 115200). 7 or 8 bits word length. Even, odd, or no-parity bit generation and detection 1 stop bit Error receive detection: Start, Parity, Frame and Overrun Receiver echo mode 2 interrupts (receive full and transmit empty) Enable receive and/or transmit Invert pad Rx and/or Tx 14.2 Overview The UART pins are PB[7], which is used as Rx - receive and PB[6] as Tx - transmit. 14.3 Registers map pos. 7 6 5-3 2 1 0 RegUartCmd SelXtal UartEnRx2 UartRcSel(2:0) UartPM UartPE UartWL rw rw rw rw rw rw rw Reset 0 resetsystem 0 resetsystem 000 resetsystem 0 resetsystem 0 resetsystem 1 resetsystem Description Select input clock: 0 = RC, 1 = xtal Enable Uart Reception RC prescaler selection Select parity mode: 0 = odd, 1 = even Enable parity: 1 = with parity, 0 = no parity Select word length: 1 = 8 bits, 0 = 7 bits Table 14-1: RegUartCmd Pos. 7 RegUartCtrl UartEcho rw rw reset 0 resetsystem 6 5 4 3 2-0 UartEnRx1 UartEnTx UartXRx UartXTx UartBR(2:0) rw rw rw rw rw 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 101 resetsystem Description Enable echo mode: 1 = echo Rx->Tx, 0 = no echo Enable uart reception Enable uart transmission Invert pad Rx Invert pad Tx Select baud rate Table 14-2: RegUartCtrl pos. 7-0 RegUartTx UartTx rw rw reset 00000000 resetsystem Description Data to be sent Table 14-3: RegUartTx (c) Semtech 2006 www.semtech.com 14-2 XE8805/05A pos. 7-2 1 0 RegUartTxSta UartTxBusy UartTxFull rw r r r reset 000000 0 resetsystem 0 resetsystem description Unused Uart busy transmitting RegUartTx full Set by writing to RegUartTx Cleared when transferring RegUartTx into internal shift register Table 14-4: RegUartTxSta pos. 7-0 RegUartRx UartRx rw r reset 00000000 resetsystem description Received data Table 14-5: RegUartRx pos. 7-6 5 4 3 2 RegUartRxSta UartRxSErr UartRxPErr UartRxFErr UartRxOErr rw r r r r rc Reset 00 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem 1 0 UartRxBusy UartRxFull r r 0 resetsystem 0 resetsystem description Unused Start error Parity error Frame error Overrun error Cleared by writing RegUartRxSta Uart busy receiving RegUartRx full Cleared by reading RegUartRx Table 14-6: RegUartRxSta 14.4 Interrupts map interrupt source Irq_uart_Tx Irq_uart_Rx default mapping in the interrupt manager IrqHig(1) IrqHig(0) Table 14-7: Interrupts map 14.5 Uart baud rate selection In order to have correct baud rates, the Uart interface has to be fed with a stable and trimmed clock source. The clock source can be the RC oscillator or the crystal oscillator. The precision of the baud rate will depend on the precision of the selected clock source. 14.5.1 Uart on the RC oscillator To select the RC oscillator for the Uart, the bit SelXtal in RegUartCmd has to be 0. In order to obtain a correct baud rate, the RC oscillator frequency has to be set to one of the frequencies given in the table on the next page. The precision of the obtained baud rate is directly proportional to the frequency deviation with respect to the values in the table. (c) Semtech 2006 www.semtech.com 14-3 XE8805/05A Frequency selection for correct Uart baud rate with RC oscillator (Hz) 2'457'600 1'843'200 1'228'800 614'400 For each of these frequencies, the baud rate can be selected with the bits UartBR(2:0) in RegUartCtrl and UartRcSel(2:0) in RegUartCmd as shown in Table 14-8 RC frequency (Hz) UartRcSel 111 110 UartBR 101 100 2'457'600 010 1'228'800 001 38400 19200 9600 Not possible 614'400 000 4800 1'843'200 000 115200 57600 28800 14400 Table 14-8: Uart baud rate with RC clock Note: The precision of the baud rate is directly proportional to the frequency deviation of the used clock from the ideal frequency given in the table. In order to increase the precision and stability of the RC oscillator, the DFLL (digital frequency locked loop) can be used with the crystal oscillator as a reference. 14.5.2 Uart on the crystal oscillator In order to use the crystal oscillator as the clock source for the Uart, the bit SelXtal in RegUartCmd has to be set. The crystal oscillator has to be enabled by setting the EnableXtal bit in RegSysClock. The baud rate selection is done using the UartBR and UartRcSel bits as shown in Table 14-9. Xtal freq. (Hz) UartRcSel 32768 001 UartBR 011 010 001 000 Baud rate 2400 1200 600 300 Table 14-9: Uart baud rate with Xtal clock Due to the odd ratio between the crystal oscillator frequency and the baud rate, the generated baud rate has a systematic error of -2.48%. 14.6 Function description 14.6.1 Configuration bits The configuration bits of the Uart serial interface can be found in the registers RegUartCmd and RegUartCtrl. The bit SelXtal is used to select the clock source (see chapter 14.5). The bits UartSelRc and UartBR select the baud rate (see chapter 14.5). The bits UartEnTx is used to enable or disable the transmission. The bits UartEnRx1 and UartEnRx2 are used to enable or disable the reception. When one is set to 1, the reception is enabled. (c) Semtech 2006 www.semtech.com 14-4 XE8805/05A The word length (7 or 8 data bits) can be chosen with UartWL. A parity bit is added during transmission or checked during reception if UartPE is set. The parity mode (odd or even) can be chosen with UartPM. Setting the bits UartXRx and UartXTx inverts the Rx respectively Tx signals. The bit UartEcho is used to send the received data automatically back. The transmission function becomes then: Tx = Rx XOR UartXTx. 14.6.2 Transmission In order to send data, the transmitter has to be enabled by setting the bit UartEnTx. Data to be sent has to be written to the register RegUartTx. The bit UartTxFull in RegUartTxSta then goes to 1, indicating to the transmitter that a new word is available. As soon as the transmitter has finished sending the previous word, it then loads the contents of the register RegUartTx to an internal shift register and clears the UartTxFull bit. An interrupt is generated on Irq_uart_Tx at the falling edge of the UartTxFull bit. The bit UartTxBusy in RegUartTxSta shows that the transmitter is busy transmitting a word. A timing diagram is shown in Figure 14-1. Data are first sent LSB. New data should be written to the register RegUartTx only while UartTxFull is 0, otherwise data will be lost. Asynchronous Transmission write to RegUartTx word 1 RegUartTx reguarttx_shift word 1 shift clock Tx start b0 b1 b6/7 parity stop UartTxBusy UartTxFull Irq_uart_Tx Asynchronous Transmission (back to back) word 1 word 2 write to RegUartTx RegUartTx reguarttx_shift word 1 word 2 word 1 word 2 shift clock Tx start b0 b6/7 stop start UartTxBusy UartTxFull Irq_uart_Tx Figure 14-1. Uart transmission timing diagram. 14.6.3 Reception On detection of the start bit, the UartRxBusy bit is set. On detection of the stop bit, the received data are transferred from the internal shift register to the register RegUartRx. At the same time, the UartRxFull bit is set (c) Semtech 2006 www.semtech.com 14-5 XE8805/05A and an interrupt is generated on Irq_uart_Rx. This indicates that new data is available in RegUartRx. The timing diagram is shown in Figure 14-2. The UartRxFull bit is cleared when RegUartRx is read. If the register was not read before the receiver transfers a new word to it, the bit UartRxOErr (overflow error) is set and the previous contents of the register is lost. UartRxOErr is cleared by writing any data to RegUartRxSta. The bit UartRxSErr is set if a start error has been detected. The bit is updated at data transfer to RegUartRx. The bit UartRxPErr is set if a parity error has been detected, i.e. the received parity bit is not equal to the calculated parity of the received data. The bit is updated at data transfer to RegUartRx. The bit UartRxFErr in RegUartRxSta shows that a frame error has been detected. No stop bit has been detected. Asynchronous Reception read of RegUartRx (software) reguartrx_shift word 1 word 1 RegUartRx shift clock Rx start b0 b6/7 parity stop UartRxBusy UartRxFull Irq_uart_Rx Figure 14-2. Uart reception timing diagram. 14.7 Interrupt or polling The transmission and reception software can be driven by interruption or by polling the status bits. Interrupt driven reception: each time an Irq_uart_Rx interrupt is generated, a new word is available in RegUartRx. The register has to be read before a new word is received. Interrupt driven transmission: each time the contents of RegUartTx is transferred to the transmission shift register, an Irq_uart_Tx interrupt is generated. A new word can then be written to RegUartTx. Reception driven by polling: the UartRxFull bit is to be read and checked. When it is 1, the RegUartRx register contains new data and has to be read before a new word is received. Transmission driven by polling: the UartTxFull bit is to read and checked. When it is 0, the RegUartTx register is empty and a new word can be written to it. (c) Semtech 2006 www.semtech.com 14-6 XE8805/05A 14.8 Software hints Example of program for a transmission with polling: 1. The RegUartCmd register and the RegUartCtrl register are initialized (for example: 8 bit word length, odd parity, 9600 baud, enable Uart transmission). 2. Write a byte to RegUartTx. 3. Wait until the UartTxFull bit in RegUartTxSta register equals 0. 4. Jump to 2 to write the next byte if the message is not finished. 5. End of transmission. Example of program for a transmission with interrupt: 1. The RegUartCmd register and the RegUartCtrl register are initialized (for example: 8 bit word length, odd parity, 9600 baud, enable Uart transmission). 2. Write a byte to RegUartTx. 3. After an interrupt and if the message is not finished, jump to 2 4. End of transmission. Example of program for a reception with polling: 1. The RegUartCmd register and the RegUartCtrl register are initialized (for example: 8 bit word length, odd parity, 9600 baud, enable Uart reception). 2. Wait until the UartRxFull bit in the RegUartRxSta register equals 1. 3. Read the RegUartRxSta and check if there is no error. 4. Read data in RegUartRx. 5. If data is not equal to End-Of-Line, then jump to 2. 6. End of reception. Example of program for a reception with interrupt: 1. The RegUartCmd register and the RegUartCtrl register are initialized (for example: 8 bit word length, odd parity, 9600 baud, enable Uart reception). 2. When there is an interrupt, jump to 3 3. Read RegUartRxSta and check if there is no error. 4. Read data in RegUartRx. 5. If data is not equal to End-Of-Line, then jump to 2. 6. End of reception. (c) Semtech 2006 www.semtech.com 14-7 XE8805/05A 15 USRT 15.1 Features 15-2 15.2 Overview 15-2 15.3 Register map 15-2 15.4 Interrupts map 15-4 15.5 Conditional edge detection 1 15-4 15.6 Conditional edge detection 2 15-4 15.7 Interrupts or polling 15-5 15.8 Function description 15-5 (c) Semtech 2006 www.semtech.com 15-1 XE8805/05A 15.1 Features The USRT implements a hardware support for software implemented serial protocols: * Control of two external lines S0 and S1 (read/write). * Conditional edge detection generates interrupts. * S0 rising edge detection. * S1 value is stored on S0 rising edge. * S0 signal can be forced to 0 after a falling edge on S0 for clock stretching in the low state. * S0 signal can be stretched in the low state after a falling edge on S0 and after a S1 conditional detection. 15.2 Overview The USRT block supports software universal synchronous receiver and transmitter mode interfaces. External lines S0 and S1 respectively correspond to clock line and data line. S0 is mapped to PB[4] and S1 to PB[5] when the USRT block is enabled. It is independent from RegPBdir (Port B can be input or output). When USRT is enabled, the configurations in port B for PB[4] and PB[5] are overwritten by the USRT configuration. Internal pull-ups can be used by setting the PBPullup[5:4] bits. Conditional edge detections are provided. RegUsrtS1 can be used to read the S1 data line from PB[5] in receive mode or to drive the output S1 line PB[5] by writing it when in transmit mode. It is advised to read S1 data when in receive mode from the RegUsrtBufferS1 register, which is the S1 value sampled on a rising edge of S0. 15.3 Register map Block configuration registers: pos. 7-1 0 RegUsrtS1 UsrtS1 rw r rw reset function 0000000 1 resetsystem Unused Write: data S1 written to pad PB[5]), Read: value on PB[5] (not UsrtS1 value). Table 15-1: RegUsrtS1 pos. 7-1 0 RegUsrtS0 UsrtS0 rw r rw Reset 0000000 1 resetsystem function Unused Write: clock S0 written to pad PB[4], Read: value on PB[4] (not UsrtS0 value). Table 15-2: RegUsrtS0 The values that are read in the registers RegUsrtS1 and RegUsrtS0 are not necessarily the same as the values that were written in the register. The read value is read back on the circuit pins, not in the registers. Since the outputs are open drain, a value different from the register value may be forced by an external circuit on the circuit pins. (c) Semtech 2006 www.semtech.com 15-2 XE8805/05A pos. 7-4 3 RegUsrtCtrl UsrtWaitS0 rw r r reset "0000" 0 resetsystem 2 UsrtEnWaitCond1 rw 0 resetsystem 1 0 UsrtEnWaitS0 UsrtEnable rw rw 0 resetsystem 0 resetsystem function Unused Clock stretching flag (0=no stretching), cleared by writing RegUsrtBufferS1 Enable stretching on UsrtCond1 detection (0=disable) Enable stretching operation (0=disable) Enable USRT operation (0=disable) Table 15-3: RegUsrtCtrl pos. 7-1 0 RegUsrtCond1 UsrtCond1 rw r r/c reset 0000000 0 resetsystem function Unused State of condition 1 detection (1 =detected), cleared when written. Table 15-4: RegUsrtCond1 pos. 7-1 0 RegUsrtCond2 UsrtCond2 rw r r/c reset 0000000 0 resetsystem function Unused State of condition 2 detection (1 =detected), cleared when written. Table 15-5: RegUsrtCond2 pos. 7-1 RegUsrtBufferS1 - 0 UsrtBufferS1 rw r r w reset 0000000 function Unused Value on S1 at last S0 rising edge. Clear RegUsrtEdgeS0 bit in RegUsrtEdgeS0 Clear UsrtWaitS0 bit in RegUsrtCtrl with any value x Table 15-6: RegUsrtBufferS1 (c) Semtech 2006 www.semtech.com 15-3 XE8805/05A pos. 7-1 0 RegUsrtEdgeS0 UsrtEdgeS0 rw r r reset 0000000 0 resetsystem function Unused State of rising edge detection (1=detected). Cleared by RegUsrtBufferS1 on S0 reading Table 15-7: RegUsrtEdgeS0 15.4 Interrupts map interrupt source Irq_cond1 Irq_cond2 default mapping in the interrupt manager RegIrqMid(7) RegIrqMid(6) Table 15-8: Interrupts map 15.5 Conditional edge detection 1 S1 S0 Figure 15-1: Condition 1 Condition 1 is satisfied when S0=1 at the falling edge of S1. The bit UsrtCond1 in RegUsrtCond1 is set when the condition 1 is detected and the USRT interface is enabled (UsrtEnable=1). Condition 1 is asserted for both modes (receiver and transmitter). The UsrtCond1 bit is read only and is cleared by all reset conditions and by writing any data to its address. Condition 1 occurrence also generates an interrupt on Irq_cond1. 15.6 Conditional edge detection 2 S1 S0 Figure 15-2: Condition 2 (c) Semtech 2006 www.semtech.com 15-4 XE8805/05A Condition 2 is satisfied when S0=1 at the rising edge of S1. The bit UsrtCond2 in RegUsrtCond2 is set when the condition 2 is detected and the USRT interface is enabled. Condition 2 is asserted for both modes (receiver and transmitter). The UsrtCond2 bit is read only and is cleared by all reset conditions and by writing any data to its address. Condition 2 occurrence also generates an interrupt on Irq_cond2. 15.7 Interrupts or polling In receive mode, there are two possibilities for detecting condition 1 or 2: the detection of the condition can generate an interrupt or the registers can be polled (reading and checking the RegUsrtCond1 and RegUsrtCond2 registers for the status of USRT communication). 15.8 Function description The bit UsrtEnable in RegUsrtCtrl is used to enable the USRT interface and controls the PB[4] and PB[5] pins. This bit puts these two port B lines in the open drain configuration requested to use the USRT interface. If no external pull-ups are added on PB[4] and PB[5], the user can activate internal pull-ups by setting PBPullup[4] and PBPullup[5] in RegPBPullup. The bits UsrtEnWaitS0, UsrtEnWaitCond1, UsrtWaitS0 in RegUsrtCtrl are used for transmitter/receiver control of USRT interface. Figure 15-3 shows the unconditional clock stretching function which is enabled by setting UsrtEnWaitS0. S0 UsrtWaitS0 write Reg UsrtBufferS1 Figure 15-3: S0 Stretching (UsrtEnWaitS0=1) When UsrtEnWaitS0 is 1, the S0 line will be maintained at 0 after its falling edge (clock stretching). UsrtWaitS0 is then set to 1, indicating that the S0 line is forced low. One can release S0 by writing to the RegUsrtBufferS1 register. The same can be done in combination with condition 1 detection by setting the UsrtEnWaitCond1 bit. Figure 15-4 shows the conditional clock stretching function which is enabled by setting UsrtEnWaitCond1. (c) Semtech 2006 www.semtech.com 15-5 XE8805/05A S1 S0 UsrtWaitS0 write Reg UsrtBufferS1 Figure 15-4: Conditional stretching (UsrtEnWaitCond1=1) When UsrtEnWaitCond1 is 1, the S0 signal will be stretched in its low state after its falling edge if the condition 1 has been detected before (UsrtCond1=1). UsrtWaitS0 is then set to 1, indicating that the S0 line is forced low. One can release S0 by writing to the RegUsrtBufferS1 register. Figure 15-5 shows the sampling function implemented by the UsrtBufferS1 bit. The bit UsrtBufferS1 in RegUsrtBufferS1 is the value of S1 sampled on PB[4] at the last rising edge of S0. The bit UsrtEdgeS0 in RegUsrtEdgeS0 is set to one on the same S0 rising edge and is cleared by a read operation of the RegUsrtBufferS1 register. The bit therefore indicates that a new value is present in the RegUsrtBufferS1 which has not yet been read. S1 S0 UsrtBufferS1 read Reg UsrtBufferS1 UsrtEdgeS0 Figure 15-5: S1 sampling (c) Semtech 2006 www.semtech.com 15-6 XE8805/05A 16. Acquisition chain 16.1 ZoomingADCTM Features.......................................................................................................... 16-2 16.2 Overview .................................................................................................................................... 16-2 16.3 Register map ............................................................................................................................. 16-2 16.4 16.4.1 16.4.2 16.4.3 ZoomingADCTM Description ..................................................................................................... 16-4 Acquisition Chain ........................................................................................................................ 16-4 Peripheral Registers ................................................................................................................... 16-6 Continuous-Time vs. On-Request .............................................................................................. 16-8 16.5 Input Multiplexers ..................................................................................................................... 16-9 16.6 16.6.1 16.6.2 16.6.3 16.6.4 Programmable Gain Amplifiers............................................................................................. 16-10 PGA & ADC Enabling ............................................................................................................... 16-12 PGA1 ........................................................................................................................................ 16-12 PGA2 ........................................................................................................................................ 16-12 PGA3 ........................................................................................................................................ 16-12 16.7 16.7.1 16.7.2 16.7.3 16.7.4 16.7.5 16.7.6 16.7.7 16.7.8 ADC Characteristics ............................................................................................................... 16-13 Conversion Sequence .............................................................................................................. 16-13 Sampling Frequency................................................................................................................. 16-14 Over-Sampling Ratio ................................................................................................................ 16-14 Elementary Conversions........................................................................................................... 16-14 Resolution................................................................................................................................. 16-15 Conversion Time & Throughput................................................................................................ 16-16 Output Code Format ................................................................................................................. 16-16 Power Saving Modes................................................................................................................ 16-18 16.8 16.8.1 16.8.2 16.8.3 16.8.3.1 16.8.3.2 16.8.4 16.8.5 16.8.6 16.8.7 Specifications and Measured Curves................................................................................... 16-18 Default Settings ........................................................................................................................ 16-19 Specifications............................................................................................................................ 16-20 Linearity .................................................................................................................................... 16-21 Integral non-linearity ................................................................................................................. 16-21 Differential non-linearity ............................................................................................................ 16-25 Noise......................................................................................................................................... 16-26 Gain Error and Offset Error....................................................................................................... 16-27 Power Consumption ................................................................................................................. 16-28 Power Supply Rejection Ratio .................................................................................................. 16-30 16.9 16.9.1 16.9.2 16.9.3 16.9.4 16.9.5 Application Hints .................................................................................................................... 16-31 Input Impedance ....................................................................................................................... 16-31 PGA Settling or Input Channel Modifications ........................................................................... 16-31 PGA Gain & Offset, Linearity and Noise................................................................................... 16-31 Frequency Response................................................................................................................ 16-32 Power Reduction ...................................................................................................................... 16-33 (c) Semtech 2006 www.semtech.com 16-1 XE8805/05A 16.1 ZoomingADCTM Features The ZoomingADCTM is a complete and versatile low-power analog front-end interface typically intended for sensing applications. The key features of the ZoomingADCTM are: Programmable 6 to 16-bit dynamic range oversampled ADC * Flexible gain programming between 0.5 and 1000 * Flexible and large range offset compensation * 4-channel differential or 8-channel single-ended input multiplexer * 2-channel differential reference inputs * Power saving modes * Direct interfacing to CoolRiscTM microcontroller 16.2 Overview Analog Inputs MUX fS 0 1 2 3 4 5 6 7 VIN fS PGA1 PGA2 VD1 GD1 Input Selection MUX 0 Reference 1 Inputs 2 PGA3 VIN,ADC VD2 GD2 GD3 OFF2 OFF3 Offset2 Offset3 ADC 16 VREF 3 Reference Selection Gain1 Gain2 Gain3 ZOOM Figure 16-1. ZoomingADCTM general functional block diagram The total acquisition chain consists of an input multiplexer, 3 programmable gain amplifier stages and an oversampled A/D converter. The reference voltage can be selected on two different channels. Two offset compensation amplifiers allow for a wide offset compensation range. The programmable gain and offset allow one to zoom in on a small portion of the reference voltage defined input range. 16.3 Register map There are eight registers in the acquisition chain (AC), namely RegAcOutLsb, RegAcOutMsb, RegAcCfg0, RegAcCfg1, RegAcCfg2, RegAcCfg3, RegAcCfg4 and RegAcCfg5. Table 16-2 to Table 16-9 show the mapping of control bits and functionality of these registers while Table 16-1 gives an overview of these eight. The register map only gives a short description of the different configuration bits. More detailed information is found in subsequent sections. (c) Semtech 2006 www.semtech.com 16-2 XE8805/05A register name RegAcOutLsb RegAcOutMsb RegAcCfg0 RegAcCfg1 RegAcCfg2 RegAcCfg3 RegAcCfg4 RegAcCfg5 Table 16-1: AC registers pos. 7:0 RegAcOutLsb Out[7:0] rw r reset 00000000 resetsystem description LSB of the output code Table 16-2: RegAcOutLsb pos. 7:0 RegAcOutMsb Out[15:8] rw r reset 00000000 resetsystem description MSB of the output code Table 16-3: RegAcOutMsb pos. 7 6:5 RegAcCfg0 Start SET_NELCONV[1:0] rw w r0 rw reset 0 resetsystem 01 resetsystem 4:2 SET_OSR[2:0] rw 010 resetsystem 1 0 CONT reserved rw rw 0 resetsystem 0 resetsystem description starts a conversion sets the number of elementary conversions sets the oversampling rate of an elementary conversion continuous conversion mode Table 16-4: RegAcCfg0 pos. 7:6 5:4 3:0 RegAcCfg1 IB_AMP_ADC[1:0] IB_AMP_PGA[1:0] ENABLE[3:0] rw rw rw rw reset 11 resetsystem 11 resetsystem 0000 resetsystem description Bias current selection of the ADC converter Bias current selection of the PGA stages Enables the different PGA stages and the ADC Table 16-5: RegAcCfg1 pos. 7:6 5:4 3:0 RegAcCfg2 FIN[1:0] PGA2_GAIN[1:0] PGA2_OFFSET[3:0] rw rw rw rw reset 00 resetsystem 00 resetsystem 0000 resetsystem description Sampling frequency selection PGA2 stage gain selection PGA2 stage offset selection Table 16-6: RegAcCfg2 (c) Semtech 2006 www.semtech.com 16-3 XE8805/05A pos. 7 6:0 RegAcCfg3 PGA1_GAIN PGA3_GAIN[6:0] rw rw rw reset 0 resetsystem 0000000 resetsystem description PGA1 stage gain selection PGA3 stage gain selection Table 16-7: RegAcCfg3 pos. 7 6:0 RegAcCfg4 reserved PGA3_OFFSET[6:0] rw r rw reset 0 0000000 resetsystem description Unused PGA3 stage offset selection Table 16-8: RegAcCfg4 pos. 7 6 5:1 0 RegAcCfg5 BUSY DEF AMUX[4:0] VMUX rw r w r0 rw rw reset 0 resetsystem 0 00000 resetsystem 0 resetsystem description Activity flag Selects default configuration Input channel configuration selector Reference channel selector Table 16-9: RegAcCfg5 16.4 ZoomingADCTM Description Figure 16-2 gives a more detailed description of the acquisition chain. 16.4.1 Acquisition Chain Figure 16-1 shows the general block diagram of the acquisition chain (AC). A control block (not shown in Figure 16-1) manages all communications with the CoolRiscTM microcontroller. Analog inputs can be selected among eight input channels, while reference input is selected between two differential channels. The core of the zooming section is made of three differential programmable amplifiers (PGA). After selection of a combination of input and reference signals VIN and VREF, the input voltage is modulated and amplified through stages 1 to 3. Fine gain programming up to 1'000V/V is possible. In addition, the last two stages provide programmable offset. Each amplifier can be bypassed if needed. The output of the PGA stages is directly fed to the analog-to-digital converter (ADC), which converts the signal VIN,ADC into digital. Like most ADCs intended for instrumentation or sensing applications, the ZoomingADCTM is an over-sampled converter (See Note1). The ADC is a so-called incremental converter, with bipolar operation (the ADC accepts both positive and negative input voltages). In first approximation, the ADC output result relative to full-scale (FS) delivers the quantity: 1 Note: Over-sampled converters are operated with a sampling frequency f much higher than the input signal's Nyquist rate (typically f is 20S S 1'000 times the input signal bandwidth). The sampling frequency to throughput ratio is large (typically 10-500). These converters include digital decimation filtering. They are mainly used for high resolution, and/or low-to-medium speed applications. (c) Semtech 2006 www.semtech.com 16-4 XE8805/05A OUT ADC V IN , ADC FS / 2 V REF / 2 (Eq. 1) in two's complement (see Sections 16.4 and 16.7 for details). The output code OUTADC is -FS/2 to +FS/2 for VIN,ADC -VREF/2 to +VREF/2 respectively. As will be shown in section 16.6, VIN,ADC is related to input voltage VIN by the relationship: V IN , ADC = GDTOT V IN - GDoff TOT V REF (V) (Eq. 2) where GDTOT is the total PGA gain, and GDoffTOT is the total PGA offset. 0 1 AC_R 2 fS fS PGA1 PGA2 PGA3 GD1 GD2 GD3 OFF2 OFF3 VIN VIN,ADC ADC VREF MUX 0 1 2 3 AC_A 4 5 6 7 MUX Inputs 3 5 2 4 7 7 Acquisition Chain Register Bank RegACCfg5 RegACCfg4 8 RegACCfg3 RegACOutLSB RegACCfg2 RegACOutMSB RegACCfg1 RegACCfg0 Power Saving Modes PGA Enabling Conversion Start Nbr of Elementary Cycles Over-Sampling Ratio Continuous vs. On-Request 8 ADC Busy Flag Default Settings Sampling Frequency fS Figure 16-2. ZoomingADCTM detailed functional block diagram (c) Semtech 2006 www.semtech.com 16-5 XE8805/05A 16.4.2 Peripheral Registers Figure 16-2 shows a detailed functional diagram of the ZoomingADCTM. In table 16-10 the configuration of the peripheral registers is detailed. The system has a bank of eight 8-bit registers: six registers are used to configure the acquisition chain (RegAcCfg0 to 5), and two registers are used to store the output code of the analog-to-digital conversion (RegAcOutMsb & Lsb). The register coding of the ADC parameters and performance characteristics are detailed in Section 16.7. Table 16-10. Peripheral registers to configure the acquisition chain (AC) and to store the analog-to-digital conversion (ADC) result Register Name Bit Position 7 6 5 4 3 RegAcOutLsb OUT[7:0] RegAcOutMsb OUT[15:8] RegAcCfg0 Default values: RegAcCfg1 Default values: RegAcCfg2 Default values: RegAcCfg3 Default values: RegAcCfg4 Default values: RegAcCfg5 Default values: START 0 SET_NELC[1:0] 01 IB_AMP_ADC[1:0] 11 FIN[1:0] 00 PGA1_G 0 0 BUSY 0 DEF 0 SET_OSR[2:0] 010 2 1 CONT 0 IB_AMP_PGA[1:0] ENABLE[3:0] 11 0001 PGA2_GAIN[1:0] PGA2_OFFSET[3:0] 00 0000 PGA3_GAIN[6:0] 0000000 PGA3_OFFSET[6:0] 0000000 AMUX[4:0] 00000 0 TEST 0 VMUX 0 With: * OUT: (r) digital output code of the analog-to-digital converter. (MSB = OUT[15]) * START: (w) setting this bit triggers a single conversion (after the current one is finished). This bit always reads back 0. SET_NELC[1:0] * SET_NELC: (rw) sets the number of elementary conversions to 2 . To compensate for offsets, the input signal is chopped between elementary conversions (1,2,4,8). (3+SET_OSR[2:0]) . OSR = 8, 16, 32, ..., * SET_OSR: (rw) sets the over-sampling rate (OSR) of an elementary conversion to 2 512, 1024. * CONT: (rw) setting this bit starts a conversion. A new conversion will automatically begin as long as the bit remains at 1. * TEST: bit only used for test purposes. In normal mode, this bit is forced to 0 and cannot be overwritten. * IB_AMP_ADC: (rw) sets the bias current in the ADC to 0.25*(1+ IB_AMP_ADC[1:0]) of the normal operation current (25, 50, 75 or 100% of nominal current). To be used for low-power, low-speed operation. * IB_AMP_PGA: (rw) sets the bias current in the PGAs to 0.25*(1+IB_AMP_PGA[1:0]) of the normal operation current (25, 50, 75 or 100% of nominal current). To be used for low-power, low-speed operation. * ENABLE: (rw) enables the ADC modulator (bit 0) and the different stages of the PGAs (PGAi by bit i=1,2,3). PGA stages that are disabled are bypassed. * FIN: (rw) These bits set the sampling frequency of the acquisition chain. Expressed as a fraction of the oscillator frequency, the sampling frequency is given as: 00 AE 1/4 fRC, 01 AE 1/8 fRC, 10 AE 1/32 fRC, 11AE ~8kHz. * PGA1_GAIN: (rw) sets the gain of the first stage: 0 AE 1, 1 AE 10. * PGA2_GAIN: (rw) sets the gain of the second stage: 00 AE 1, 01 AE 2, 10 AE 5, 11 AE 10. * PGA3_GAIN: (rw) sets the gain of the third stage to PGA3_GAIN[6:0]1/12. * PGA2_OFFSET: (rw) sets the offset of the second stage between -1 and +1, with increments of 0.2. The MSB gives the sign (0 positive, 1 negative); amplitude is coded with the bits PGA2_OFFSET[5:0]. * PGA3_OFFSET: (rw) sets the offset of the third stage between -5.25 and +5.25, with increments of 1/12. The MSB gives the sign (0 positive, 1 negative); amplitude is coded with the bits PGA3_OFFSET[5:0]. (c) Semtech 2006 www.semtech.com 16-6 XE8805/05A * * * * BUSY: (r) set to 1 if a conversion is running. Note that the flag is set at the effective start of the conversion. Since the ADC is generally synchronized on a lower frequency clock than the CPU, there might be a small delay (max. 1 cycle of the ADC sampling frequency) between the writing of the START or CONT bits and the appearance of BUSY flag. DEF: (w) sets all values to their defaults (PGA disabled, max speed, nominal modulator bias current, 2 elementary conversions, over-sampling rate of 32) and starts a new conversion without waiting the end of the preceding one. AMUX(4:0): (rw) AMUX[4] sets the mode (0 AE 4 differential inputs, 1 AE 7 inputs with A(0) = common reference) AMUX(3) sets the sign (0 AE straight, 1AE cross) AMUX[2:0] sets the channel. VMUX: (rw) sets the differential reference channel (0 AE R(1) and R(0), 1 AE R(3) and R(2)). (r = read; w = write; rw = read & write) (c) Semtech 2006 www.semtech.com 16-7 XE8805/05A 16.4.3 Continuous-Time vs. On-Request The ADC can be operated in two distinct modes: "continuous-time" and "on-request" modes (selected using the bit CONT). In "continuous-time" mode, the input signal is repeatedly converted into digital. After a conversion is finished, a new one is automatically initiated. The new value is then written in the result register, and the corresponding internal trigger pulse is generated. This operation is sketched in Figure 16-3. The conversion time in this case is defined as TCONV. TCONV Internal Trig Ouput Code RegACOut[15:0] BUSY IRQ Figure 16-3. ADC "continuous-time" operation T CONV Internal Trig Request START Ouput Code RegACOut[15:0] BUSY IRQ Figure 16-4. ADC "on-request" operation In the "on-request" mode, the internal behaviour of the converter is the same as in the "continuous-time" mode, but the conversion is initiated on user request (with the START bit). As shown in Figure 16-4, the conversion time is also TCONV. Note that the flag is set at the effective start of the conversion. Since the ADC is generally synchronized on a lower frequency clock than the CPU, there might be a small delay (max. 1 cycle of the ADC sampling frequency) between the writing of the START or CONT bits and the appearance of BUSY flag. (c) Semtech 2006 www.semtech.com 16-8 XE8805/05A 16.5 Input Multiplexers The ZoomingADCTM has eight analog inputs AC_A(0) to AC_A(7) and four reference inputs AC_R(0) to AC_R(3). Let us first define the differential input voltage VIN and reference voltage VREF respectively as: VIN = VINP - VINN (V) (Eq. 3) VREF = VREFP - VREFN (V) (Eq. 4) and: As shown in Table 16-11 the inputs can be configured in two ways: either as 4 differential channels (VIN1 = AC_A(1) - AC_A(0),..., VIN4 = AC_A(7) - AC_A(6)), or AC_A(0) can be used as a common reference, providing 7 signal paths all referred to AC_A(0). The control word for the analog input selection is AMUX[4:0]. Notice that the bit AMUX[3] controls the sign of the input voltage. AMUX[4:0] (RegAcCfg5[5:1]) VINP VINN AMUX[4:0] (RegAcCfg5[5:1]) VINP VINN 00x00 00x01 00x10 00x11 AC_A(1) AC_A(3) AC_A(5) AC_A(7) AC_A(0) AC_A(2) AC_A(4) AC_A(6) 01x00 01x01 01x10 01x11 AC_A(0) AC_A(2) AC_A(4) AC_A(6) AC_A(1) AC_A(3) AC_A(5) AC_A(7) 10000 10001 10010 10011 10100 10101 10110 10111 AC_A(0) AC_A(1) AC_A(2) AC_A(3) AC_A(4) AC_A(5) AC_A(6) AC_A(7) AC_A(0) 11000 11001 11010 11011 11100 11101 11110 11111 AC_A(0) AC_A(0) AC_A(1) AC_A(2) AC_A(3) AC_A(4) AC_A(5) AC_A(6) AC_A(7) Table 16-11. Analog input selection Similarly, the reference voltage is chosen among two differential channels (VREF1 = AC_R(1)-AC_R(0) or VREF2 = AC_R(3)-AC_R(2)) as shown in Table 16-12. The selection bit is VMUX. The reference inputs VREFP and VREFN (common-mode) can be up to the power supply range. (c) Semtech 2006 www.semtech.com 16-9 XE8805/05A VMUX (RegAcCfg5[0]) 0 1 VREFP VREFN AC_R(1) AC_R(3) AC_R(0) AC_R(2) Table 16-12. Analog Reference input selection 16.6 Programmable Gain Amplifiers As seen in Figure 16-1, the zooming function is implemented with three programmable gain amplifiers (PGA). These are: * PGA1: coarse gain tuning * PGA2: medium gain and offset tuning * PGA3: fine gain and offset tuning All gain and offset settings are realized with ratios of capacitors. The user has control over each PGA activation and gain, as well as the offset of stages 2 and 3. These functions are examined hereafter. ENABLE[3:0] Block xxx0 xxx1 xx0x xx1x x0xx x1xx 0xxx 1xxx ADC disabled ADC enabled PGA1 disabled PGA1 enabled PGA2 disabled PGA2 enabled PGA3 disabled PGA3 enabled Table 16-13. ADC & PGA enabling PGA1 Gain GD1 (V/V) 1 10 PGA1_GAIN 0 1 Table 16-14. PGA1 Gain Settings PGA2 Gain GD2 (V/V) 1 2 5 10 PGA2_GAIN[1:0] 00 01 10 11 Table 16-15. PGA2 gain settings (c) Semtech 2006 www.semtech.com 16-10 XE8805/05A PGA2_OFFSET[3:0] 0000 0001 0010 0011 0100 0101 1001 1010 1011 1100 1101 PGA2 Offset GDoff2 (V/V) 0 +0.2 +0.4 +0.6 +0.8 +1 -0.2 -0.4 -0.6 -0.8 -1 Table 16-16. PGA2 offset settings PGA3 Gain GD3 (V/V) 0 1/12(=0.083) ... 6/12 ... 12/12 16/12 PGA3_GAIN[6:0] 0000000 0000001 ... 0000110 ... 0001100 0010000 ... 0100000 ... 1000000 ... 1111111 32/12 64/12 127/12(=10.58) Table 16-17. PGA3 gain settings PGA3_OFFSET[6:0] 0000000 0000001 0000010 ... 0010000 ... 0100000 ... 0111111 1000000 1000001 1000010 ... 1010000 ... 1100000 ... 1111111 PGA3 Offset GDoff3 (V/V) 0 +1/12(=+0.083) +2/12 ... +16/12 ... +32/12 ... +63/12(=+5.25) 0 -1/12(=-0.083) -2/12 ... -16/12 ... -32/12 ... -63/12(=-5.25) Table 16-18. PGA3 offset settings (c) Semtech 2006 www.semtech.com 16-11 XE8805/05A 16.6.1 PGA & ADC Enabling Depending on the application objectives, the user may enable or bypass each PGA stage. This is done according to the word ENABLE and the coding given in Table 16-13. To reduce power dissipation, the ADC can also be inactivated while idle. 16.6.2 PGA1 The first stage can have a buffer function (unity gain) or provide a gain of 10 (see Table 16-14). The voltage VD1 at the output of PGA1 is: V D1 = GD1 V IN (V) (Eq. 5) where GD1 is the gain of PGA1 (in V/V) controlled with the bit PGA1_GAIN. 16.6.3 PGA2 The second PGA has a finer gain and offset tuning capability, as shown in Table 16-15 and Table 16-16. The voltage VD2 at the output of PGA2 is given by: V D 2 = GD2 V D1 - GDoff 2 V REF (V) (Eq. 6) where GD2 and GDoff2 are respectively the gain and offset of PGA2 (in V/V). These are controlled with the words PGA2_GAIN[1:0] and PGA2_OFFSET[3:0]. As shown in equation 6, the offset correction is directly proportional to the reference voltage. All drifts and perturbations on the reference voltage will affect the precision of the offset compensation. 16.6.4 PGA3 The finest gain and offset tuning is performed with the third and last PGA stage, according to the coding of Table 16-17 and Table 16-18. The output of PGA3 is also the input of the ADC. Thus, similarly to PGA2, we find that the voltage entering the ADC is given by: V IN , ADC = GD3 V D 2 - GDoff 3 V REF (V) (Eq. 7) where GD3 and GDoff3 are respectively the gain and offset of PGA3 (in V/V). The control words are PGA3_GAIN[6:0] and PGA3_OFFSET[6:0] . To remain within the signal compliance of the PGA stages, the condition: V D1 , V D 2 < V DD (V) (Eq. 8) must be verified. As shown in equation 7, the offset correction is directly proportional to the reference voltage. All drifts and perturbations on the reference voltage will affect the precision of the offset compensation. (c) Semtech 2006 www.semtech.com 16-12 XE8805/05A Finally, combining equations Eq. 5 to Eq. 7 for the three PGA stages, the input voltage VIN,ADC of the ADC is related to VIN by: V IN , ADC = GDTOT V IN - GDoff TOT V REF (V) (Eq. 9) where the total PGA gain is defined as: GDTOT = GD3 GD2 GD1 (V/V) (Eq. 10) and the total PGA offset is: GDoff TOT = GDoff 3 + GD3 GDoff 2 (V/V) (Eq. 11) 16.7 ADC Characteristics The main performance characteristics of the ADC (resolution, conversion time, etc.) are determined by three programmable parameters: * * * sampling frequency fS, over-sampling ratio OSR, and number of elementary conversions NELCONV. The setting of these parameters and the resulting performances are described hereafter. 16.7.1 Conversion Sequence A conversion is started each time the bit START or the bit DEF is set. As depicted in Figure 16-5, a complete analog-to-digital conversion sequence is made of a set of NELCONV elementary incremental conversions and a final quantization step. Each elementary conversion is made of (OSR+1) sampling periods TS=1/fS, i.e.: TELCONV = (OSR + 1) / f S (s) (Eq. 12) The result is the mean of the elementary conversion results. An important feature is that the elementary conversions are alternatively performed with the offset of the internal amplifiers contributing in one direction and the other to the output code. Thus, converter internal offset is eliminated if at least two elementary sequences are performed (i.e. if NELCONV 2). A few additional clock cycles are also required to initiate and end the conversion properly. TELCONV= (OSR+1)/f S Init Elementary Conversion Conversion index Offset 1 + Elementary Conversion Elementary Conversion Elementary Conversion 2 - NELCONV- 1 + N ELCONV - T CONV End Conversion Result Figure 16-5. Analog-to-digital conversion sequence (c) Semtech 2006 www.semtech.com 16-13 XE8805/05A 16.7.2 Sampling Frequency The word FIN[1:0] is used to select the sampling frequency fS (Table 16-19). Three sub-multiples of the internal RC-based frequency fRCEXT can be chosen. For FIN = "11", sampling frequency is about 8kHz. Additional information on oscillators and their control can be found in the clock block documentation. FIN[1:0] 00 01 10 11 Sampling Frequency fS (Hz) 01/05 02 1/4fRC 1/8fRCEXT 1/8fRC 1/16fRCEXT 1/32fRC 1/64fRCEXT 8kHz 4kHz Table 16-19. Sampling frequency settings (fRC= RC-based frequency) 16.7.3 Over-Sampling Ratio The over-sampling ratio (OSR) defines the number of integration cycles per elementary conversion. Its value is set with the word SET_OSR[2:0] in power of 2 steps (see Table 16-20) given by: 3 + SET_OSR[2 : 0] OSR = 2 (-) SET_OSR[2:0] (RegAcCfg0[4:2]) 000 001 010 011 100 101 110 111 (Eq. 13) Over-Sampling Ratio OSR (-) 8 16 32 64 128 256 512 1024 Table 16-20. Over-sampling ratio settings 16.7.4 Elementary Conversions As mentioned previously, the whole conversion sequence is made of a set of NELCONV elementary incremental conversions. This number is set with the word SET_NELC[1:0] in power of 2 steps (see Table 16-21) given by: SET_NELC[1: 0] N ELCONV = 2 (-) (Eq. 14) (c) Semtech 2006 www.semtech.com 16-14 XE8805/05A # of Elementary Conversions NELCONV (-) 1 2 4 8 SET_NELC[1:0] (RegAcCfg0[6:5]) 00 01 10 11 Table 16-21. Number of elementary conversion settings As already mentioned, NELCONV must be equal or greater than 2 to reduce internal amplifier offsets. 16.7.5 Resolution The theoretical resolution of the ADC, without considering thermal noise, is given by: n = 2 log 2 (OSR) + log 2 ( N ELCONV ) (Bits) (Eq. 15) 17 Resolution - n [Bits] 15 13 11 SET_NELC= 11 10 01 00 9 7 5 000 001 010 011 100 101 110 111 SET_OSR Figure 16-6. Resolution vs. SET_OSR[2:0] and SET_NELC[2:0] SET_OSR [2:0] 000 001 010 011 100 101 110 111 00 SET_NELC 01 10 11 6 8 10 12 14 16 16 16 7 9 11 13 15 16 16 16 9 11 13 15 16 16 16 16 8 10 12 14 16 16 16 16 (shaded area: resolution truncated to 16 bits due to output register size RegAcOut[15:0]) Table 16-22. Resolution vs. SET_OSR[2:0] and SET_NELC[1:0] settings Using look-up Table 16-22 or the graph plotted in Figure 16-6, resolution can be set between 6 and 16 bits. Notice that, because of 16-bit register use for the ADC output, practical resolution is limited to 16 bits, i.e. n 16. Even if the resolution is truncated to 16 bit by the output register size, it may make sense to set OSR and NELCONV to higher values in order to reduce the influence of the thermal noise in the PGA (see section 16.8.4). (c) Semtech 2006 www.semtech.com 16-15 XE8805/05A 16.7.6 Conversion Time & Throughput As explained using Figure 16-5, conversion time is given by: TCONV = ( N ELCONV (OSR + 1) + 1) / f S (s) (Eq. 16) and throughput is then simply 1/TCONV. For example, consider an over-sampling ratio of 256, 2 elementary conversions, and a sampling frequency of 500kHz (SET_OSR = "101", SET_NELC = "01", fRC = 2MHz, and FIN = "00"). In this case, using Table 16-23, the conversion time is 515 sampling periods, or 1.03ms. This corresponds to a throughput of 971Hz in continuous-time mode. The plot of Figure 16-7 illustrates the classic trade-off between resolution and conversion time. SET_OSR [2:0] 000 001 010 011 100 101 110 111 SET_NELC[1:0] 01 10 19 37 35 69 67 133 131 261 259 517 515 1029 1027 2053 2051 4101 00 10 18 34 66 130 258 514 1026 11 73 137 265 521 1033 2057 4105 8201 Table 16-23. Normalized conversion time (TCONV fS) vs. SET_OSR[2:0] and SET_NELC[1:0](normalized to sampling period 1/fS) Resolution - n [Bits] 16.0 14.0 12.0 10.0 8.0 6.0 10 00 11 SET_NELC 01 4.0 10.0 100.0 1000.0 10000.0 Normalized Conversion Time - TCONV*fS [-] Figure 16-7. Resolution vs. normalized conversion time for different SET_NELC[1:0] 16.7.7 Output Code Format The ADC output code is a 16-bit word in two's complement format (see Table 16-24). For input voltages outside the range, the output code is saturated to the closest full-scale value (i.e. 0x7FFF or 0x8000). For resolutions smaller than 16 bits, the non-significant bits are forced to the values shown in Table 16-25. The output code, expressed in LSBs, corresponds to: OUTADC = 216 VIN , ADC OSR + 1 (LSB) VREF OSR (Eq.17) Recalling equation Eq. 9, this can be rewritten as: (c) Semtech 2006 www.semtech.com 16-16 XE8805/05A OUT ADC = 216 V IN V REF V GDTOT - GDoff TOT REF VIN OSR + 1 OSR (LSB) (Eq. 18) where, from Eq. 10 and Eq. 11, the total PGA gain and offset are respectively: GDTOT = GD3 GD2 GD1 (V/V) and: GDoff TOT = GDoff 3 + GD3 GDoff 2 (V/V) ADC Input Voltage VIN,ADC % of Full Scale (FS) +2.49505V +0.5FS +2.49497V ... ... +76.145V 0V -76.145V ... ... ... 0 ... ... -2.49505V ... -2.49513V -0.5FS Output in LSBs +215-1 =+32'767 15 +2 -2 =+32'766 ... +1 0 -1 ... 15 -2 -1 =-32'767 -215 =-32'768 Output Code in Hex 7FFF 7FFE ... 0001 0000 FFFF ... 8001 8000 Table 16-24. Basic ADC Relationships (example for: VREF = 5V, OSR = 512, n = 16 bits) SET_OSR [2:0] 000 001 010 011 100 101 110 111 SET_NELC = 00 SET_NELC = 01 SET_NELC = 10 SET_NELC = 11 1000000000 10000000 100000 1000 10 - 100000000 1000000 10000 100 1 - 10000000 100000 1000 10 - 1000000 10000 100 1 - Table 16-25. Last forced LSBs in conversion output registers for resolution settings smaller than 16 bits (n < 16) (RegAcOutMsb[7:0] & RegAcOutLsb[7:0]) (c) Semtech 2006 www.semtech.com 16-17 XE8805/05A The equivalent LSB size at the input of the PGA chain is: LSB = 1 V REF OSR n 2 GDTOT OSR + 1 (V) (Eq. 19) Notice that the input voltage VIN,ADC of the ADC must satisfy the condition: VIN , ADC 1 OSR (V REFP - VREFN ) 2 OSR + 1 (V) (Eq. 20) to remain within the ADC input range. 16.7.8 Power Saving Modes During low-speed operation, the bias current in the PGAs and ADC can be programmed to save power using the control words IB_AMP_PGA[1:0] and IB_AMP_ADC[1:0] (see Table 16-26). If the system is idle, the PGAs and ADC can even be disabled, thus, reducing power consumption to its minimum. This can considerably improve battery lifetime. ADC PGA [1:0] Bias Current Bias Current 00 01 10 11 x 1/4IADC 1/2IADC 3/4IADC IADC x x 00 01 10 11 x 1/4IPGA 1/2IPGA 3/4IPGA IPGA IB_AMP_ADC [1:0] IB_AMP_PGA Max. fS [kHz] 62.5 125 250 500 62.5 125 250 500 Table 16-26. ADC & PGA power saving modes and maximum sampling frequency 16.8 Specifications and Measured Curves This section presents measurement results for the acquisition chain. A summary table with circuit specifications and measured curves are given. (c) Semtech 2006 www.semtech.com 16-18 XE8805/05A 16.8.1 Default Settings Unless otherwise specified, the measurement conditions are the following: * * * * * * Temperature TA = +25C VDD = +5V, GND = 0V, VREF = +5V, VIN = 0V RC frequency fRC = 2MHz, sampling frequency fS = 500kHz Offsets GDOff2 = GDOff3 = 0 Power operation: normal (IB_AMP_ADC[1:0] = IB_AMP_PGA[1:0] = '11') Resolution: for n = 12 bits: OSR = 32 and NELCONV = 4 for n = 16 bits: OSR = 512 and NELCONV = 2 (c) Semtech 2006 www.semtech.com 16-19 XE8805/05A 16.8.2 Specifications Unless otherwise specified: Temperature TA = +25C, VDD = +5V, GND = 0V, VREF = +5V, VIN = 0V, RC frequency fRC = 2MHz, sampling frequency fS = 500kHz, Overall PGA gain GDTOT = 1, offsets GDOff2 = GDOff3 = 0. Power operation: normal (IB_AMP_ADC[1:0] = IB_AMP_PGA[1:0] = '11'). For resolution n = 12 bits: OSR = 32 and NELCONV = 4. For resolution n = 16 bits: OSR = 512 and NELCONV = 2. VALUE PARAMETER UNITS COMMENTS/CONDITIONS MIN TYP MAX ANALOG INPUT CHARACTERISTICS -2.42 V Differential Input Voltage Ranges Gain = 1, OSR = 32 (Note 1) +2.42 -24.2 mV VIN = (VINP - VINN) Gain = 100, OSR = 32 +24.2 -2.42 mV Gain = 1000, OSR = 32 +2.42 Reference Voltage Range VREF = (VREFP - VREFN) PROGRAMMABLE GAIN AMPLIFIERS (PGA) Total PGA Gain, GDTOT PGA1 Gain, GD1 PGA2 Gain, GD2 PGA3 Gain, GD3 Gain Setting Precision (each stage) Gain Temperature Dependence Offset PGA2 Offset, GDoff2 PGA3 Offset, GDoff3 Offset Setting Precision (PGA2 or 3) Offset Temperature Dependence Input Impedance PGA1 PGA2, PGA3 Output RMS Noise PGA1 PGA2 PGA3 ADC STATIC PERFORMANCE Resolution, n No Missing Codes Gain Error Offset Error 0.5 1 1 0 -3 -1 -127/12 -3 Throughput Rate (Continuous Mode), 1/TCONV Nbr of Initialization Cycles, NINIT Nbr of End Conversion Cycles, NEND PGA Stabilization Delay DIGITAL OUTPUT ADC Output Data Coding 0.5 5 V 1000 10 10 127/12 +3 V/V V/V V/V V/V % ppm/C +1 +127/12 +3 V/V V/V % ppm/C k k k 1500 150 150 V V V 205 340 365 6 Integral Non-Linearity, INL Resolution n = 16 Bits Differential Non-Linearity, DNL Resolution n = 16 Bits Power Supply Rejection Ratio, PSRR DYNAMIC PERFORMANCE Sampling Frequency, fS Conversion Time, TCONV 0.5 5 VDD 16 Bits See Table 16-14 See Table 16-15 Step=1/12 V/V, 16-17 See Step=0.2 V/V, See Table 16-16 Step=1/12 V/V, See Table 16-18 (Note 2) PGA1 Gain = 1 (Note 3) PGA1 Gain = 10 (Note 3) Maximal gain (Note 3) (Note 4) (Note 5) (Note 6) (Note 7) (Note 8) (Note 9) n = 16 bits (Note 10) 0.15 1 % of FS LSB 1.0 LSB (Note 11) 0.5 78 72 LSB dB dB (Note 12) VDD = 5V 0.3V (Note 13) VDD = 3V 0.3V (Note 13) 3 133 1027 3.76 0.49 0 0 2 5 OSR kHz cycles/fS cycles/fS kSps kSps cycles cycles cycles Table n = 12 bits (Note 14) n = 16 bits (Note 14) n = 12 bits, fS = 500kHz n = 16 bits, fS = 500kHz (Note 15) Binary Two's Complement See Table 16-24 and Table 16-25 (c) Semtech 2006 www.semtech.com 16-20 XE8805/05A Specifications (Cont'd) PARAMETER POWER SUPPLY Voltage Supply Range, VDD Analog Quiescent Current Consumption, Total (IQ) ADC Only PGA1 PGA2 PGA3 Analog Power Dissipation Normal Power Mode 3/4 Power Reduction Mode 1/2 Power Reduction Mode 1/4 Power Reduction Mode TEMPERATURE Specified Range Operating Range MIN VALUE TYP MAX +2.4 +5 +5.5 -40 -40 UNITS V 720/620 250/190 165/150 130/120 175/160 A A A A A 3.6/1.9 2.7/1.4 1.8/0.9 0.9/0.5 mW mW mW mW +85 +125 COMMENTS/CONDITIONS Only Acquisition Chain VDD = 5V/3V VDD = 5V/3V VDD = 5V/3V VDD = 5V/3V VDD = 5V/3V All PGAs & ADC Active VDD = 5V/3V (Note 16) VDD = 5V/3V (Note 17) VDD = 5V/3V (Note 18) VDD = 5V/3V (Note 19) C C Notes: (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) = GD1GD2GD3. Maximum input voltage is given by: Gain defined as overall PGA gain GDTOT VIN,MAX = (VREF/2)(OSR/OSR+1). Offset due to tolerance on GDoff2 or GDoff3 setting. For small intrinsic offset, use only ADC and PGA1. Measured with block connected to inputs through AMUX block. Normalized input sampling frequency for input impedance is fS = 512kHz. This figure must be multiplied by 2 for fS = 256kHz, 4 for fS = 128kHz. Input impedance is proportional to 1/fS. See model of Figure 16-18(a). Figure independent from PGA1 gain and sampling frequency fS. See equation Eq. 21 to calculate equivalent input noise. Figure independent on PGA2 gain and sampling frequency fS. See model of Figure 16-18(a). See equation Eq. 21 to calculate equivalent input noise. Figure independent on PGA3 gain and sampling frequency fS. See model of Figure 16-18(a) and equation Eq. 21 to calculate equivalent input noise. Resolution is given by n = 2log2(OSR) + log2(NELCONV). OSR can be set between 8 and 1024, in powers of 2. NELCONV can be set to 1, 2, 4 or 8. If a ramp signal is applied to the input, all digital codes appear in the resulting ADC output data. Gain error is defined as the amount of deviation between the ideal (theoretical) transfer function and the measured transfer function (with the offset error removed). (See Figure 16-19) Offset error is defined as the output code error for a zero volt input (ideally, output code = 0). For 1 LSB offset, NELCONV must be 2. INL defined as the deviation of the DC transfer curve of each individual code from the best-fit straight line. This specification holds over the full scale. DNL is defined as the difference (in LSB) between the ideal (1 LSB) and measured code transitions for successive codes. Figures for Gains = 1 to 100. PSRR is defined as the amount of change in the ADC output value as the power supply voltage changes. Conversion time is given by: TCONV = (NELCONV (OSR + 1) + 1) / fS. OSR can be set between 8 and 1024, in powers of 2. NELCONV can be set to 1, 2, 4 or 8. PGAs are reset after each writing operation to registers RegAcCfg1-5. The ADC must be started after a PGA or inputs commonmode stabilisation delay. This is done by writing bit Start several cycles after PGA settings modification or channel switching. Delay between PGA start or input channel switching and ADC start should be equivalent to OSR (between 8 and 1024) number of cycles. This delay does not apply to conversions made without the PGAs. Nominal (maximum) bias currents in PGAs and ADC, i.e. IB_AMP_PGA[1:0] = `11' and IB_AMP_ADC[1:0] = `11'. Bias currents in PGAs and ADC set to 3/4 of nominal values, i.e. IB_AMP_PGA[1:0] = `10', IB_AMP_ADC[1:0] = `10'. Bias currents in PGAs and ADC set to 1/2 of nominal values, i.e. IB_AMP_PGA[1:0] = `01', IB_AMP_ADC[1:0] = `01'. Bias currents in PGAs and ADC set to 1/4 of nominal values, i.e. IB_AMP_PGA[1:0] = `00', IB_AMP_ADC[1:0] = `00'. 16.8.3 Linearity 16.8.3.1 Integral non-linearity The integral non-linearity depends on the selected gain configuration. First of all, the non-linearity of the ADC (all PGA stages bypassed) is shown in Figure 16-8. (c) Semtech 2006 www.semtech.com 16-21 XE8805/05A Figure 16-8. Integral non-linearity of the ADC (PGA disabled, reference voltage of 4.8V) The different PGA stages have been designed to find the best compromise between the noise performance, the integral non-linearity and the power consumption. To obtain this, the first stage has the best noise performance and the third stage the best linearity performance. For large input signals (small PGA gains, i.e. up to about 50), the noise added by the PGA is very small with respect to the input signal and the second and third stage of the PGA should be used to get the best linearity. For small input signals (large gains, i.e. above 50), the noise level in the PGA is important and the first stage of the PGA should be used. The following figures give the non-linearity for different gain settings of the PGA, selecting the appropriate stage to get the best noise and linearity performance. Figure 16-9 shows the non-linearity when the third stage is used with a gain of 1. It is of course not very useful to use the PGA with a gain of 1 unless it is used to compensate offset. By increasing the gain, the integral non-linearity becomes even smaller since the signal in the amplifiers reduces. Figure 16-10 shows the non-linearity for a gain of 2. Figure 16-11 shows the non-linearity for a gain of 5. Figure 16-12 shows the non-linearity for a gain of 10. By comparing these figures to Figure 16-8, it can be seen that the third stage of the PGA does not add significant integral non-linearity. Figure 16-13 shows the non-linearity for a gain of 20 and Figure 16-14 shows the non-linearity for a gain of 50. In both cases the PGA2 is used at a gain of 10 and the remaining gain is realized by the third stage. It can be seen again that the second stage of the PGA does not add significant non-linearity. For gains above 50, the first stage PGA1 should be selected in stead of PGA2. Although the non-linearity in the first stage of the PGA is larger than in stage 2 and 3, the gain in stage 3 is now sufficiently high so that the non-linearity of the first stage does become negligible as is shown in Figure 16-15 for a gain of 100. Therefor, the first stage is preferred over the second stage since it has less noise. Increasing the gain further up to 1000 will further increase the linearity since the signal becomes very small in the first two stages. The signal is full scale at the output of stage 3 and as shown in Figure 16-9 to Figure 16-12, this stage has very good linearity. (c) Semtech 2006 www.semtech.com 16-22 XE8805/05A Figure 16-9. Integral non-linearity of the ADC and with gain of 1 (PGA1 and PGA2 disabled, PGA3=1, reference voltage of 5V) Figure 16-10. Integral non-linearity of the ADC and gain of 2 (PGA1 and PGA2 disabled, PGA3=2 reference voltage of 5V) Figure 16-11. Integral non-linearity of the ADC and gain of 5 (PGA1 and PGA2 disabled, PGA3=5, reference voltage of 5V) (c) Semtech 2006 www.semtech.com 16-23 XE8805/05A Figure 16-12. Integral non-linearity of the ADC and gain of 10 (PGA1 and PGA2 disabled, PGA3=10, reference voltage of 5V) Figure 16-13. Integral non-linearity of the ADC and gain of 20 (PGA1 and PGA2=10, PGA3=2, reference voltage of 5V) (c) Semtech 2006 www.semtech.com 16-24 XE8805/05A Figure 16-14. Integral non-linearity of the ADC and gain of 50 (PGA1 disabled, PGA2=10, PGA3=5, reference voltage of 5V) Figure 16-15. Integral non-linearity of the ADC and gain of 100 (PGA1=10 and PGA3=10, PGA2 disabled, reference voltage of 5V) 16.8.3.2 Differential non-linearity The differential non-linearity is generated by the ADC. The PGA does not add differential non-linearity. Figure 16-16 shows the differential non-linearity. (c) Semtech 2006 www.semtech.com 16-25 XE8805/05A Figure 16-16. Differential non-linearity of the ADC converter. 16.8.4 Noise Ideally, a constant input voltage VIN should result in a constant output code. However, because of circuit noise, the output code may vary for a fixed input voltage. Thus, a statistical analysis on the output code of 1200 conversions for a constant input voltage was performed to derive the equivalent noise levels of PGA1, PGA2, and PGA3. The extracted rms output noise of PGA1, 2, and 3 are given in Table 16-27: standard output deviation and output rms noise voltage. Figure 16-17 shows the distribution for the ADC alone (PGA1, 2, and 3 bypassed). Quantization noise is dominant in this case, and, thus, the ADC thermal noise is below 16 bits. The simple noise model of Figure 16-18(a) is used to estimate the equivalent input referred rms noise VN,IN of the acquisition chain in the model of Figure 16-18(b). This is given by the relationship: V N , IN = 2 (V N 1 / GD1 ) 2 + (V N 2 /(GD1 GD2 )) 2 + (V N 3 /(GD1 GD2 GD3 )) 2 (OSR N ELCONV ) (V2rms) (Eq. 21) where VN1, VN2, and VN3 are the output rms noise figures of Table 16-27, GD1, GD2, and GD3 are the PGA gains of stages 1 to 3 respectively. As shown in this equation, noise can be reduced by increasing OSR and NELCONV (increases the ADC averaging effect, but reduces noise). Parameter Standard deviation at ADC output (LSB) Output rms noise (V) 1 PGA1 PGA2 PGA3 0.85 1.4 1.5 205 (VN1) 340 (VN2) 365 (VN3) Note: see noise model of Figure 16-18 and equation Eq. 21. Table 16-27. PGA noise measurements (n = 16 bits, OSR = 512, NELCONV = 2, VREF = 5V) (c) Semtech 2006 www.semtech.com 16-26 XE8805/05A Occurences [% of total samples] 80 60 40 20 0 -5 -4 -3 -2 -1 0 1 2 3 4 5 Output Code Deviation From Mean Value [LSB] Figure 16-17. ADC noise (PGA1, 2 & 3 bypassed, OSR=512,NELCONV=2) fS PGA1 VN1 GD1 PGA2 VN2 GD2 PGA3 VN3 GD3 ADC (a) fS PGA1 PGA2 PGA3 GD1 GD2 GD3 VN,IN ADC (b) Figure 16-18. (a) Simple noise model for PGAs and ADC and (b) total input referred noise As an example, consider the system where: GD2 = 10 (GD1 = 1; PGA3 bypassed), OSR = 512, NELCONV = 2, VREF = 5V. In this case, the noise contribution VN1 of PGA1 is dominant over that of PGA2. Using equation Eq. 21, we get: VN,IN = 6.4V (rms) at the input of the acquisition chain, or, equivalently, 0.85 LSB at the output of the ADC. Considering a 0.2V (rms) maximum signal amplitude, the signal-to-noise ratio is 90dB. Noise can also be reduced by implementing a software filter. By making an average on a number of subsequent measurements, the apparent noise is reduced the square root of the number of measurement used to make the average. 16.8.5 Gain Error and Offset Error Gain error is defined as the amount of deviation between the ideal transfer function (theoretical equation Eq. 18) and the measured transfer function (with the offset error removed). The actual gain of the different stages can vary depending on the fabrication tolerances of the different elements. Although these tolerances are specified to a maximum of 3%, they will be most of the time around 0.5%. Moreover, the tolerances between the different stages are not correlated and the probability to get the maximal error in the same direction in all stages is very low. Finally, these gain errors can be calibrated by the software at the same time with the gain errors of the sensor for instance. Figure 16-19 shows gain error drift vs. temperature for different PGA gains. The curves are expressed in % of FullScale Range (FSR) normalized to 25C. (c) Semtech 2006 www.semtech.com 16-27 XE8805/05A Offset error is defined as the output code error for a zero volt input (ideally, output code = 0). The offset of the ADC and the PGA1 stage are completely suppressed if NELCONV > 1. The measured offset drift vs. temperature curves for different PGA gains are depicted in Figure 16-20. The output offset error, expressed in LSB for 16-bit setting, is normalized to 25C. Notice that if the ADC is used alone, the output offset error is below 1 LSB and has no drift. NORMALIZED TO 25C Gain Error [% of FSR] 0.2 0.1 0.0 -0.1 1 5 20 100 -0.2 -0.3 -0.4 -50 -25 0 25 50 75 100 Temperature [C] Figure 16-19. Gain error vs. temperature for different PGA gains Output Offset Error [LSB] NORMALIZED TO 25C 100 1 5 20 100 80 60 40 20 0 -20 -40 -50 -25 0 25 50 75 100 Temperature [C] Figure 16-20. Offset error vs. temperature for different PGA gains 16.8.6 Power Consumption Figure 16-21 plots the variation of quiescent current consumption with supply voltage VDD, as well as the distribution between the 3 PGA stages and the ADC (see Table 16-28). As shown in Figure 16-22, if lower sampling frequency is used, the quiescent current consumption can be lowered by reducing the bias currents of the PGAs and the ADC with registers IB_AMP_PGA [1:0] and IB_AMP_ADC [1:0]. (In Figure 16-22, IB_AMP_PGA/ADC[1:0] = '11', '10', '00' for fS = 500, 250, 62.5kHz respectively.) Quiescent current consumption vs. temperature is depicted in Figure 16-23, showing a relative increase of nearly 40% between -45 and +85C. Figure 16-24 shows the variation of quiescent current consumption for different frequency settings of the internal RC oscillator. It can be seen that the quiescent current varies by about 20% between 100kHz and 2MHz. (c) Semtech 2006 www.semtech.com 16-28 XE8805/05A 800 Quiescent Current - IQ [ A] 700 PGA1, 2 & 3 + ADC 600 500 PGA1 & 2 + ADC 400 PGA1 + ADC 300 200 No PGAs, ADC only 100 2.5 3.0 3.5 4.0 4.5 5.0 5.5 Supply Voltage - VDDA [V] Figure 16-21. Quiescent current consumption vs. supply voltage 800 Sampling Frequency fS : 500kHz Quiescent Current - IQ [ A] 700 600 250kHz 500 400 300 62.5kHz 200 100 2.5 3.0 3.5 4.0 4.5 5.0 5.5 Supply Voltage - VDDA [V] Figure 16-22. Quiescent current consumption vs. supply voltage for different sampling frequencies 20 Relative Quiescent Current Change IQ / IQ,25C [%] Quiescent Current - IQ [ A] 900 850 800 750 700 650 600 550 15 10 5 0 -5 -10 -15 -20 -25 500 -50 -25 0 25 50 75 100 125 -50 -25 0 25 50 75 100 125 Temperature [C] Temperature [C] (a) (b) Figure 16-23. (a) Absolute and (b) relative change inquiescent current consumption vs. temperature (c) Semtech 2006 www.semtech.com 16-29 XE8805/05A Supply VDD = 5V VDD = 3V ADC PGA1 PGA2 PGA3 TOTAL Unit 250 190 165 150 130 120 175 160 720 620 A A 15 850 10 800 Quiescent Current - IQ [ A] Relative Quiescent Current Change IQ / IQ,2MHz [%] Table 16-28. Typical quiescent current distributions in acquisition chain (n = 16 bits, fS = 500kHz) 5 0 -5 -10 -15 750 700 650 600 550 500 -20 0 500 1000 1500 2000 2500 3000 0 3500 500 1000 1500 2000 2500 3000 3500 Frequency - fRC [kHz] Frequency - fRC [kHz] (a) (b) Figure 16-24. (a) Absolute and (b) relative change in quiescent curent consumption vs. RC oscillator frequency (all PGAs active, VDD = 5V) 16.8.7 Power Supply Rejection Ratio Figure 16-25 shows power supply rejection ratio (PSRR) at 3V and 5V supply voltage, and for various PGA gains. PSRR is defined as the ratio (in dB) of voltage supply change (in V) to the change in the converter output (in V). PSRR depends on both PGA gain and supply voltage VDD. 105 100 VDD=3V VDD=5V PSRR [dB] 95 90 85 80 75 70 65 60 1 5 10 20 100 PGA Gain [V/V] Figure 16-25. Power supply rejection ratio (PSRR) (c) Semtech 2006 www.semtech.com 16-30 XE8805/05A Supply VDD = 5V VDD = 3V GAIN = 1 GAIN =5 GAIN = 10 GAIN = 20 GAIN =100 Unit 79 72 78 79 100 90 99 90 97 86 dB dB Table 16-29. PSRR (n = 16 bits, VIN = VREF = 2.5V, fS = 500kHz) 16.9 Application Hints 16.9.1 Input Impedance The PGAs of the acquisition chain employ switched-capacitor techniques. For this reason, while a conversion is done, the input impedance on the selected channel of the PGAs is inversely proportional to the sampling frequency fS and to stage gain as given in equation 22. Z in 768 10 9 Hz f s gain (Eq. 22) The input impedance observed is the input impedance of the first PGA stage that is enabled or the input impedance of the ADC if all three stages are disabled. PGA1 (with a gain of 10), PGA2 (with a gain of 10) and PGA3 (with a gain of 10) each have a minimum input impedance of 150k at fS = 512kHz (see Specification Table). Larger input impedance can be obtained by reducing the gain and/or by reducing the sampling frequency. Therefor, with a gain of 1 and a sampling frequency of 100kHz, Zin > 7.6M. The input impedance on channels that are not selected is very high (>100M). 16.9.2 PGA Settling or Input Channel Modifications PGAs are reset after each writing operation to registers RegAcCfg1-5. Similarly, input channels are switched after modifications of AMUX[4:0] or VMUX. To ensure precise conversion, the ADC must be started after a PGA or inputs common-mode stabilization delay. This is done by writing bit START several cycles after PGA settings modification or channel switching. Delay between PGA start or input channel switching and ADC start should be equivalent to OSR (between 8 and 1024) number of cycles. This delay does not apply to conversions made without the PGAs. If the ADC is not settled within the specified period, there is most probably an input impedance problem (see previous section). 16.9.3 PGA Gain & Offset, Linearity and Noise Hereafter are a few design guidelines that should be taken into account when using the ZoomingADCTM: 1) 2) 3) 4) Keep in mind that increasing the overall PGA gain, or "zooming" coefficient, improves linearity but degrades noise performance. Use the minimum number of PGA stages necessary to produce the desired gain ("zooming") and offset. Bypass unnecessary PGAs. For high gains (>50), use PGA stage 1. For low gains (<50) use stages 2 and 3. For the lowest noise, set the highest possible gain on the first (front) PGA stage used in the chain. For example, in an application where a gain of 20 is needed, set the gain of PGA2 to 10, set the gain of PGA3 to 2. (c) Semtech 2006 www.semtech.com 16-31 XE8805/05A 4) For highest linearity and lowest noise performance, bypass all PGAs and use the ADC alone (applications where no "zooming" is needed); i.e. set ENABLE[3:0] = '0001'. For low-noise applications where power consumption is not a primary concern, maintain the largest bias currents in the PGAs and in the ADC; i.e. set IB_AMP_PGA[1:0] = IB_AMP_ADC[1:0] = '11'. For lowest output offset error at the output of the ADC, bypass PGA2 and PGA3. Indeed, PGA2 and PGA3 typically introduce an offset of about 5 to 10 LSB (16 bit) at their output. Note, however, that the ADC output offset is easily calibrated out by software. 5) 6) 16.9.4 Frequency Response The incremental ADC is an over-sampled converter with two main blocks: an analog modulator and a low-pass digital filter. The main function of the digital filter is to remove the quantization noise introduced by the modulator. As shown in Figure 16-26, this filter determines the frequency response of the transfer function between the output of the ADC and the analog input VIN. Notice that the frequency axes are normalized to one elementary conversion period OSR/fS. The plots of Figure 16-26 also show that the frequency response changes with the number of elementary conversions NELCONV performed. In particular, notches appear for NELCONV 2. These notches occur at: f NOTCH (i ) = i fS OSR N ELCONV for i = 1,2,..., ( N ELCONV - 1) (Hz) (Eq. 23) and are repeated every fS/OSR. 1.2 1 Normalized Magnitude [-] Normalized Magnitude [-] Information on the location of these notches is particularly useful when specific frequencies must be filtered out by the acquisition system. For example, consider a 5Hz-bandwidth, 16-bit sensing system where 50Hz line rejection is needed. Using the above equation and the plots below, we set the 4th notch for NELCONV = 4 to 50Hz, i.e. 1.25fS/OSR = 50Hz. The sampling frequency is then calculated as fS = 20.48kHz for OSR = 512. Notice that this choice yields also good attenuation of 50Hz harmonics. NELCONV = 1 0.8 0.6 0.4 0.2 0 1.2 1 NELCONV = 2 0.8 0.6 0.4 0.2 0 0 1 2 3 4 0 1.2 1 N ELCONV = 4 0.8 0.6 0.4 0.2 0 0 1 2 3 1 2 3 4 Normalized Frequency - f *(OSR/fS) [-] Normalized Magnitude [-] Normalized Magnitude [-] Normalized Frequency - f *(OSR/fS) [-] 4 1.2 NELCONV = 8 1 0.8 0.6 0.4 0.2 0 0 1 2 3 4 Normalized Frequency - f *(OSR/fS) [-] Normalized Frequency - f *(OSR/fS) [-] Figure 16-26. Frequency response: normalized magnitude vs. frequency for different NELCONV (c) Semtech 2006 www.semtech.com 16-32 XE8805/05A 16.9.5 Power Reduction The ZoominADCTM is particularly well suited for low-power applications. When very low power consumption is of primary concern, such as in battery operated systems, several parameters can be used to reduce power consumption as follows: 1) 2) 3) 4) 5) Operate the acquisition chain with a reduced supply voltage VDD. Disable the PGAs which are not used during analog-to-digital conversion with ENABLE[3:0]. Disable all PGAs and the ADC when the system is idle and no conversion is performed. Use lower bias currents in the PGAs and the ADC using the control words IB_AMP_PGA[1:0] and IB_AMP_ADC[1:0]. (This reduces the maximum sampling frequency according to Table 16-26.) Reduce internal RC oscillator frequency and/or sampling frequency. Finally, remember that power reduction is typically traded off with reduced linearity, larger noise and slower maximum sampling speed. (c) Semtech 2006 www.semtech.com 16-33 XE8805/05A 17. Vmult (Voltage Multiplier) 17.1 Features 17-2 17.2 Overview 17-2 17.3 Control register 17-2 17.4 External component 17-2 (c) Semtech 2006 www.semtech.com 17-1 XE8805/05A 17.1 Features * * Generates a voltage that is higher or equal to the supply voltage. Can be easily enabled or disabled 17.2 Overview The Vmult block generates a voltage (called "Vmult") that is higher or equal to the supply voltage. This output voltage is used in the acquisition chain. The voltage multiplier should be on (bit ENABLE in RegVmultCfg0) when using the acquisition chain or analog properties of the Port B while VBAT is below 3V. If the multiplier is enabled, the external capacitor on the pin VMULT is mandatory. The source clock of Vmult is selected by FIN[1:0] in RegVmultCfg0. It is strongly recommended to use the same settings as in the ADC. 17.3 Control register There is only one register in the Vmult. Table 177-1 describes the bits in the register. Pos. 2 RegVmultCfg0 Enable rw rw Reset 0 resetsystem 1-0 Fin rw 0 resetsystem Function enable of the vmult `1' : enabled `0' : disabled system clock division factor `00' : 1/2, `01' : 1/4, `10' : 1/16, `11' : 1/64 Table 177-1. RegVmultCfg0 17.4 External component When the multiplier is enabled, a capacitor has to be connected to the VMULT pin. If the multiplier is disabled, the pin may remain floating. Min. Max. 1.0 3.0 Capacitor on VMULT (c) Semtech 2006 Note nF www.semtech.com 17-2 XE8805/05A 18. Signal D/A (DAS) 18.1 Features 18-2 18.2 Overview of Signal DAC - The generic DAC 18-2 18.3 Registers Map 18-3 18.4 18.4.1 18.4.2 18.4.3 The D/A description What is a noise shaper ? Advantages/disadvantages D/A setup and resolution 18-4 18-4 18-4 18-5 18.5 Amplifier 18-7 18.6 18.6.1 18.6.2 Low pass filter First order low pass filter Second order low pass filter 18-8 18-8 18-9 18.7 18.7.1 18.7.2 4-20mA loop 2-wire loop with first order filtering 2-wire loop with second order filtering 18-10 18-10 18-12 (c) Semtech 2006 www.semtech.com 18-1 XE8805/05A 18.1 Features * * * * * * * 16-bits maximum input word width Synchronization mechanism to guarantee data integrity when writing LSB and MSB 8-bits data Programmable noise shaper order: second, first or order zero Programmable PWM modulation between 4 and 11-bits Programmable clock input frequency: fin or fin/2 Programmable output polarity: active high or low On chip amplifier for analog filtering, voltage output or 4-20 mA loop 18.2 Overview of Signal DAC - The generic DAC The generic DAC block consists of two major parts: the noise shaper (sigma-delta modulator) and the PWM modulator as shown in Figure 18-1. CPU bus PWM modulator Sigma-delta modulator control 16 DAS Out 4-11 m amp Figure 18-1. General block diagram A D/A converter that is built with a digital PWM modulator needs a high clock frequency for a small signal bandwidth. For a 10 bit digital PWM modulator for instance, a 10 bit counter is needed in order to create a pulse with a resolution of 1024. This means that, in case an infinitely sharp analog output filter is used, the clock frequency has to be at least 1024 times the output bandwidth. In practice however, in order to be able to build the analog filter, the clock frequency needs to be much higher. In order to reduce this frequency requirement, the input digital word is broken down into n words with a smaller width m by a noise shaper so that the "average" (average for first order noise shaper, more complicated for higher order noise shapers) value of the n m-bit words represents the full width input code. Instead of 1 pulse with the full resolution, the PWM modulator now generates n pulses with a smaller resolution m. This increases the output pulse repetition frequency with a factor n for identical clock frequency. Therefore, the analog output filtering is easier to implement. Higher order noise shapers (order >1) allow to decrease the clock frequency for identical signal bandwidth. Another advantage is that the signal distortion is less dependent on the signal value. A disadvantage is however, that the output signal after filtering is more dependent on the rise and fall times of the PWM output since there are many more pulses. The maximum word width at the input is 16-bit. If the word is narrower, 0's have to be added after the LSB. In order to maintain maximum flexibility, the order of the noise shaper and the resolution of the PWM modulation are programmable by writing the codes CodeLmax and NsOrder to the configuration register. The possible noise shaper order is 0 (which means no noise shaping), 1 or 2. The possible PWM modulation resolution m can be set between 4 and 11. (c) Semtech 2006 www.semtech.com 18-2 XE8805/05A 18.3 Registers Map All registers are reset with the system reset. The contents of the registers RegDasInLsb and RegDasInMsb are transferred to the D/A converter when after data have been written into RegDasInMsb. Therefore, in order to maintain the synchronisation between the LSB and MSB, the LSB should always be written before the MSB. Pos. 7-0 RegDasInLsb DasInLsb(7:0) rw rw reset 0 resetsystem function Data to convert LSB Table 18-1. RegDasInLsb Pos. 7-0 RegDasInMsb DasInMsb(7:0) rw rw reset 0 resetsystem function Data to convert MSB Table 18-2. RegDasInMsb Pos. 7:6 RegDasCfg0 NsOrder(1:0) rw rw reset 00 resetsystem 5:3 CodeLmax(2:0) rw 000 resetsystem 2:1 Enable(1:0) rw 0 Fin rw 00 resetsystem 0 resetsystem function Noise Shaper order 00 : order 0 01 : order 1 1x : order 2 PWM pulse resolution : 000 : 4 bits 001 : 5 bits 010 : 6 bits 011 : 7 bits 100 : 8 bits 101 : 9 bits 110 : 10 bits 111 : 11 bits Bit 0 : enables the D/A Bit 1 : enables the amplifier Input frequency of modulator as a fraction of oscillator frequency 0 : 1.fRC, 1 : 1/2.fRC Table 18-3. RegDasCfg0 Pos. 7:2 1 RegDasCfg1 BW rw rw rw reset 000000 0 resetsystem 0 INV rw 0 resetsystem function Unused Amplifier bandwidth 0 : small bandwidth 1 : large bandwidth Inverts the PWM output 0 : normal, active high 1 : inverted, active low Table 18-4. RegDasCfg1 (c) Semtech 2006 www.semtech.com 18-3 XE8805/05A 18.4 The D/A description The D/A converter consists of 2 parts: a classic PWM modulator which is preceded by a noise shaper (Figure 18-1). The PWM signal has then to be low pass filtered using the amplifier and external components to obtain the analog signal. 18.4.1 What is a noise shaper? The major disadvantage of using a PWM modulator to generate a high resolution analog signal is that it requires a high ratio between the PWM switching frequency and the useful bandwidth of the output analog signal after low pass filtering. Example: assuming the switching frequency of the PWM modulator is 1MHz and one wants to resolve 16 bit, i.e. 216=65536 steps. In this case, the PWM has to code each step in increments of 1s=1/1MHz and needs therefore 65536s per pulse. This means that the PWM pulse repetition rate is 1/65536s=15.25Hz. So, even with a higher order low pass filter, more than 1 frequency decade will be required to filter the PWM signal down to a 16 bit accurate analog signal. This leaves a useful bandwidth below 1Hz. The goal of the noise shaper is to reduce the ratio between the PWM switching frequency and the useful bandwidth. The noise shaper will not reduce the "truncation noise" and "PWM modulation noise" but move it to higher frequencies. It "shapes" the frequency spectrum ("noise") of the generated PWM signal, hence its name. In practice, the noise shaper allows the generation of a signal with a given resolution using a PWM modulator that has a lower resolution. The noise shaper then generates a series of different subsequent low resolution codes for the PWM so that the average value corresponds to the high resolution code. The first order noise shaper interpolates between two adjacent PWM codes to obtain a higher resolution. The second order noise shaper can use non-adjacent PWM codes. Example for first order noise shaper: assuming again the resolution of 16 bits using a 1MHz PWM switching frequency using the noise shaper with order 1. If a PWM modulator with 4 bits, i.e. 16 steps is used, the PWM repetition frequency becomes then 1MHz/16=62.5kHz. The PWM modulator can convert only the 4 MSB's of the 16 bit input such as h0000, h1000 until hF000. In order to convert the code h5800, which is between h5000 and h6000? In this case, the first order noise shaper will interpolate by presenting alternatively the code h5 and h6 to the PWM so that after filtering a signal is obtained halfway between the normal PWM steps. To convert the code h5400, it will present h5 3 times and h6 once to the PWM and so on. It is clear from this that the PWM repetition frequency is much higher than for the simple PWM and can be filtered out more easily. The quantization noise frequency will depend on the code to be converted: for this example for instance we need two PWM pulses to implement the code h5800, but we need four to implement h5400 etc. Example for second order noise shaper: if we use the same conditions as for the example above, we will obtain the same PWM repetition frequency. However, to implement the code h5400, the noise shaper now can present the following sequence to the PWM modulator: h6, h5, h6, h4. This increases the frequency components at the PWM pulse repetition frequency and 1/2 of the PWM pulse repetition frequency but at the same time reduces energy at 1/4 of the PWM pulse repetition frequency with respect to the first order noise shaper. The low pass cut-off frequency can therefore be higher than for a first order noise shaper. A disadvantage of the second order noise shaper is however that the resolution will drop when the code is very close to h0000 or hFFFF. Example: if we assume the same conditions as above, but we want to convert the code h0400. It is now impossible to use a similar sequence as above (which would be h1, h0, h1, h(-1) ) due to saturation of the code. There is no choice left but the sequence h1 h0 h0 h0 which is the same sequence as in the first order noise shaper. 18.4.2 Advantages/disadvantages Advantages: Using a high order noise shaper together with a PWM modulator with low resolution reduces the ratio between the low pass cut off frequency and the PWM switching frequency for the same total resolution. This can be used to (c) Semtech 2006 www.semtech.com 18-4 XE8805/05A increase the output signal bandwidth or to reduce the PWM switching frequency and therefore the power consumption of the D/A. Signal distortion is less dependent on the signal value. Disadvantages: Using a high order noise shaper together with a PWM modulator with low resolution will use lots of short pulses in stead of 1 long pulse. The D/A is therefore more sensitive to rise and fall times of the PWM resulting in a slightly higher non-linearity and temperature dependence. The second order noise shaper also has a reduced resolution for codes very close to zero or full scale. 18.4.3 D/A setup and resolution In this section, the resolution that can be obtained with the D/A as a function of settings is calculated. These calculations are based on the quantization and PWM modulation noise. Noise on the reference, i.e. the supply voltage is not taken into account. High frequency noise on the supply voltage can be filtered by the output low pass filter, but in band noise on the reference will show up in the output signal with amplitude that will depend on the signal value. Therefore, when using the D/A, one should take care to minimize the switching activity on the digital ports and/or to limit the load on these ports. 18.4.3.1 Noise shaper of order 0 Setting the noise shaper to order 0 (NsOrder=00), reduces the D/A to a regular PWM. Two parameters are setting the resolution of the D/A: the resolution of the modulator itself and the amount of low pass filtering at the output. The modulation width m of the PWM modulator is given by: m = 4 + CodeLmax The cut-off frequency f c of the low pass filter required to get the resolution is calculated below. The PWM modulator repetition frequency f PWM can be calculated as a function of the selected modulation width m , the frequency of the RC oscillator of the circuit f RC and the selected frequency division set by Fin : f PWM 1 f RC 1 + Fin = 2m To obtain an analog signal with the required solution, the PWM signal has to be low pass filtered. The resolution that can be obtained depends on the filter order and the ratio between the PWM modulation frequency f PWM and the filter cut-off frequency f c . For a low pass filter of LpOrder, we obtain: f resolution PWM = LpOrder log 2 PWM fc The total resolution of the D/A is then the minimal value of both criteria: resolution = min(m, resolution PWM ) In Table 18-5 the required cut-off frequency of the low pass filter is shown for a noise shaper of order 0 as a function of the desired resolution for both a first and second order low pass filter. The PWM modulation factor m should be chosen equal to the desired resolution. (c) Semtech 2006 www.semtech.com 18-5 XE8805/05A resolution (bit) 4 5 6 7 8 9 10 11 m 4 5 6 7 8 9 10 11 fc for LpOrder=1 (Hz) 7812 1953 488 122 30 7.6 1.9 0.48 fc for LpOrder=2 (Hz) 31250 11048 3906 1381 488 172 61 22 Table 18-5. Signal bandwidth as a function of the required resolution for the PWM without noise shaper (Fin=0, NsOrder=00, fRC=2MHz). 18.4.3.2 Noise shaper of order 1 or 2 The calculation on the required low pass cut-off frequency given in 18.4.3.1 remains valid in this case. However, the noise shaper allows using smaller PWM modulation for the same resolution. This increases the PWM modulation frequency and as a consequence increases the output bandwidth. An additional criterion however shows up: the filtering of the quantization noise. As can be seen from the examples in 18.4.1, the interpolation between PWM codes generated by the noise shaper introduce sequences at frequencies below the PWM modulation frequency. Assuming a low pass filter that has at least the same order as the noise shaper, the resolution is given by (NsOrder1) : f resolution quant = 0.359 + m + NsOrder log 2 ( PWM ) - 2.65 fc The total resolution of the D/A is then the minimal of both criteria: resolution = min(resolutionquant , resolution PWM ) Table 18-6 and Table 18-7 show the signal bandwidth that can be obtained as a function of required resolution and PWM modulation for first and second order noise shapers. It can be seen that these options are useful to obtain high resolution using low PWM modulation m. For high PWM modulation m, the resolution is limited by the PWM modulator and adding a noise shaper does not change anything. NsOrder=1, fRC=2MHz, Fin=0, LpOrder=2 Resolution PWM modulation m (bit) 4 5 6 8 1596.4 1596.4 1596.4 9 798.2 798.2 798.2 10 399.1 399.1 399.1 11 199.5 199.5 199.5 12 99.8 99.8 99.8 13 49.9 49.9 49.9 14 24.9 24.9 24.9 15 12.5 12.5 12.5 16 6.2 6.2 6.2 7 976.6 690.5 399.1 199.5 99.8 49.9 24.9 12.5 6.2 8 488.3 345.3 244.1 172.6 99.8 49.9 24.9 12.5 6.2 9 244.1 172.6 122.1 86.3 61.0 43.2 24.9 12.5 6.2 10 122.1 86.3 61.0 43.2 30.5 21.6 15.3 10.8 6.2 11 61.0 43.2 30.5 21.6 15.3 10.8 7.6 5.4 3.8 Table 18-6. Low pass cut-off frequency as a function of the selected PMW modulation and required resolution for a first order noise shaper. (c) Semtech 2006 www.semtech.com 18-6 XE8805/05A NsOrder=2, fRC=2MHz, Fin=0, LpOrder=2 Resolution PWM modulation m (bit) 4 5 6 8 5638.4 3906.3 1953.1 9 3986.9 2762.1 1381.1 10 2819.2 1953.1 976.6 11 1993.5 1381.1 690.5 12 1409.6 976.6 488.3 13 996.7 690.5 345.3 14 704.8 488.3 244.1 15 498.4 345.3 172.6 16 352.4 244.1 122.1 7 976.6 690.5 488.3 345.3 244.1 172.6 122.1 86.3 61.0 8 488.3 345.3 244.1 172.6 122.1 86.3 61.0 43.2 30.5 9 244.1 172.6 122.1 86.3 61.0 43.2 30.5 21.6 15.3 10 122.1 86.3 61.0 43.2 30.5 21.6 15.3 10.8 7.6 11 61.0 43.2 30.5 21.6 15.3 10.8 7.6 5.4 3.8 Table 18-7. Low pass cut-off frequency as a function of the selected MPW modulation and required resolution for a second order noise shaper. The output range of the D/A is for code 0h0000 is VSS and for code 0hFFFF is (VBAT-VSS)(2m-1)/2m. 18.5 Amplifier The amplifier can be used to implement the low pass filter and/or a 4-20mA loop. The amplifier is enabled using the bit Enable(1)=1. The amplifier has two different modes selected by the bit BW: a low frequency mode (BW=0) that allows driving a high capacitive load and a high frequency mode (BW=1). The first mode is particularly adapted when a voltage output is used. The second mode is more adapted for a 420mA loop since loads are small and higher bandwidth is required to reject current consumption changes in the loop. Table 18-8 shows the specification of the amplifier. Note that the amplifier can not be used to generate signals that are larger than the supply voltages VBAT and VSS since the amplifier inputs and outputs are clamped to these voltages. The amplifier inputs and outputs should stay within the input and output ranges specified below. (c) Semtech 2006 www.semtech.com 18-7 XE8805/05A sym gain GBW0 CL0 GBW1 CL1 m RL SR CMR OR Voff CMRR noise PSRR Iquie Ioff 1. 2. 3. 4. 5. 6. 7. 8. 9. description gain at DC gain bandwidth product capacitive load gain bandwidth product capacitive load phase margin resistive load slew rate common mode input range output range offset common mode rejection integrated input noise power supply rejection ratio quiescent bias current off current min 80 25 typ 100 70 max 5 250 450 55 5 10 VSS-0.2 VSS+0.2 65 200 30 VBAT-1.2 VBAT-0.2 5 60 20 50 60 100 150 1 unit dB kHz nF kHz pF k kV/s V V mV dB uVrms dB Comment 1 6 6 7 7 8 5 9 2 3 4 uA uA For the minimal resistive load and the maximal capacitive load The amplifier common mode is VSS in the 4-20mA loop. At DC At DC. Only a low rejection ratio is needed since the D/A output refers directly to the power supplies. Short circuit protection at >3mA. GBW when the maximal load is cl0 and with the bit BW=0 GBW when the maximal load is cl1 and with the bit BW=1 In both cases BW=0 and BW=1 for the maximal capacitive load and the minimal resistive load. For maximal load CL0, BW=0 and maximal resistive load RL Table 18-8. Specification of the amplifier. 18.6 Low pass filter Several low pass filters are proposed here as examples. Other filter types are possible depending on the features or constraints of the application. If the filter is inverting the signal, the bit INV can be used to invert the D/A output. This inversion does not need to be done by calculation. A first or second order low pas filter can be built with the amplifier. If higher order filters are needed, additional first or second order sections can be added using external amplifiers. 18.6.1 First order low pass filter Figure 18-2 shows a possible implementation of a first order low pass filter. Ideally, the analog ground should be halfway between VBAT and VSS. The gain G and cut-off frequency fc of such a filter are given by: G= R2 R1 fc = 1 2R2 C As an example, to obtain a 1kHz filter with unity gain, we can choose C=1nF and R1=R2=150k. (c) Semtech 2006 www.semtech.com 18-8 XE8805/05A XE88xx c o n t r o l D DAS_OUT A R1 DAS_AI_M R2 amp C DAS_AO DAS_AI_P analog ground Figure 18-2. First order low pass filter. 18.6.2 Second order low pass filter Figure 18-3 shows an example of a second order low pass filter using the multi-feedback architecture. The gain G, cut-off frequency fc and the damping factor (or quality factor Q) as a function of the factors k and m (see Figure 18-3) are given by: G = -k 1 (n + 1) km = 2Q 2 n 1 fc = 2RC kmn = For a second order Butterworth filter, = 2 2 . For smaller damping factors, the filter is under damped resulting in overshoots on the step response. For higher damping factors, the filter is over damped resulting in a smooth but slower step response. An example of a 1dB ripple Chebychev filter with a cut-off frequency of about 1.5kHz and a DC gain of 1 is given by choosing m=0.22, k=1, n=0.5, R=330k and C=1nF. The resistor nR can be rounded to 180k. A 60Hz unity gain low pass Butterworth filter can be built choosing R=180k, C=12nF, k=1, m=0.183, n=8.33. Note that parasitic capacitors between the DAS_OUT node and the filter output DAS_AO will adversely affect the high frequency behavior of the filter. Care should be taken when routing these signals. (c) Semtech 2006 www.semtech.com 18-9 XE8805/05A XE88xx c o n t r o l D DAS_OUT A R C nR DAS_AI_M DAS_AO mC kR amp DAS_AI_P analog ground Figure 18-3. Second order low pass filter. 18.7 4-20mA loop 18.7.1 2-wire loop with first order filtering The amplifier can be used to build a 4-20mA loop externally. Figure 18-4 shows the principle of such a 2-wire loop using a first order low pass filter. In a 2-wire loop, the current consumption of the sensor and read-out electronics is drawn on the same wires as the signal current. The current consumption of the sensor and read-out electronics should therefore remain below 4mA. The signal current is then added by the bipolar transistor. The resistors Rlim1 and Rlim2 are added to protect the bipolar transistor against high transient currents during power-up. Rlim1 is generally set to a few k. The value of Rlim2 is chosen as a function of the external loop voltage VEXT and the transistor saturation voltage VCEsat. Rlim 2 = (V EXT - VCEsat - Rsense 20mA) 20mA If VEXT is larger than 5.5V, a voltage regulator has to be inserted. Since the quiescent current of the regulator adds up to the 4mA budget, a component with sufficiently low quiescent current has to be selected. The resistor Rsense measures the total current in the loop (if Rsense<<Rf2). The resistors Rf1 and Rf2 are used to set the gain and Rf2 and Cf to set the bandwidth of the filter. The resistor Roffset adds an offset to the filter voltage so that the code 0 of the D/A corresponds to 4mA. The amplifier will force a current through the bipolar transistor so that the voltage on the filter Vf and VSS on Rsense is equal. This transforms the filter voltage into a loop current Iloop=(VSS-VEXT-)/Rsense. (c) Semtech 2006 www.semtech.com 18-10 XE8805/05A The resistor value Rsense is generally chosen between 50 and 150 resulting in a 1V to 3V voltage drop at maximal loop current. The resistor Rf2 is then chosen much larger depending on the current error requirement. Allowing for an error of 0.1% gives Rf2=Rsense/0.001. The resistor Roffset may be omitted but it will reduce the useful code range of the D/A. Using the large bandwidth of the amplifier is recommended since this increases the rejection of supply current variations of the other components in the loop. A bypass capacitor between VBAT and VSS will also reduce the high frequency current variations. Values will depend on the voltage regulator used. The software in the XE8805 should keep the current supply of the circuit as stable as possible. This means that the clock frequency should kept constant, peripherals should not be switched on and off, the current in the sensor is kept constant, the processor should not use the halt or sleep modes, etc. XE88xx 4-20mA Voltage regulator VBAT VEXT+ VBAT c o n t r o l D DAS_OUT Roffset A Rf1 Rlim2 DAS_AI_P Vf amp Rlim1 DAS_AO Sensor DAS_AI_M VSS VSS Cf Rf2 Rsense VEXT4-20mA Figure 18-4. 2-wire 4-20mA loop with first order filter The resistor Rf1 can then be calculated to set the full scale D/A code range (depends on the PWM modulation m, see section 18.4.3.2) equal to the full scale signal current of 16mA: Rf1 (VBAT - VSS ) (2 m - 1) R f 2 2 m Rsense 16mA The resistor Roffset can be calculated in order to obtain a 4mA current while the D/A code is 0: Roffset (VBAT - VSS ) R f 2 Rsense 4mA (c) Semtech 2006 www.semtech.com 18-11 XE8805/05A The capacitor Cf can be calculated from the first order filter cut-off frequency fc : Cf = Rf1 + Rf 2 2f c R f 1 R f 2 As an example, for VBAT-VSS=5V, m=4, Rsense=50, VEXT=30V and a 1kHz low pass filter, we can use: Rlim1=1k, Rlim2=1k, Rf1=560k, Rf2=100k, Roffset=2.4M, Cf=1.8nF. 18.7.2 2-wire loop with second order filtering A second order filter function can be implemented by replacing the resistor Rf1 in Figure 18-4 by another first order filter section as shown in Figure 18-5. The value of Rf1a+Rf2a has to be chosen the same way as Rf1 in the first order schematic. XE88xx VBAT c o n t r o l 4-20mA Voltage regulator VBAT VEXT+ Roffset D DAS_OUT Rf1a A Rf1b Cf1 Vf DAS_AI_P amp Rlim2 Rlim1 DAS_AO Sensor DAS_AI_M VSS VSS Cf2 Rf2 Rsense VEXT4-20mA Figure 18-5. 2-wire 4-20mA loop with second order filtering. Another possibility is shown in Figure 18-6. The advantage this solution is that it is easier to stabilize depending on the component parasitics and board layout. But since it limits the bandwidth of the current regulation loop, it reduces the rejection of the supply current variations. For this schematic, all the equations of the first order schematic remain valid. The values of Rfs and Cfs can be calculated from the cut-off frequency: R fs C fs = 1 2f c For the 1kHz example, we can chose Rfs=150k and Cfs=1nF (set BW=0 in this case). (c) Semtech 2006 www.semtech.com 18-12 XE8805/05A XE88xx VBAT c o n t r o l 4-20mA Voltage regulator VBAT VEXT+ Roffset D DAS_OUT Rf1 A Rlim2 Vf DAS_AI_P amp Rlim1 DAS_AO Sensor Cfs DAS_AI_M VSS Rfs VSS Cf Rf2 Rsense VEXT4-20mA Figure 18-6. 2-wire 4-20mA with second order filter and increased stability (c) Semtech 2006 www.semtech.com 18-13 XE8805/05A 19. Bias D/A (DAB) 19.1 Features 19-2 19.2 General description 19-2 19.3 Register map 19-2 19.4 D/A specification 19-3 19.5 Amplifier specification 19-3 19.6 19.6.1 19.6.2 Application examples Voltage controlled sensor bias Current controlled sensor bias 19-4 19-4 19-4 (c) Semtech 2006 www.semtech.com 19-1 XE8805/05A 19.1 Features * * * 8 bit low frequency A/D On-chip amplifier with 2 terminals MOS output Current and voltage controlled applications can be implemented 19.2 General description XE8805A - Bias D/A block DAB R P DAB OUT DAB AIP control registers D/A VBATT amp DAB AOP MOS VSS DabIn Enable DAB AOM DAB AIM DAB R M As the amplifier is attacking a MOS, effective polarity of the feedback loop depends on load and network. Figure 19-1. General block diagram Figure 19-1 shows the general block diagram of the D/A peripheral. It consists of a control block that manages all communication with the CPU and sets the configuration of the peripheral. The D/A converts the digital data in an analog output signal. An amplifier is added that can be used to drive the large sensor currents. 19.3 Register map The bias D/A has two registers. Pos. 7-0 RegDabIn DabIn(7:0) rw rw reset 0 resetsystem function Data to convert Table 19-1. RegDabIn Pos. 7-2 1-0 RegDabCfg Enable(1:0) rw r rw reset 0 00 resetsystem function Unused bit 1: enables the amplifier when 1 bit 0: enables the D/A when 1 Table 19-2. RegDabCfg (c) Semtech 2006 www.semtech.com 19-2 XE8805/05A 19.4 D/A specification The D/A generates a voltage on node DAB_OUT that is proportional to the code RegDabIn in between the positive reference voltage DAB_R_P and the negative reference voltage DAB_R_M. The voltage on DAB_R_P always has to be larger than the voltage on DAB_R_M. The DAB is intended for very low frequency use. The specification is given in Table 19-3. The DAB is based on a resistive ladder. Capacitors larger than 100pF are allowed on the node DAB_OUT, but the step response will increase proportionally. sym wda tstep OR refp refn Rdab description number of input bits step response D/A output range DAB_R_P range DAB_R_M range impedance between DAB_R_P and DAB_R_M min typ 8 0.25 max 1 DAB_R_P VBAT VBAT-2.3 DAB_R_M VSS+2.3 VSS 1.6 unit bits ms note 1 V V Mohm 1. Time to reach the final value within 5%, CL on DAB_OUT smaller than 100pF. Table 19-3. D/A specification. 19.5 Amplifier specification The amplifier output stage is a single transistor follower that is able to drive large currents. This transistor is not connected internally so that different circuit configurations are possible (see next section). In order to guarantee correct functionality, the voltage on the output pins has to respect the specifications VRAOP and VRAOM as indicated in Table 19-4. sym gain GBW m RL CL CMR VRAOM VRAOP Voff noise description gain at DC gain bandwidth product phase margin resistive load capacitive load common mode input range DAB_AOM voltage range DAB_AOP voltage range offset integrated input noise Isource PSRR Ibias Ioff max source current power supply rejection ratio quiescent bias current off current min 60 100 60 300 typ 90 4500 80 100000 1 VBAT DAB_AOP-0.2 VBAT VSS VSS+0.2 VSS+2.3 10 max 10 60 100 40 80 5 0.1 10 1 unit dB Hz nF V V V mV Vrm s mA dB A A Note 1 1 1 1 1 2 1. For all possible combinations of resistive load and capacitive load. 2. At DC. Table 19-4. Amplifier specification. (c) Semtech 2006 www.semtech.com 19-3 XE8805/05A 19.6 Application examples 19.6.1 Voltage controlled sensor bias Figure 19-2 shows the basic connectivity to have a voltage controlled sensor bias. The D/A will generate a voltage between vrep and vrefn proportional to the input code. The amplifier will copy the D/A voltage to the sensor. The D/A code can be used to do a software temperature calibration of the sensor for instance. Filter capacitors can be added in parallel with the sensor reference and signal, on Vrefp, Vrefn and DAB_OUT. The voltages Vrefp and Vrefn can be filtered before being connected to the D/A reference inputs. The reference voltages can be connected directly to VBAT and VSS for simplicity. They can also be connected to VBAT and VSS through a low pass filter that rejects the high frequency supply noise. Finally, in most cases, the voltage range of interest for the voltage Vsensor on the sensor is only a fraction of the supply voltage. By generating Vrefp and Vrefn equal to the limits of the voltage range of interest, the resolution of the D/A can be increased. Example: if a supply of 5V is used and the reference voltage is equal to the supply, the D/A can generate a sensor voltage between 0V and 5V in steps of about 5V/25520mV. If the sensor voltage is always to be between 3V and 4V, and by connecting Vrefn=3V and Vrefp=4V, the sensor voltage is adjustable between 3V and 4V with steps of about 1V/2554mV. XE8805A - DAB block DAB_R_P DAB_R_M D/A Vrefp VBAT or a Reference voltage source Vrefn VSS DAB_OUT * High impedance node, impedance depends on D/A code DAB_AIP DAB_AOP amp VBAT * Needs a 300 - 100 k load * Max capacitive load is 1 nF Vsensor DAB_AOM DAB_AIM reference input of the ZoomingADC * In this configuration the bridge current and DAB_AOM increase when DAB_OUT increases. VSS signal input of the ZoomingADC Figure 19-2. Voltage controlled bridge bias principle. Note that the voltage on the sensor can not be higher than VBAT-0.2V in the example of Figure 19-2 (specification VRAOM in Table 19-4). 19.6.2 Current controlled sensor bias Figure 19-3 shows the principle of a current controlled sensor bias schematic. In this case, the amplifier forces the voltage VR to be equal to the D/A output voltage VD/A. The current Isensor through the sensor is given by: (c) Semtech 2006 www.semtech.com 19-4 XE8805/05A I sensor = Vrefp - VR Rsense = Vrefp - VD / A Rsense = (Vrefp - Vrefn ) (1 - code 255) Rsense The voltage Vsensor can be calculated as a function of the current Isensor and the sensor impedance. Note that the voltage VR>VSS+2.3V and VR-Vsensor>0.2V (VRAOP and VRAOM specifications in Table 19-4) to guarantee correct functionality of this schematic. Choosing the Vrefp equal to the supply voltage or close to the supply voltage in order to have the highest possible voltage on the sensor is recommended. From the equation, it can be seen that the sensor current step per LSB can be made smaller by reducing the voltage between Vrefp and Vrefn or by increasing the sense resistor value. As for the voltage controlled sensor bias, capacitors can be added on several nodes to filter out the noise. XE8805A DAB_R_P D DAB_R_M * Voltage must remain above VSS + 2.3 V Vrefp Vrefn * Voltage must remain below VBAT - 2.3 V VSS DAB_OUT VD/A DAB_AIM A Rsense * Voltage must remain above VSS + 2.3 V DAB_AOP VR * Needs a 300 - 100 k load * Max capacitive load is 1 nF amp DAB_AOM Vsensor DAB_AIP reference * In this configuration, the bridge current and DAB_AOM decrease when DAB_OUT increases. VSS signal Figure 19-3. Current controlled bridge bias In Figure 19-4, the sense resistor is inserted between the negative reference voltage and the sensor. This schematic has the same principle as above, but it is easier to respect the limits on VRAOP when VBAT is low. The sensor current is now: I sensor = VR - Vrefn Rsense = VD / A - Vrefn Rsense = (Vrefp - Vrefn ) (code 255) Rsense In this case, it is recommended to choose Vrefn equal to VSS or close to VSS in order to have the highest possible voltage on the sensor. The only limit is now Vsensor<VBAT-0.2V. (c) Semtech 2006 www.semtech.com 19-5 XE8805/05A XE8805A DAB_R_P D DAB_R_M Vrefp VBAT or another reference voltage Vrefn * Vrefp must remain above VSS + 2.3 V VSS DAB_OUT VD/A DAB_AIP A VBAT DAB_AOP amp Vsensor * Needs a 300 - 100 k kohm load * Max capacitive load is 1 nF DAB_AOM DAB_AIM reference input of ZoomingADC * In this configuration, the bridge current and DAB_AOM increase when DAB_OUT increases signal input of ZoomingADC VR Rsense VSS Figure 19-4. Current controlled sensor bias. . (c) Semtech 2006 www.semtech.com 19-6 XE8805/05A 20. Counters/Timers/PWM 20.1 Features 20-2 20.2 Overview 20-2 20.3 Register map 20-2 20.4 Interrupts and events map 20-3 20.5 Block schematic 20-4 20.6 General counter registers operation 20-4 20.7 Clock selection 20-5 20.8 Counter mode selection 20-5 20.9 Counter / Timer mode 20-6 20.10 PWM mode 20-8 20.11 Capture function 20-9 (c) Semtech 2006 www.semtech.com 20-1 XE8805/05A 20.1 Features * * * * * * * * * * 4 x 8-bits timer/counter modules or 2 x 16-bits timers/counter modules Each with 4 possible clock sources Up/down counter modes Interrupt and event generation Capture function (internal or external source) Rising, falling or both edge of capture signal PA[3:0] can be used as clock inputs (debounced or direct) 2 x 8 bits PWM or 2 x 16 bits PWM PWM resolution of 8, 10, 12, 14 or 16 bits Complex mode combinations are possible between counter, capture and PWM modes 20.2 Overview CounterA and CounterB are 8-bit counters and can be combined to form a 16-bit counter. CounterC and CounterD exhibit the same features. The counters can also be used to generate two PWM outputs on PB[0] and PB[1]. In PWM mode one can generate PWM functions with 8, 10, 12, 14 or 16 bit wide counters. The counters A and B can be captured by events on an internal or an external signal. The capture can be performed on both 8-bit counters running individually on two different clock sources or on both counters chained to form a 16-bit counter. In any case, the same capture signal is used for both counters. When the counters A and B are not chained, they can be used in several configurations: A and B as counters, A and B as captured counters, A as PWM and B as counter, A as PWM and B as captured counter. When the counters C and D are not chained, they can be used either both as counters or counter C as PWM and counter D as counter. 20.3 Register map Bit 7-0 7-0 RegCntA CounterA CounterA rw r w reset xxxxxxxx xxxxxxxx function 8-bits counter value 8-bits comparison value Table 20-1. RegCntA bit 7-0 7-0 RegCntB CounterB CounterB rw r w reset xxxxxxxx xxxxxxxx function 8-bits counter value 8-bits comparison value Table 20-2. RegCntB Note: When writing to RegCntA or RegCntB, the processor writes the counter comparison values. When reading these locations, the processor reads back either the actual counter value or the last captured value if the capture mode is active. bit 7-0 7-0 RegCntC CounterC CounterC rw r w reset xxxxxxxx xxxxxxxx function 8-bits counter value 8-bits comparison value Table 20-3. RegCntC (c) Semtech 2006 www.semtech.com 20-2 XE8805/05A bit 7-0 7-0 RegCntD CounterD CounterD rw r w reset xxxxxxxx xxxxxxxx function 8-bits counter value 8-bits comparison value Table 20-4. RegCntD Note: When writing RegCntC or RegCntD, the processor writes the counter comparison values. When reading these locations, the processor reads back the actual counter value. bit 7-6 5-4 3-2 1-0 RegCntCtrlCk CntDCkSel(1:0) CntCCkSel(1:0) CntBCkSel(1:0) CntACkSel(1:0) rw rw rw rw rw reset xx xx xx xx function Counter d clock selection Counter c clock selection Counter b clock selection Counter a clock selection Table 20-5. RegCntCtrlCk bit 7 6 5 4 3 2 1 0 RegCntConfig1 CntDDownUp CntCDownUp CntBDownUp CntADownUp CascadeCD CascadeAB CntPWM1 CntPWM0 rw rw rw rw rw rw rw rw rw Reset x x x x x x 0 resetsystem 0 resetsystem function Counter d up or down counting (0=down) Counter c up or down counting (0=down) Counter b up or down counting (0=down) Counter a up or down counting (0=down) Cascade counter c & d (1=cascade) Cascade counter a & b (1=cascade) Activate pwm1 on counter c or c+d (PB(1)) Activate pwm0 on counter a or a+b (PB(0)) Table 20-6. RegCntConfig1 bit 7-6 5-4 3-2 1-0 RegCntConfig2 CapSel(1:0) CapFunc(1:0) Pwm1Size(1:0) Pwm0Size(1:0) rw rw rw rw rw Reset 00 resetsystem 00 resetsystem xx xx function Capture source selection Capture function Pwm1 size selection Pwm0 size selection Table 20-7. RegCntConfig2 bit 7-4 3 2 1 0 RegCntOn -CntDEnable CntCEnable CntBEnable CntAEnable rw r rw rw rw rw Reset 0000 0 resetsystem 0 resetsystem 0 resetsystem 0 resetsystem Function Reserved Enable counter d Enable counter c Enable counter b Enable counter a Table 20-8. RegCntOn 20.4 Interrupts and events map Interrupt source IrqA IrqB IrqC IrqD Mapping in the interrupt manager RegIrqHigh(4) RegIrqLow(5) RegIrqHigh(3) RegIrqLow(4) Mapping in the event manager RegEvn(7) RegEvn(3) RegEvn(6) RegEvn(2) Table 20-9. Interrupt and event mapping. (c) Semtech 2006 www.semtech.com 20-3 XE8805/05A 20.5 Block schematic ck1k ck32k ck128 ckrcext/4 ckrcext RegCntA (write) RegCntA (read) Counter A PA(0) Capture RegCntB (write) RegCntB (read) Counter B PA(1) RegCntC (write) ck1k ck32k RegCntC (read) Counter C PA(2) RegCntD (write) RegCntD (read) Counter D PA(3) PB(0) PWM PB(1) Figure 20-1: Counters/timers block schematic 20.6 General counter registers operation Counters are enabled by CntAEnable, CntBEnable, CntCEnable, and CntDEnable in RegCntOn. To stop the counter X, CntXEnable must be reset. To start the counter X, CntXEnable must be set. When counters are cascaded, CntAEnable and CntCEnable also control respectively the counters B and D. In the control registers, all registers must be written in this order: RegCntCtrlCk, RegCntConfig1, RegCntConfig2 and all RegCntX because several bits have no default values at reset. All counters have a corresponding 8-bit read/write register: RegCntA, RegCntB, RegCntC, and RegCntD. When read, these registers contain the counter value (or the captured counter value). When written, they modify the counter comparison values. For a correct acquisition of the counter value, use one of the three following methods: 1) Stop the concerned counter, perform the read operation and restart the counter. While stopped, the counter content is frozen and the counter does not take into account the clock edges delivered on the external pin. (c) Semtech 2006 www.semtech.com 20-4 XE8805/05A 2) For slow operating counters (typically at least 8 times slower than the CPU clock), oversample the counter content and perform a majority operation on the consecutive read results to select the correct actual content of the counter. 3) Use the capture mechanism. When a value is written into the counter register while the counter is in counter mode, both the comparison value is updated and the counter value is modified. In upcount mode, the register value is reset to zero. In downcount mode, the comparison value is loaded into the counter. Due to the synchronization mechanism between the processor clock domain and the external clock source domain, this modification of the counter value can be postponed until the counter is enabled and it receives it's first valid clock edge. In the PWM mode, the counter value is not modified by the write operation in the counter register. Changing the counter mode, does not update the counter value (no load in downcount mode). 20.7 Clock selection The clock source for each counter can be individually selected by writing the appropriate value in the register RegCntCtrlCk. Table 20-10 gives the correspondence between the binary codes used for the configuration bits CntACkSel(1:0), CntBCkSel(1:0), CntCCkSel(1:0) or CntDCkSel(1:0) and the clock source selected respectively for the counters A, B, C or D. Clock source for CntXCkSel(1:0) 11 10 01 00 CounterA CounterB CounterC CounterD Ck128 CkRc/4 CkRc PA(0) Ck1k Ck32k PA(1) PA(2) PA(3) Table 20-10: Clock sources for counters A, B, C and D The CkRc clock is the RC oscillator. The clocks below 32kHz can be derived from the RC oscillator or the crystal oscillator (see the documentation of the clock block). A separate external clock source can be delivered on Port A for each individual counter. The external clock sources can be debounced or not by setting the Port A configuration registers. The clock source can be changed only when the counter is stopped. 20.8 Counter mode selection Each counter can work in one of the following modes: 1) Counter, downcount & upcount 2) Captured counter, downcount & upcount (only counters A&B) 3) PWM, downcount The counters A and B or C and D can be cascaded or not. In cascaded mode, A and C are the LSB counters while B and D are the MSB counters. Table 20-11 shows the different operation modes of the counters A and B as a function of the mode control bits. For all counter modes, the source of the down or upcount selection is given (either the bit CntADownUp or the bit CntBDownUp). Also, the mapping of the interrupt sources IrqA and IrqB and the PWM output on PB(0) in these different modes is shown. (c) Semtech 2006 www.semtech.com 20-5 XE8805/05A CascadeAB CountPWM0 CapFunc(1:0) 0 0 00 1 0 00 0 1 00 1 1 00 0 0 1 0 0 1 1x or x1 1x or x1 1x or x1 Counter A mode Counter B mode IrqA source IrqB source PB(0) function Counter 8b Counter 8b Downup: A Downup: B Counter 16b AB Downup: A PWM 8b Counter 8b Down Down PWM 10 - 16b AB Down Captured Captured counter 8b counter 8b Downup: A Downup: B Counter A Counter AB Counter B PB(0) - PB(0) - Counter B PWM A - - PWM AB Capture A Capture B PB(0) Captured counter 16b AB Downup: A Capture AB Capture AB PB(0) Captured counter 8b Downup: B Must not be used Capture B PWM A PWM 8b Down Table 20-11: Operating modes of the counters A and B Table 20-12 shows the different operation modes of the counters C and D as a function of the mode control bits. For all counter modes, the source of the down or upcount selection is given (either the bit CntCDownUp or the bit CntDDownUp). The mapping of the interrupt sources IrqC and IrqD and the PWM output on PB(1) in these different modes is also shown. The switching between different modes must be done while the concerned counters are stopped. While switching capture mode on and off, unwanted interrupts can appear on the interrupt channels concerned by this mode change. CascadeCD CountPWM1 0 0 1 0 0 1 1 1 Counter C mode Counter D mode IrqC Source IrqD source PB(1) function Counter 8b Counter 8b Downup: C Downup: D Counter 16b CD Downup: C PWM 8b Counter 8b Down Down PWM 10 - 16b CD Down Counter C Counter CD - Counter D - PB(1) - Counter D - PB(1) PWM C PWM CD Table 20-12: Operating modes of the counters C and D 20.9 Counter / Timer mode The counters in counter / timer mode are generally used to generate interrupts after a predefined number of clock periods applied on the counter clock input. Each counter can be set individually either in upcount mode by setting CntXDownUp in the register RegCntConfig1 or in downcount mode by resetting this bit. Counters A and B can be cascaded to behave as a 16 bit counter by setting CascadeAB in the RegCntConfig1 register. Counters C and D can be cascaded by setting CascadeCD. When cascaded, the up/down count modes of the counters B and D are defined respectively by the up/down count modes set for the counters A and C. (c) Semtech 2006 www.semtech.com 20-6 XE8805/05A When in upcount mode, the counter will start incrementing from zero up to the target value which has been written in the corresponding RegCntX register(s). When the counter content is equal to the target value, an interrupt is generated at the next falling edge of counter clock. Then the counter is loaded again with the zero value at the next rising edge of counter clock (Figure 20-2). When in downcount mode, the counter will start decrementing from the initial load value which has been written in the corresponding RegCntX register(s) down to the zero value. Once the counter content is equal to zero, an interrupt is generated at the next falling edge of counter clock. Then the counter is loaded again with the load value at the next rising edge of counter clock (Figure 20-2). Be careful to select the counter mode (no capture, not PWM, specify cascaded or not and up or down counting mode) before writing any target or load value to the RegCntX register(s). This ensures that the counter will start from the correct initial value. When counters are cascaded, both counter registers must be written to ensure that both cascaded counters will start from the correct initial values. The stopping and consecutive starting of a counter in counter mode without a target or load value write operation in between can generate an interrupt if this counter has been stopped at the zero value (downcount) or at it's target value (upcount). This interrupt is additional to the interrupt which has already been generated when the counter reached the zero or the target value. dow n counting clock counter X RegcntX_r XX RegCntX_w XX 3 2 1 0 3 2 1 0 3 2 1 0 3 write RegCntX CntXDownUp IrqX CntXEnable up counting clock counter X RegCntX_r RegCntX_w XX XX 0 1 2 3 0 1 2 3 0 1 2 3 3 write RegCntX CntXDownUp IrqX CntXEnable Figure 20-2. Up and down count interrupt generation. (c) Semtech 2006 www.semtech.com 20-7 XE8805/05A 20.10 PWM mode The counters can generate PWM signals (Pulse Width Modulation) on the Port B outputs PB(0) and PB(1). The PWM mode is selected by setting CntPWM1 and CntPWM0 in the RegCntConfig1 register. See Table 20-11 and Table 20-12 for an exact description of how the setting of CntPWM1 and CntPWM0 affects the operating mode of the counters A, B, C and D according to the other configuration settings. When CntPWM0 is enabled, the PWMA or PWMAB output value overrides the value set in bit 0 of RegPBOut in the Port B peripheral. When CntPWM1 is enabled, the PWMC or PWMCD output value overrides the value set in bit 1 of RegPBOut. The corresponding ports (0 and/or 1) of Port B must be set in digital mode and as output and either open drain or not and pull up or not through a proper setting of the control registers of the Port B. Counters in PWM mode always count down, the CntXDownUp bit setting must be reset. No interrupts and events are generated by the counters which are in PWM mode. Counters do count circularly: they restart at the maximal value (either 0xFF when not cascaded or 0xFFFF when cascaded) when respectively an underflow condition occurs in the counting. The internal PWM signals are low as long as the counter contents are higher than the PWM code values written in the RegCntX registers. They are high when the counter contents are smaller or equal to these PWM code values. The PWM resolution is always 8 bits when the counters used for the PWM signal generation are not cascaded. PWM0Size(1:0) and PWM1Size(1:0) in the RegCntConfig2 register are used to set the PWM resolution for the counters A and B or C and D respectively when they are in cascaded mode. The different possible resolutions in cascaded mode are shown in Table 20-13. Choosing a 16 bit PWM code higher than the maximum value that can be represented by the number of bits chosen for the resolution, results in a PWM output which is always tied to 1. PwmXsize(1:0) 11 10 01 00 Resolution 16 bits 14 bits 12 bits 10 bits Table 20-13: Resolution selection in cascaded PWM mode Small PWM code Tlsmall Thsmall Large PWM code Tllarge Thlarge Tper Figure 20-3: PWM modulation examples The period of the PWM signal is given by the formula: Tper = 2 resolution f ckcnt (c) Semtech 2006 www.semtech.com 20-8 XE8805/05A The duty cycle ratio DCR of the PWM signal is defined as: DCR = DCR can be selected between 0 % and Th Tper 2 resolution - 1 * 100 %. 2 resolution DCR in % in function of the RegCntX content(s) is given by the relation: DCR = 100 * RegCntX 2 resolution 20.11 Capture function The 16-bit capture register is provided to facilitate frequency measurements. It provides a safe reading mechanism for the counters A and B when they are running. When the capture function is active, the processor does not read anymore the counters A and B directly, but instead reads shadow registers located in the capture block. An interrupt is generated after a capture condition has been met when the shadow register content is updated. The capture condition is user defined by selecting either internal capture signal sources derived from the prescaler or from the external PA(2) or PA(3) ports. Both counters use the same capture condition. When the capture function is active, the A and B counters must be written with the value 0xFF and can either upcount or downcount. They do count circularly: they restart at zero or at the maximal value (either 0xFF when not cascaded or 0xFFFF when cascaded) when respectively an overflow or an underflow condition occurs in the counting. CapFunc(1:0) in register RegCntConfig2 determines if the capture function is enabled or not and selects which edges of the capture signal source are valid for the capture operation. The source of the capture signal can be selected by setting CapSel(1:0) in the RegCntConfig2 register. For all sources, rising, falling or both edge sensitivity can be selected. Table 20-14 shows the capture condition as a function of the setting of these configuration bits. CapSel(1:0) Selected capture signal 11 1K 10 32 K 01 PA3 00 PA2 CapFunc 00 01 10 11 00 01 10 11 00 01 10 11 00 01 10 11 Selected condition Capture disabled Rising edge Falling edge Both edges Capture disabled Rising edge Falling edge Both edges Capture disabled Rising edge Falling edge Both edges Capture disabled Rising edge Falling edge Both edges Capture condition 1 K rising edge 1 K falling edge 1 K both edges 32 K rising edge 32 K falling edge 32 K both edges PA3 rising edge PA3 falling edge PA3 both edges PA2 rising edge PA2 falling edge PA2 both edges Table 20-14: Capture condition selection CapFunc(1:0) and CapSel(1:0) can be modified only when the counters are stopped otherwise data may be corrupted during one counter clock cycle. Due to the synchronization mechanism of the shadow registers and depending on the frequency ratio between the capture and counter clocks, the interrupts may be generated one or only two counter clock pulses after the effective capture condition occurred. When the counters A and B are not cascaded and do not operate on the (c) Semtech 2006 www.semtech.com 20-9 XE8805/05A same clock, the interruptions on IrqA and IrqB which inform that the capture condition was met, may appear at different moments. In this case, the processor should read the shadow register associated to a counter only if the interruption related to this counter has been detected. It must be noted that when counters A and B are cascaded, the capture might happen at different cycles for the A and B registers. This is due to the asynchronous relationship between counter and capture clock and to the fact that the capture condition detection is independent for A and B counters. (c) Semtech 2006 www.semtech.com 20-10 XE8805/05A 21 VLD (Voltage Level Detector) 21.1 Features 21-2 21.2 Overview 21-2 21.3 Register map 21-2 21.4 Interrupt map 21-2 21.5 VLD operation 21-2 (c) Semtech 2006 www.semtech.com 21-1 XE8805/05A 21.1 Features * * Can be switched off, on or simultaneously with CPU activities Generates an interrupt if power supply is below a pre-determined level 21.2 Overview The Voltage Level Detector monitors the state of the system battery. It returns a logical high value (an interrupt) in the status register if the supplied voltage drops below the user defined level (Vsb). 21.3 Register map There are two registers in the VLD, namely RegVldCtrl and RegVldStat. Table 221-1 shows the mapping of control bits and functionality of RegVldCtrl while Table 221-2 describes that for RegVldStat. pos. 7-4 3 RegVldCtrl -VldRange rw r rw reset 0000 0 resetsystem 2-0 VldTune[2:0] rw 000 resetsystem function reserved VLD detection voltage range for VldTune = "011": 0 : 1.3V 1 : 2.55V VLD tuning: 000 : +19 % 111 : -18 % Table 221-1: RegVldCtrl pos. 7-3 2 RegVldStat -VldResult rw r r reset 00000 0 resetsystem 1 0 VldValid VldEn r rw 0 resetsystem 0 resetsystem function reserved is 1 when battery voltage is below the detection voltage Indicates when VldResult can be read VLD enable Table 221-2: RegVldStat 21.4 Interrupt map interrupt source IrqVld mapping in the interrupt manager RegIrqMid(2) Table 221-3: Interrupt map 21.5 VLD operation The VLD is controlled by VldRange, VldTune and VldEn. VldRange selects the voltage range to be detected, while VldTune is used to fine-tune this voltage level in 8 steps. VldEn is used to enable (disable) the VLD with a 1(0) value respectively. Disabled, the block will dissipate no power. (c) Semtech 2006 www.semtech.com 21-2 XE8805/05A symbol description min typ max unit Note 1 Vth TEOM TPW Threshold voltage VldTune 1.53 VldRange 0 1.44 0 001 1.36 0 010 1.29 0 011 1.22 0 100 1.16 0 101 1.11 0 110 0 111 3.06 1 000 2.88 1 001 2.72 1 010 2.57 1 011 2.44 1 100 2.33 1 101 2.22 1 110 2.13 1 111 V 1.06 duration of measurement Minimum pulse width detected comments trimming values: 2.0 2.5 ms Note 2 875 1350 us Note 2 000 Table 221-4: Voltage level detector operation Note 1: absolute precision of the threshold voltage is 10%. Note 2: this timing is respected in case the internal RC or crystal oscillators are enabled To start the voltage level detection, the user sets bit VldEn. The measurement is started. After 2ms, the bit VldValid is set to indicate that the measurement results are valid. From that time on, as long as the VLD is enabled, a maskable interrupt request is sent if the voltage level falls below the threshold. One can also poll the VLD and monitor the actual measurement result by reading the VldResult bit of the RegVldStat. This result is only valid as long as the VldValid bit is `1'. An interrupt is generated on each rising edge of VldResult. (c) Semtech 2006 www.semtech.com 21-3 XE8805/05A 22 Physical dimensions CONTENTS 22.1 QFP type package 22-2 (c) Semtech 2006 www.semtech.com 22-1 XE8805/05A 22.1 QFP type package The QFP package dimensions are given in Figure 22-1 and Table 22-1. The dimensions conform to JEDEC MS026 Rev. C. Figure 22-1. QFP type package package LQFP-64 A B C D E F G mm 10.0 mm 12.0 mm 1.4 mm 0.10 mm 0.22 mm 0.5 mm Table 22-1. QFP package dimensions (c) Semtech 2006 www.semtech.com 22-2 XE8805/05A (c) Semtech 2005 All rights reserved. Reproduction in whole or in part is prohibited without the prior written consent of the copyright owner. The information presented in this document does not form part of any quotation or contract, is believed to be accurate and reliable and may be changed without notice. No liability will be accepted by the publisher for any consequence of its use. Publication thereof does not convey nor imply any license under patent or other industrial or intellectual property rights. Semtech. assumes no responsibility or liability whatsoever for any failure or unexpected operation resulting from misuse, neglect improper installation, repair or improper handling or unusual physical or electrical stress including, but not limited to, exposure to parameters beyond the specified maximum ratings or operation outside the specified range. SEMTECH PRODUCTS ARE NOT DESIGNED, INTENDED, AUTHORIZED OR WARRANTED TO BE SUITABLE FOR USE IN LIFE-SUPPORT APPLICATIONS, DEVICES OR SYSTEMS OR OTHER CRITICAL APPLICATIONS. INCLUSION OF SEMTECH PRODUCTS IN SUCH APPLICATIONS IS UNDERSTOOD TO BE UNDERTAKEN SOLELY AT THE CUSTOMER'S OWN RISK. Should a customer purchase or use Semtech products for any such unauthorized application, the customer shall indemnify and hold Semtech and its officers, employees, subsidiaries, affiliates, and distributors harmless against all claims, costs damages and attorney fees which could arise. Contact Information Semtech Corporation Wireless and Sensing Products Division 200 Flynn Road, Camarillo, CA 93012 Phone (805) 498-2111 Fax : (805) 498-3804 (c) Semtech 2006 www.semtech.com 22-3