19-0597; Rev 3; 7/98 MAA AIL/VI Pin Programmable Universal and Bandpass Filters General Description The MAX263/264 and MAX267/268 CMOS switched- capacitor active filters are designed for precision filtering applications. Center frequency, Q, and oper- ating mode are all selected via pin-strapped inputs. The MAX263/264 uses no external components for a variety of bandpass, lowpass, highpass, notch and allpass filters. The MAX267/268 is dedicated to bandpass applications and includes an uncommitted op-amp. Two second-order filter sections are included in both devices. An input clock and a 5-bit programming input precisely set the filter center/corner frequency. Q is also programmed from 0.5 to 64. Separate clock inputs for each filter half operate with either an external clock or a crystal. The MAX263 and 267 operate with center frequencies up to 57kHz while the MAX264 and 268 extend the fo range to 140kHz by employing lower fo_k/fo ratios. The MAX263/264 is supplied in 28 pin wide DIP and small outline packages while the MAX267/268 is supplied in 24 pin narrow DIP and wide SO packages. All devices are available in commercial, extended, and military temperature ranges. Applications Sonar and Avionics Instruments Anti-Aliasing Filters Digital Signal Processing Vibration and Audio Analysis Matched Tracking Filters Typical Application SU AAIAI MAX263 INPUT ouTPUT FILTER B (120Hz NOTCH) 20kKHz FILTER A +BV (BOHz NOTCH) -5V fp LOGIC 60Hz-120Hz NOTCH FILTER eeeeee Filter Design Software Available 32-Step Center Frequency Control 128-Step Q Control Independent Q and f) Programming Guaranteed Clock to fp Ratio1% (A grade) 75kHz f, Range (MAX264/268) Single +5V and +5V Operation Features Ordering Information PART TEMP. RANGE PACKAGE* ACCURACY MAX263ACPI 0C to +70C ~ Plastic DIP 1% MAX263BCPI 0C to +70C ~- Plastic DIP 2% MAX263AEPI ~40C to +85C - Plastic DIP 1% MAX263BEPI -40C to +86C Plastic DIP 2% MAX263ACWI 0C to +70C = Wide SO 1% MAX263BCWI 0C to +70C = Wide SO 2% MAX263AMJI _-55C to +126C }=CERDIP 1% MAX263BMJI = -55C to +126C ~=CERDIP 2% MAX264ACPI 0C to +70C Plastic DIP 1% MAX264BCPI 0C to +70C ~Pilastic DIP 2% (Ordering Information continued at end of data sheet.) * MAX263/264 packages are 28-pin 0.6" wide DIP and 28-pin 0.3 wide SO (Small Outline). MAX267/268 packages are 24-pin 0.3" narrow DIP and 24-pin 0.3" wide SO (Small Outline). co . Pin Configuration TOP VIEWS INB rf [28] LPs SI LP, [2 | [37] oP, Na 1d 124] oP a Be, (3 36] NHP_ BPA ie 33 )a4 NMP TT] , 38] a4 op out [3] [22] Fo {MAXIM op [a] Hai re NaS] = MAX269 [24I Fo MAKLMI as fe] MAX264 Tastee wal] = Maxze7 [20] 03 MAX268 mi OF [zz] a3 as[61 /ganppass [2)% mo [8 | 21] a2 afr] ONLY) [is] osc our as [3] Bosc our [8 7] ono v: Lo! [9] GNO Fa[ a ji] v- ra li ra] v- 3 To a5 Fs Fa [92 a7) 1 CLR Fi | [1a jo1 cura [73] i] ar Coke [12] [33 00 CLKe [74] 18] a6 MAAXIMA For small orders, phone 1-800-835-8769. Maxim Integrated Products 1 For free samples & the latest literature: http:/;www.maxim-ic.com, or phone 1-800-998-8800. 89ST XVW/LOSTXVW/P9OCXVW/ESSXVINMAX263/MAX264/MAX26 7/MAX268 Pin Programmable Universal and Bandpass Filters ABSOLUTE MAXIMUM RATINGS Total Supply Voltage (V" to V) Input Voltage, any pin Input Current, any pin Power Dissipation eee eee eneenee eens 15V vane een ee ee renee +50mA Plastic DIP (derate 8.38mW/C above 70C) ... 660mW CERDIP (derate 12.5mW/C above 70C) ..... 1000mW Wide SO (derate 11.8mW/C above 70C) tenes 944mW Operating Temperature MAX26XXCXX ow. eee eee ee OC to +70C MAX26XXEXX 200. cee ~40C to +85C MAX26XXMXX oo. eee eee ~5C to +125C Storage Temperature .............-..066 ~65C to +160C Lead Temperature (Soldering, 10 seconds) ....... +300C Stresses above those listed under Absolute Maximum Ratings may cause permanent damage to the device. These are stress ratings only, and functional operation of the device at these or any other conditions above those indicated in the operational sections of the specification is not implied. Exposure to absolute Maximum ratings conditions for extended periods may affect the device reliability. ELECTRICAL CHARACTERISTICS (V" = +5V, Vo = -5V, CLK, = CLK, = + 5V, 1.5MHZz, fo. x/fg = 197.92 for MAX263/67 and 138.23 for MAX264/68, Filter Mode 1, 1 = Vv and 0 = V" on F and Q inputs, T, = +25C unless otherwise noted.) PARAMETER CONDITIONS MIN Typ MAX UNITS fg Center Frequency Range See Table 1 Maximum Clock Frequency See Table 1 foi x/f, Ratio Error (Note 1) Ta = Tay to T MAX26XA +0.2 +1.0 on owe MAS MAX26XB +02 +20 | * fo Temperature Coefficient ~5 ppm/c Q Accuracy (deviation from Ta =TMmin to TMAX ideal continuous filter) Q=0.5t08 MAX26XA 1 +6 (Note 2) Q=0.5t0o 8 MAX26XB +1 +10 Q= 16 to 32 MAX26XA +2 #10 % Q= 16 to 32 MAX26XB tke +15 Q=64 MAX26XA +4 +20 Q = 64 MAX26XB +4 +25 Q Temperature Coefficient +20 ppm/*Cc DC Lowpass Gain Accuracy MAX263/4 +0.1 +0.5 dB Gain Temperature Coefficient | Lowpass (at D.C.) ~5 mec Bandpass (at f,) +20 pp Output Offset Voltage Ta = Twin tO Tyax Q = 4 (Note 3) Mode 1 MAX263/67A +0.05 0.20 BP Output MAX263/67B +0.05 +0.30 MAX264/68A +0.05 +0.20 MAX264/68B +0.05 +0.30 Mode 1 MAX263A +0.40 1.00 LPN Outputs MAX263B 0.80 +1.60 MAX264A +0.40 +1.20 MAX264B +0.80 +1.60 Vv Mode 3 MAX263A 0.10 +0.20 BP HP Outputs MAX263B +0.10 +0.30 MAX264A +0.10 +0.20 MAX264B +0.10 +0.30 Mode 3 MAX263A 0.50 +1.10 LP Output MAX263B +0.90 +1.60 MAX264A +0.50 +1.30 MAX264B +0.90 +1.60 Offset Voitage Temperature fo r/fp = 100.53, Q = 4 o Coefficient Ta = Tain tO Trax 0.75 mv? Glock Feedthrough +4 mV Crosstalk -70 dB Wideband Noise (Note 4) Q = 1, 2nd-Order, LP/BP See Typ. Oper. Char. 4th-Order LP 90 Vans 4th-Order BP 100 MAAXKIMELECTRICAL CHARACTERISTICS (Continued) (V" = +5V, V" = -5V, CLK, = CLK, = + 5V, 1.5MHz, foi y/fp = 197.92 for MAX263/67 and 138.23 for MAX264/68, Filter Mode 1, 1" = V* and 0 = V on F and Q inputs, 7, = +25C unless otherwise noted.) Pin Programmable Universal and Bandpass Filters PARAMETER CONDITIONS MIN TYP MAX | UNITS Harmonic Distortion at f, Q = 4, Vin = 1.5Vpp 67 dB Supply Voltage Range Ta = Tyin tO Trax 2.37 +5 6.3 V Power Supply Current Ta = Twin tO Tax MAX263/67 14 20 mA (Note 5) MAX264/68 14 20 Shutdown Supply Current Q0-Q6 = all 0 2.5 mA (Note 5) fp, Q Programming Inputs Ta = Twin tO Trax, FO-F4, Q0-Q6 High Threshold V*-0.5 Vv Low Threshold V+O.5 Clock Inputs Ta = Twin tO Tax, CLK,, CLK High Threshold 2.4 V Low Threshold 0.8 Input Leakage Current Ta = Tun.tO Trax CLK, = V" or V- 10 CLK, = V or V 6 60 HA MO, M1, FO-F4, Q0-Q6 = V*-0.5V or V"+0.5V 20 200 MO, M1, FO-F4, Q0-Q6 = V or V 5 INTERNAL AMPLIFIERS Output Signal Swing Ta = Tyan tO Tray, 10kQ load +4.75 V Output Short Circuit Current source . mA Power Supply Rejection Ratio | OHz to 10kHz -70 dB Gain Bandwidth Product 2.5 MHz Slew Rate 6 Vius ELECTRICAL CHARACTERISTICS (for V+ = +2.5V +5%) (V" = +2.37V, V = -2.37V, CLK, = CLKg = +2.5V 1MHz fo.,/fg = 197.92 for the MAX263/67 and 138.23 for MAX264/68, Filter Mode 1, Ta = +25C unless otherwise noted.) PARAMETER CONDITIONS MIN TYP MAX | UNITS fy Center Frequency Range (Note 6) Maximum Clock Frequency (Note 6) forw?/fp Ratio Error Q=8 MAX26XA +01 +1 % (Notes 1, 7) MAX26XB +01 2 @ Accuracy (deviation from Q=8 ideal continuous filter) fork/fp = 197.92 MAX263/67A 2 46 (Notes 2, 7) MAX263/67B +2 +10 o% four/fp = 138.23 MAX264/68A +2 +6 MAX264/68B +2 10 Output Signal Swing All Outputs +2 Vv Power Supply Current 7 mA Shutdown Current 0.45 mA Note 1: Note 2: Note 3: feik/fo accuracy is tested at 197.92 on the MAX263/67, and at 138.23 on the MAX264/68. Q accuracy tested at Q = 8, 32, and 64. Q of 32 and 64 tested at 1/2 stated clock frequency. The Offset Voltage is specified for the entire filter. Offset is virtually independent of Q and fc.k/fo ratio setting. The test clock frequency for Mode 3 is 750kHz. Output noise is measured with an RC output smoothing filter at 4 x fo to remove clock feedthrough. TTL logic levels are: HIGH = 2.4V, LOW = 0.8V. Power supply current is typically 4mA higher with TTL clock input levels. At #2.5V supplies, the fo range and maximum clock frequency are typically 75% of values listed in Table 1. fcik/fo and Q accuracy are a function of the accuracy of internal capacitor ratios. No increase in error is expected at +2.5V as compared to +5V, however these parameters are only tested to the extent indicated by the MIN or MAX limits. Note 4: Note 5: Note 6: Note 7: MAXIM ce een neue 3 89CXVW/L9OCXVIN/P9ICXVW/ESSTXVINMAX263/MAX264/MAX26 7/MAX268 Pin Programmable Universal and Bandpass Filters Typical Operating Characteristics IDD vs POWER SUPPLY VOLTAGE 25 25C 20 CLK FREQ = 500KHz CONTROL PINS (5V, 0V) Lf / % | | CLOCKS (5V, 0 Lf | w/ | IDD (mA) a yr a YJ cLOcks (5 A | V, -5V) 5 6 7 8 9 10 vt TO v- (V) teix/fp ERROR (%) n 12 IDD (ma) iDD vs CLOCK FREQUENCY torx/fp vs CLOCK FREQUENCY 0.2 0.0 ~0.2 -0.4 -0.6 ~0.8 -1.0 1.2 1.0 15 2.0 25 Q vs CLOCK FREQUENCY 25 4 30 35 CLOCK FREQUENCY (MHz) 20 a CLOCK (2.4V, 0.8V} 19 ] L cennjeamnesy 1 i ? T { 18 CLOCK @v, ov) -] +5V a = 17 | CONTROL PINS (Sv, OV) ~ 25C a 5 16 = Ww 15 - o Lo CLOCK (SV, -5V) 44 io 13 0.5 15 25 35 0540 #415 20 CLOCK FREQUENCY (MHz) Wy VS fo 1M + Ta = 25C 100K ALL MODES /| _ oon 7 G /) Ww @ 10K 7 a 2 S z 1K 7 g L. a = 100 2 a 3 S| 10 V4 1 30 3.5 CLOCK FREQUENCY (MHz) 1 10 100 1K 10K = 100K Wideband RMS Noise (db ref. to 2.47Vams, 7Vp-p), +5V Operation Mod Q=1 Q=8 Q = 64 le LP BP | HP/AP/N | LP BP | HPYAP/N | LP BP | HP/AP/N 1 ~84 ~90 ~B4 ~80 ~82 ~85 -72 ~73 ~85 2 -88 ~90 ~88 ~84 ~82 ~84 ~77 -73 -76 3 ~84 ~90 -88 -80 -82 -82 -73 -73 ~74 4 -83 -89 ~84 ~79 ~81 ~B5 -71 -73 ~85 Notes: 1. fork 7 IMHz 2. forty ratio programmed at N = 31 (see Table 2) 3. Clock feedthrough is removed with an RC lowpass at 4fp, i.e. R = 3.9kO, C = 2000pF for MAX263. fo (Hz) Noise Spectral Distribution (MAX263/67, fork = 1 MHz, dB ref. to 2.47Veus, 7Vo-p) Measurement - - = Bandwidth | O1|@=8 | Q=64 Wideband ~84 | -80 | -72 SkHz ~87 | -87 | ~-86 C Message Weighted 93 | -83 | -93 MAXIMPin Programmable Universal and Bandpass Filters Pin Description MAX263 MAX267 MAX264 MAX268 PIN # PIN # NAME FUNCTION 10 8 ve Positive supply voltage 18 16 Vv Negative supply voltage 19 17 GND Analog Ground. Connect to the system ground for dual supply operation or mid-supply for single supply operation. GND should be well bypassed in single supply applications. 13 ae) CLK, Input to the oscillator and clock input to section A. This clock is internally divided by 2. 14 12 CLKs Clock input to filter B. This clock is internally divided by 2. 20 18 OSC OUT Connects to crystal for self clocked operation 5,1 5,1 INa, INe Filter inputs 3, 27 2, 24 BPs, BPs Bandpass outputs 2, 28 LPs, LPs Lowpass outputs (MAX263/264 only) 4, 26 HPa, HPs Highpass/Notch/Alipass outputs (MAX263/264 only) 8,7 MO, M1 Mode select inputs (MAX267/268 are fixed in Mode 1) 24, 17, 23 22, 15, 21 FO-F4 Clock/center frequency ratio (fcix/fo) programming inputs 12, 11 10, 9 16, 16, 21 13, 14, 19 Q0-Q6 Q programming inputs 22, 25, 6 20, 23, 6 9 7 4 OP IN Inverting input of uncommitted op-amp on MAX267/268 only. Noninverting input is internally connected to ground. 3 OP OUT Output of uncommitted op-amp on MAX267/268 only. MAKLAM 5 89SXVW/ZLOSXVIN/P9CXVIN/ESEXVWMAX263/MAX264/MAX267/MAX268 Pin Programmable Universal and Bandpass Filters SCN = SWITCH-CAPACITOR NETWORK N/HP/AP (MAX263/64 ONLY) SCN pi BP IN SCN p LP (MAX263/64 ONLY) $1 $2 $3 | MODE q SCN SELECT $3 | 1 SCN oS Mo M1 Lt (MAX263/64 ONLY) (TABLE 3) (TABLE 2) THE MAX267/68 OPERATES ONLY AS A MODE 1 BANDPASS. INTERNAL SWITCHES S1, S2, S3 ARE SHOWN FOR MODE 1. Figure 1. Filter Block Diagram (One Second-Order Section) Introduction Each MAX26X device contains two second-order filters. In Figure 1, a block diagram of the state variable topology employed in one filter section shows how on-chip switched capacitor networks provide adjustable feedback to control fo and Q. Shared pro- gramming inputs require that both halves of the filter be set for the same foux/fo ratio and Q. In the MAX263 and MAX264 universal filters, switches $1-S3 are con- trolled by inputs MO and M1 to set the filter operating mode. The MAX267/68 bandpass filter operates only in Mode 1. The MAX264/68 uses a lower range of sampling (fex/fo) ratios than the MAX263/67 to allow higher signal bandwidths and a wider programming range. The reduced foix/fo ratios result in somewhat more deviation from ideal continuous filter parameters than with the MAX263/67, however these differences can be compensated using Figure 17 (See Applications Hints) or Maxims filter design software. The second-order sections in the MAX263/64/67/68 are identical and may be used as matched dual track- ing filters, or can be cascaded to form higher-order filters. They can also be combined with external resistors and amplifiers for multiple feedback all-pole bandpass filters. In all MAX26X series filters, the internal sample rate is one half the input clock rate (CLK, or CLKa) due to an internal division by two. All clock related data, tables, and other discussions in this data sheet refer to the frequency at the CLK, or CLK, input, i.e. twice the internal sample rate, unless specifically stated otherwise. 6 ____._.. Quick Look Design Procedure MAX26xX series filters, with Maxims filter design soft- ware, greatly simplify the design procedure for many active filters. Most designs can be realized using the steps in this section. If the filter software is not used, or if the complexity is beyond the scope of this section, refer to the remainder of this data sheet for more detailed application information. Step 1Filter Design Starting with the design program PZ, determine what type of filter is needed. PZ helps determine the type (Butterworth, Chebyshev, etc.) and the number of poles for the optimum choice. The program also plots the frequency response and calculates the pole/zero (fo) and Q values for each second-order section. Each MAX26X contains two such sections and devices may be cascaded for higher order filters. An alternate technique for bandpass filters uses multiple feedback (see Figure 13). If this is employed, the filter design program BP should be used instead of PZ and Step 2 is not used. Step 2Generate Programming Coefficients If multiple feedback is not used, start with the fo and Q values obtained with PZ in Step 1 and use the program MPP to generate the digital program codes for feix/fo and Q. MPP displays N values for fo and Q where N is the decimal equivalent of the binary pin- program codes. These are listed in Tables 2 and 3. MAXIMPin Programmable Universal and Bandpass Filters An input clock and filter Mode must also be selected in this step, however, if a specific clock rate is not selected, MPP will pick one. With regard to mode selection, Mode 1 (only possible mode for MAX267/68) is the most convenient choice for most bandpass and lowpass filters except for elliptics which require Mode 3. Highpass filters also use Mode 3, while allpass filters require Mode 4. For details regarding mode selection see Filter Operating Modes. When a clock frequency (or frequencies) is selected and the pro- gramming codes for fe.x/fo and Q are determined, the filter can then be programmed and operated. Filter Design Software Maxim provides software programs to help speed the transition from frequency response design require- ments to working hardware. A series of programs are available, including: Program PZ. Given the requirements, such as center frequency, Q, passband ripple, and stopband attenua- tion, PZ will calculate the pole frequencies, Qs, zeros, and the number of stages needed. Program MPP. For programmed filters, MPP computes the input codes to use and describes the expected performance of the design. Program BP. In the special case of bandpass filters, an alternate mode of operation is the Multiple Feedback Technique. BP calculates the resistor values and the bandpass frequency response for this mode. An advantage of multiple feedback is that identical MAXIMA programming and one clock frequency can be used for all stages. Program FR. When a design of one or more stages is completed, FR checks the final cascaded assembly. The output frequency response can be compared with that expected from PZ. Detailed Description f, and @ Programming Figure 2 shows a block diagram of a compiete filter. Each 2nd-order filter section has its own clock input, however, package pin limitations require that fo, Q, and Mode control be shared by both sections. The actual center frequency is a function of the filters clock rate, Pu fo control word (see Table 2), and operating ode. For some filter designs, the MAX263/64/67/68 may require separate clocks for each second-order section since separate programming inputs are not provided. Such designs may be implemented with different clock inputs, or, in the case of bandpass filters, by using multiple feedback and one clock (see Descrip- tion of Filter Functions). When implementing two or more matched filters, however, the programming restrictions are easily overcome and one clock can still be used as demonstrated by the design example in Figure 21. Another alternative is to use the MAX260/261/262 microprocessor programmed filters or the MAX265/266 resistor programmed filters which allow independent programming of each filter section. Refer to the device data sheets for further details on those products. 89SXVW/LOSXVIN/PISXVN/EITXVMAX263/MAX264/MAX267/MAX268 Pin Programmable Universal and Bandpass Filters Table 1. Typical Clock and Center Frequency Limits (MAX267/268 are operated in Made 1 only.) Mo fp LOGIC Mir CLKa Q LOGIC OSC OUT CLKg OP IN Me YF _- GND PART | @ | MODE fox fo PART | O | MODE fork fy MAX263/1 1 1 40Hz-4.0MHz 0.4H2-40kHz MAX264/| 1 1 40Hz-4.0MHz 1.0Hz-100kHz 267 1 2 40Hz-4.0MHz 0.5Hz-57kHz 268 1 2 40Hz-4.0MHz 1.4H2-140kHz 4 3 40Hz-4,0MHz 0.4Hz-40kHz 1 3 40Hz-4.0MHz 1.0H2z-100kHz 1 4 40Hz-4.0MHz 0.4Hz-40kHz 1 4 40Hz-4.0MHz 1.0Hz-100kHz 8 1 40Hz-2.7MHz 0.4Hz-27kHz 8 1 40Hz-2 5MHz 1.0Hz-80kHz 8 2 40Hz-2.1MHz 0.5H2-30kHz 8 2 40Hz-1.4MHz 1.4Hz-50kHz 8 3 40Hz-1.7MHz 0.4Hz-17kHz 8 3 40H2-1.4MHz 1.0Hz-35kHz & 4 40Hz-2.7MHz 0.4Hz-27kHz 8 4 40Hz-2.5MHz 1,.QHz-60kHz 64 1 40Hz-2.0MHz 0.4H2-20kHz 64 1 40H2-1.5MHz 1.0Hz-37kHz 90 2 40Hz-1.2MHz 0.4Hz-18kHz 90 2 40Hz-6.9MHz 1.4Hz-32kHz 64 3 40Hz-1.2MHz 0.4Hz-12kHz 64 3 40Hz-0.9MHz 1.OHz-22kHz 64 4 40Hz-2.0MHz 0.4Hz-20kHz 64 4 40H2-1.5MHz 1.0Hz-37kHz INA N/HPYAP,* BP, LP, INg N/HP/APp* BPg LBa* *MAX263/64 ONLY MAXK267/68 ONLY OP OUT Figure 2. MAX263/264/267/268 Block Diagram MAKLMPin Programmable Universal and Bandpass Filters Table 2. fo_x/fp Program Selection Table 8S9CXVIW/LOCXVIN/P9SCXVW/ESSXVIN forx/lg RATIO PROGRAM CODE MAX263/67 MAX264/68 MODE 1,3,4 MODE 2 MODE 1,3,4 MODE 2 N F4 F3 F2 Fi Fo 100.53 71.09 40.84 28.88 0 0 0 0 0 0 103.67 73.31 43.98 31.10 1 0 0 0 0 j 106.81 75.53 4712 33.32 2 0 0 0 1 0 109.96 77.75 $0.27 35.54 3 0 0 0 1 1 113.10 79.97 3.41 37.76 4 0 0 1 0 0 116.24 82.19 56.55 39.99 5 0 0 1 0 4 119.38 84.42 59.69 42.21 6 0 0 1 1 0 122.52 86.64 62.83 44,43 7 0 0 1 { 1 125.66 88.86 65.97 46.65 8 0 1 0 0 0 128.81 91.80 69.12 48.87 9 0 1 0 0 1 131.95 93.30 72.26 51.10 10 0 1 0 1 0 135.08 95.52 75.40 53.31 W 0 1 0 4 1 138.23 97.74 78.53 55.54 12 0 1 1 0 0 141.37 99.97 81.68 $7.76 13 0 1 1 0 1 144.51 102.89 84.82 59.98 14 0 1 1 1 0 147.65 104.41 87.96 62.20 15 0 1 1 1 1 150.80 106.63 9111 64.42 16 1 0 0 0 0 153.98 108.85 94.25 66.64 17 1 0 0 0 1 157.08 111.07 97.39 68.86 18 1 0 0 1 0 160.22 113.29 100.53 71.09 19 1 0 0 1 4 163.36 115.52 102.67 73.31 20 1 0 1 0 0 166.50 117.74 106.81 75.53 21 1 0 1 0 4 169.65 119.96 109.96 7775 22 1 0 1 1 0 172.79 122.18 113.10 79.97 23 1 0 1 1 1 175.93 124,40 116.24 82.19 24 1 1 0 0 0 179.07 126.62 119.38 84.81 25 1 1 0 0 1 182.21 128.84 122.52 86.64 26 1 1 0 1 0 185.35 131.07 125.66 88.86 27 1 1 0 1 1 188.49 133.29 128.81 91.08 28 1 1 1 0 0 191.64 135.51 131.95 93.30 29 1 1 4 0 1 194.78 137.73 135.09 95.52 30 1 1 1 1 0 197.92 139.95 138.23 97.74 31 1 1 1 1 1 Notes: 1) For the MAX263/67, fox/fo = m(N+32) in Mode 1, 3, and 4, where N varies form 0 to 31. 2) For the MAX264/68, foix/fo = 77(N+13) in Mode 1, 3, and 4, where N varies 0 to 31. 3) In Mode 2, all foux/fo ratios are divided by /2. MAKIM 9MAX263/MAX264/MAX267/MAX268 Pin Programmable Universal and Bandpass Filters Table 3. Q Program Selection Table (Continued on following page) PROGRAMMED Q PROGRAM CODE PROGRAMMED Q PROGRAM CODE MODE 1,3,4| MODE 2 | N Q6 Q5 Q4 Q3 Q2 Q1 QO MODE 13,4) MODE 2 | N Q6 Q5 Q4 Q3 G2 Qi QO Note 4 Note 4 0 0006 0 0 0 0 0.800 1.13 48 0 1 1 0 0 0 686 0.504 0.713 100 0 0 0 0 1 0.810 1.15 49 0 1 141 0 0 0 1 0.508 0.718 2 0 0 0 0 0 1 8 0.821 1.16 50 0 1 1 0 0 1 #0 0.512 0.724 300 0 060 0 1 41 0.831 1.18 51 0 1 17 0 0 1 3 0.516 0.730 4 00060 0 1 0 0 0.842 1.19 520 1 1 0 41 0 90 0.520 0.736 5 00 00 1 0 1 0.853 1.24 53 0 1 1 0 1 0 7 0.525 0.742 6 0 0 0 0 1 1 +90 0.865 1.22 6 0 1 1 0 141 14 ~=0 0.529 0.748 7 0 0 0 0 141 1 1 0.877 1.24 55 0 1 171 0 4 7 21 0.533 0,754 8 0 6 60 1 0 0 0 0.889 1.26 56 0 1 4 1 0 0 0 0.538 0.761 9 0 00 10 0 1 0.901 1.27 57 0 1 1 141 0 0 41 0.542 0.767 10 0 0 0 1 0 1 90 0.914 1.29 58 0 1 141 1 0 1 90 0.547 0.774 170001 0 1 =1 0.928 1.31 59 0 1 47 1 0 7141 0.552 0.780 12 0 0 0 1 141 0 90 0.941 1.33 60 0 1 1 1 4 @ =O 0.556 0.787 1300 0% 1 0 1 0.955 1.35 61 0 1 1 17 7 0 7 0.561 0.794 1400 0 14 1 t 0 0.969 1.37 62 0 1 1 7 14 1 8 0.566 0.801 15 0 0 0 1 17 1 ~=41 0.985 1.39 68 0 1 14 141 4 7 ~3 0.571 0.808 140 0 1 0 0 0 0 1.00 1.41 64 1 0 0 0 0 0 0 0.577 0.815 17 0 0 1 0 0 0 1 1.02 1.44 6 1 0 0 0 0 0 1 0.582 0.823 1% #0 0 1 00 1 =0 1.03 1.46 66 1 0 0 0 0 1 6 0.587 0.830 19 0 0 10 0 1 1 1.05 1.48 67 1 0 0 0 01 4 0.593 0.838 20 0 01 0 1 0 #0 1.07 1.51 68 1 0 0 0 1 0 #6 0.598 0.846 210 0 1 01 0 1 1.08 1.53 69 1 0 0 0 1 0 1 0.604 0.854 22 0 0 1 0 1 +1 =0 1.10 1.56 7o 41 0 0 0 1 1 9 0.609 0.862 23.0 0 41 0 1 14 4 1.12 1.59 mM 1 0 0 0 41 #14 ~4 0.615 0.870 24001 1 0 0 0 1.14 1.62 72 1 0 0 1 0 0 90 0.621 0.879 2 0 0 1 1 0 0 1 1.16 1.65 723 #1 0 0 190 0 1 0.627 0.887 260 0 1 1 0 1 +0 4.19 1.68 74 1 0 0 1 0 4 +80 0.634 0.896 27 0 0 1 1 0 1 4 1.21 171 7 1 0 0 4 0 744 0.640 0.905 28 0 0 1 141 7 0 0 1.23 1.74 7674 00 1: 4 ~0 0 0.646 0.914 29 0 0 1 71 1 0 41 1.25 1.77 77 406<8dC 0.653 0.924 30 6 0 1 1 #1 1 0 1.28 1.81 78 1 0 0 1 1 4 = 8 0.660 0.933 310 0 17 4 4 71 4 1.31 1.85 79 -160 0 4 ~4~7~1 0.667 0.943 320 1 90 0 0 0 0 1.33 1.89 80 1 0 1 0 0 0 8 0.674 0.953 33.0 1 0 0 0 0 4 1.36 1.93 81 1 0 1 0 0 0 +1 0.681 0.963 340 1 0 0 0 1 0 1.39 1.97 82 1 0 1 0 0 1 90 0.688 0.973 35 0 1 0 0 0 1 1 1.42 2.01 8 1 0 1 0 0 1 1 0.696 0.984 36 0 10 0 14 0 0 1.45 2.06 84 1 0 1 0 1 #0 8 0.703 0.995 370 10 0 1 0 =41 1.49 2.10 8% 1 0 1 0 1 0 1 0711 1.01 38 Oo 1 0 0 141 #7 0 1.52 2.16 86 1 0 1 0 1 741 90 0.719 1.02 39 0 1 0 0 4 4 41 1.56 2.21 87 1 0 1 0 1 71 =1 0.727 1.03 40 0 1 0 1 0 0 0 1.60 2.26 s8 1 0 1 1 40 0 OG 0.736 1.04 414.0 1 0 1 0 0 1 1.64 2.32 89 1 0 141 1 0 0 1 0.744 1,05 42 0 1 0 1 0 1 ~=0 1.68 2.40 90 1 0 1 1 0 4 9 0.753 1.06 44 0 41 0 1 0 141 1.73 2.45 91 $1 01 1 0 4 3 0.762 1.08 401 0 1 1 0 0 1.78 2.51 g2 1 01 1 4 90 9 0.77% 1.09 45 0 1 0 1 1 0 1 1.83 2.59 93 1 0 1 41 1 0 1 0.780 1.10 46 0 1 0 1 14 1 ~=90 1,88 2.66 9410 1 41 41 1 ~90 0.790 1.12 47 0 41 0 4 4 1 ~1 1.94 2.74 9 1 0 1 41 4 71 71 Notes: 4) Writing all Cs into Q0-6 activates a low power shutdown mode. BOTH filter sections are deactivated. 0 = Vt Ve. 10 MAXIMPin Programmable Universal and Bandpass Filters Table 3. Q Program Selection Table (Continued) PROGRAMMED Q PROGRAM CODE PROGRAMMED Q PROGRAM CODE MODE 1,3,4| MODE 2 | N Q6 Q5 Q4 Q3 Q2 Q1 QO MODE 1,3,4| MODE 2 | N Q6 Q5 Q4 Q3 Q2 Q1 Q0 2.00 2.83 96 11 0 0 0 0 0 4.00 5.66 12 11 1 00 0 0 2.06 2.92 97 1 1 0 0 0 0 =4 4.27 6.03 71317 =4 + O 0 0 1 2.13 3.02 98 110 0 0 1 0 4.57 6.46 1141 4 1 0 0 1 0 221 3.12 99 1 1 0 0 01 1 4.92 6.96 115 1 #14 7 0 0 1 41 2.29 3.23 100 71 1 0 0 747 +0 0 .33 7,54 1176 1 174 1 0 1 0 90 2.37 3.35 101 1 41 0 0 1 0 4 5.82 8.23 4171 #4 4 0 1 0 ~4 2.46 3.48 102 1 1 0 0 14 1 0 6.40 9.05 118 1 #4 71 0 41 71 96 2.56 3.62 103 1 14 G6 0 717 1 ~=41 7.11 10.1 W9 41 #4 7 0 14 74 1 2.67 3.77 1041 1 0 1 0 0 0 8.00 11.3 120 1 1 141 71 00 90 2.78 3.96 105 1 1 0 1 0 0 1 9.14 12.9 1211 4 1 1 0 0 1 2.91 4.41 106 1 1 0 717 0 414 +0 10.7 15.1 122 1 141 1 71 0 14 =0 3.05 431 107 71 #1 0 1 0 717 4 12.8 18.1 123 1 #4 #1 #74 0 7 ~7 3.20 4.53 108 1 #10 141 1 0 0 16.0 22.6 124 1 41 1 41 1 0 06 3.37 4.76 109 1 #410 14 41 0 1 21.3 30.2 125 1 14 1 1 1 0 41 3.56 5.03 110 1 #1 0 41 14 ~=1 ~=~90 32.0 45.3 126 1 #4 1 1 64~6648t~COD 3.76 5.32 1111 #4 O@ 14 4 1 ~=4 64.0 90.5 127 1 #9 1 74 4 ~74~49 Notes: 5) In Modes 1, 3, and 4: Q = 64/(128-N) 6) In Mode 2, the listed Q values are those of Mode 1 multiplied by V2. Then Q = 90.51/(128-N) Shutdown Mode FILTER A FILTER B | | osc CLKa ouT lon {| ------4 CRYSTAL FILTER A FILTER B 4 , osc CLKa OUT CLKg T Ne. | ee | JUL EXTERNAL CLOCK IN (ANY DUTY-CYCLE) Figure 3. Clock Input Connections Oscillator and Clock Inputs The clock circuitry of the MAX263/64/67/68 can operate with a crystal or an external clock generator as shown in Figure 3. The duty cycle of the clock at CLK, and CLKg is unimportant because the input is internally divided by two to generate the sampling clock for each filter section. It is important to note that this internal division also halves the sample rate when considering aliasing and other sampied system phenomenon. The filter enters a shutdown mode when all Q inputs, Q0-Q6, are tied low. When shut down, power con- sumption with +5V supplies typically drops to 25mW. When reactivating the filter after shutdown, allow 2ms to return to full operation. Filter Operating Modes (MAX263/264 Only) The MAX263/264s filter sections can be configured in four basic Modes as selected by inputs MO and Mi (see Table 4). The MAX267/68 operates only in Mode 1. A fifth mode, 3A, uses an external op amp and resistors but is selected the same way and uses the same internal configuration as Mode 3. Figures 4 through 8 show symbolic representations of the MAX263/64 filter modes. Only one second-order section is shown in each case, however the fp, Q, and Mode select inputs are common to both haives of the IC. The fp, fy (Notch), Q, and various output gains for each mode are shown in Table 4. Filter Mode Selection All operating modes listed in this section can be used with the MAX263/64. The MAX267/68 bandpass filter operates only in Mode 1. MODE 1 (Figure 4) is useful when implementing all- pole lowpass and bandpass fillers such as Butterworth, Chebyshev, Bessel, etc. It can also be used for notch filters, but only second-order notches because the rela- tive pole and zero locations are fixed. Higher order notch filters require more latitude in fo and fy, which is why they are more easily implemented with Mode 3A. 89ZXVW/LOCXVIN/POSXVW/ESCXVNMAX263/MAX264/MAX26 7/MAX 268 Pin Programmable Universal and Bandpass Filters Table 4. Filter Modes for Second-Order Functions MAX263/264 (MAX267/268 = MODE 1, BP only) FILTER Hons Hone MODE | M1,Mo | FUNCTIONS | f, | @ | fy | How | Hosp | (f-9) | (t-> foun) OTHER 1 0, 0 LP BPN fo -1 -Q 1 = 2 0, 1 LP BPN a ay fov2 | 05 |-Q//2] -05 4 3 1,0 | LP BPR HP a B 1 -a Honp =~! 3A 10 |uRBRHPN| Rr) fy /Ruy -q | +8 + Bs Hone = -1 w | uw R, R, Ri, a | o Hoap = 1 = ~ OaP ~~ 4 14 LR BP, AP 2 2Q fz = fy, Q2 = Q Notes: f, = Center Frequency fy = Notch Frequency Hop = Lowpass Gain at DC Hogp = Bandpass Gain at f, Hone = Highpass Gain as f approaches f>, ,/4 Mode 7, along with Mode 4, supports the highest clock frequencies (see Table 1) because the input summing amplifier is outside the filters resonant loop (Figure 4). The gain of the lowpass and notch outputs is 1, while the bandpass gain at the center frequency is Q. For bandpass gains other than Q, the filter input or output can be scaled by a resistive divider or op amp. In multiple feedback filters, the gain is set by the feedback resistors. SCN = SWITCHED-CAPACITOR NETWORK Figure 4. Filter Mode 1: Second-Order Bandpass, Lowpass and Notch MODE 2 (Figure 5) is used for all-pole lowpass and bandpass filters. Key advantages compared to Mode 1 are higher available Qs (see Table 3) and lower output noise. Mode 2's available fo, ,/fy ratios are V2 less than with Mode 1 (see Table 2) so a wider overall range of fps can be selected from a single clock when both modes are used together. MODE 3 (Figure 6) is the only mode which produces high-pass filters. The maximum clock frequency is somewhat less than with Mode 1 (see Table 1). MODE 3A (Figure 7) uses a separate op amp to sum the highpass and lowpass outputs of Mode 3, 42 Hon: = Notch Gain as f approaches DC Hone = Notch Gain as f approaches fo, ,/4 Hoap = Allpass Gain f,, Qz = f and Q of Complex Pole Pair creating a separate notch output. This output allows the notch to be set independently of fp by adjusting the op amps feedback resistor ratio (Ri, R,). Ru, R,, and Re are external resistors. Because the notch can be independently set, Mode 3A is also useful when designing pole-zero filters such as elliptics. SCN = SWITCHED-CAPACITOR NETWORK Figure 5. Filter Mode 2: Second-Order Bandpass, Lowpass and Notch SCN = SWITCHED-CAPACITOR NETWORK Figure 6. Filter Mode 3: Second-Order Bandpass, Lowpass and Highpass MAXIMPin Programmable Universal and Bandpass Filters MODE 3A Re SCN = SWITCHED-CAPACITOR NETWORK Figure 7 Filter Mode 3A: Second-Order Bandpass, Lowpass, Highpass and Notch. For elliptic LP BP. HP and Notch, the N output is used. MODE 4 (Figure 8) is the only mode that provides an allpass output. This is useful when implementing group delay equalization. In addition to this, Mode 4 can also be used in all pole lowpass and bandpass filters. Along with Mode 1, it is the fastest operating mode for the filter, allhough the gains are different than in Mode 1. When the allpass function is used, note that some amplitude peaking occurs (approximately 0.3dB when Q = 8) at fp. Also note that fp and Q sampling errors are highest in Mode 4 (see Figure 17). MODE 4 > AP(N/HP) -~ BP Dp SCN = SWITCHED-CAPACITOR NETWORK Figure 8. Filter Mode 4: Second-Order Bandpass, Lowpass and Allpass Description of Filter Functions The MAX263/64 performs all filter functions listed in this section. The MAX267/68 operates only as a bandpass filter. MAKIM BANDPASS (Figure 9) For all pole bandpass and lowpass filters (Butterworth, Bessel, Chebyshev) use Mode 1 if possible. If appro- priate fo. /fp or Q values are not available in Mode 1, Mode 2 may provide a selection that is closer to the required values. Mode 1 however has the highest bandwidth (see Table 1). For pole-zero filters such as elliptics see Mode 3A. -H $(W/Q) OBP $2 + S(w/Q) + wo? Hopp = Bandpass output gain at w = w, G(s) fo = @,/2m = The center frequency of the complex pole pair. Input-output phase shift is ~-180 at O- Q= The quality factor of the complex pole pair. Also the ratio of fp to -3dB bandwidth of the second-order bandpass response. LOWPASS See Bandpass text. (Figure 10) Wo? S? + S(w,/Q) + Ww? Ho_p = Lowpass output gain at DC fp = W/2r HIGHPASS (Figure 11) Mode 3 is the only mode with a highpass output. It will work for all pole filter types such as Butterworth, Bessel and Chebyshev. Use mode 3A for filters em- ploying both poles and zeros such as elliptics. ge $? + $(W/Q) + wo? Hoxp = Highpass output gain as f approaches fp, ,/4 fo = W/o G(s) = Hote G(s) = Hone BANDPASS OUTPUT Hopp 0.707 Hogp GAIN (W/V) Y fl fo fH 1(LOG SCALE) f Q= FOF to > \/ futw Figure 9. Second-Order Bandpass Characteristics 13 89CXVW/LOSXVIN/P9EXVW/ESSCXVINMAX263/MAX264/MAX26 7/MAX268 Pin Programmable Universal and Bandpass Filters LOWPASS OUTPUT A Hop om H = OLP = 0.707 Hop z o- 4 3 fp fc - f(LOG SCALE) tetas Lr adah~ J (aga ip=to \/ 14 aa 1 Hop = Hore * 1 /,. a ~ aQ2 Figure 10. Second-Order Lowpass Characteristics { HIGHPASS OUTPUT a Hop |__... ... a S Hoxp > z 0.707 Honp <q So fc tp - (LOG SCALE) 4 fo = fo * ., Ly (1 - 3q3) + (1-sqa)P +1 ory 1 fp ~ fo = a | i ee 1 /,.4 Q 4qz Figure 11. Second-Order Highpass Characteristics Hop = Houp < NOTCH (Figure 12) Mode 3A is recommended for multi-pole notch filters. In 2nd order filters, Mode 1 can also be used. The advantages of Mode 1 are higher bandwidth compared 14 to mode 3 (Higher fy can be implemented) and no need for external components as required in Mode 3A. g2 + Wy? $? + (w/Q) + Wo? Hone = Notch output gain as f approaches fo, /4 Hon: = Notch output gain as f approaches DC fn = w,/2m G(s) = Hone Honz Honi GAIN (VV) Hon La in f(LOG SCALE) Figure 12, Second-Order Notch Characteristics ALL PASS Mode 4 is the only configuration in which an allpass function can be realized. s2 - s(w,/Q) + wo? G s) = H esas OO (8) = Hoar coy 8(Wp/Q) + Wo? Hoap = All pass output gain for DO <f < fo, n/4 fo= wo 2 _...... Filter Design Procedure The procedure for most filter designs is to first convert the required frequency response specifications to fps and Qs for the appropriate number of second-order sections that implement the filter. This can be done by using design equations or tables in available liter- ature, or can be conveniently calculated using Maxims filter design software. Once the fp and Qs have been found, the next step is to turn them into the digital program coefficients required by the filter. An oper- ating Mode and clock frequency (or clock/center frequency ratio) must also be selected. Next, if the sample rate (fo. ,/2) is low enough to cause significant errors, the selected fs and Qs should be corrected to account for sampling effects by using Figure 17 or Maxims design software. In most cases, the sampling errors are small enough to require no correction, i.e. less than 1%. In any case, with or without correction, the required fps and Qs can then be selected from Tables 2 and 3. Maxims filter design software can also perform this last step. The desired fps and Qs are stated, and the appropriate digital coefficients are supplied. MAXIMPin Programmable Universal and Bandpass Filters Multiple Feedback Bandpass Filters An alternate implementation of all-pole bandpass filers (i.e. Butterworth, Chebyshev) requires only one clock and common programming for ali second- order sections. This can be useful with MAX26X pin-programmed filters since the two second-order halves must be programmed with the same foik/fo ratio and Q (although they may use different clocks). As shown in Figure 13, external resistors connect the outputs of cascaded filter sections to a summing op-amp at the input. Since each 2nd-order section inverts (gain = ~Q) the output from odd numbered sec- tions (except for the first) must be inverted before being fed back as in the 8th-order example in Figure 13. The MAX267/68 has an on-chip amplifier for this purpose but the MAX263/64 requires external op-amp(s). In multiple feedback filters, the bandpass response is a function of the clock, fo.x/fy ratio, Q, and feedback resistor ratios. In Table $ constants for calculating resistor ratios in common bandpass configurations are listed. Maxim's filter design program BP also selects resistors for multiple feedback bandpass Gesigns. A 4th-order design example (Figure 13) best illustrates how Table 5 is used. Multiple Feedback Example Requirements: 4th-order Chebyshev with 1 dB pass- band ripple, fp = 10kHz, and bandwidth (BW) = 2kHz. 1) The overall filter Q is Qe = fp/BW = 10KHz/2kHz = 5 Table 5. Multiple Feedback Bandpass Filter Constants 2) From Table 5: Ka = 1.8219 3) The Q of each 2nd-order selection is Qp = Qe x Kg = 5x 1.8219 = 9.09 4) Rr is selected, 10kQ is a convenient value. 5) Ro = KoRe(Qp/2)? = 1.5039 x 10k x (9.109/2)" = 312k In higher order filters, the general equation is: Ry = KyRe(Qp/2)8 6) Ro sets the overall gain, A: Ry = KoR-(Qp/2)2/A, so for a gain of 1: Rp = 1.0930 x 10k x (9.109/2)2/1 = 226.8k. In higher order filters the general equation is Ro = KoRe(Qp/2) where M = (order of filter)/2. 7) The filter fy can be programmed using a wide range of clock frequencies and feix/fo ratios. If feck = 1MHz, then forx/fp = 100 (code 00000 = 10053) results in fo = 10kHz. 8) A 2.5pF to 10pF capacitor may be required across Ro to prevent response peaking. Cascading Filters in some designs, such as very narrow band filters, several second-order sections with identical center frequency may be cascaded without multiple feed- back. The total Q of the resultant filter is: -__.@. Total Qy Vain 4) Q is the Q of each individual fitter section, and N is the number of sections. In Table 6, the Total Q and TYPE (RIPPLE) ORDER KO K2 K3 K4 KQ Butterworth 4 2.0000 4.0000 1.4142 (3.0 dB) 6 2.3704 2.6667 9.1429 4.5000 8 2.9142 2.000 5.8284 14.315 1.5307 Chebyshev 4 1.6983 2.9512 0.8430 (0.1 dB) 6 1.3183 1.2137 4.5125 1.5473 8 0.7986 0.5782 1.8809 2.0343 2.2176 Chebyshev 4 4.5757 2.5998 1.0378 (0.2 dB) 6 1.1128 0.9894 3.7271 1.8413 8 0.5891 0.4551 1.4954 1.3309 2.6057 Chebyshev 4 1.3405 2.0161 1.4029 (0.5 dB) 6 0.8143 0.6897 26447 2.3944 8 0.3389 0.3040 1.0114 0.6365 3.3406 Chebyshev 4 1.0930 1.5039 1.8219 (1.0 dB) 6 0.5822 0.4756 1.8475 3.0354 8 0.1869 0.2038 0.6840 0.3002 41981 Chebyshev 4 0.9192 4.1934 2.1688 (1.5 dB) 6 0.4515 0.3616 1.4145 3.5705 Chebyshev 4 0.7866 0.9767 2.4881 (2.0 dB) 6 0.3641 0.2878 1.1308 4.0660 Chebyshev 4 0.6769 0.8148 2.7962 (2.5 dB) 6 0.3005 0.2353 0.9275 4 5462 Chebyshev 4 0.5875 0.6886 31013 (3.6 dB) 6 0.2519 0.1959 0.7739 5.0231 MA AKI 15 89TXVW/LOSXVIN/P9IEXVW/ESCXVANMAX 263/MAX264/MAX26 7/MAX268 Pin Programmable Universal and Bandpass Table 6. Cascading Identical Bandpass Filter Sections Total Sections Total B.W. Total Q 1 1.000 B 1.00 Q 2 0.644 B 1.55 QO 3 0.510 B 1.96 Q 4 0.435 B 2.30 Q 5 0.386 B 2.60 Q Note: B = individual stage bandwidth, Q = individual stage Q. bandwidth are listed for up to five identical second- order sections. B is the bandwidth of each section. In high order bandpass filters that do not use multiple feedback, stages with different fos and Qs may also be cascaded. When this happens the overall filter gain at the bandpass center frequency is not simply the product of the individual gains because fos, the frequency where each sections gain is specified, is different for each second-order section. The gain of each section at the cascaded filter's center frequency must be determined to obtain the total gain. Filters For all-pole filters the gain, H(f,), at each second-order section's fy is divided by an adjustment factor, G, to obtain that sections gain, H(fogp), at the overall center frequency: Hy (foge) = H(fo1)/G, = Section 1's Gain at fggp _ QaL(F,2 = 1)? + (F,/Q,)217 = = where F, = fo:/fogp G,, Q,, and fg; are the gain adjustment factor, Q, and fy for the first of the cascaded second-order sections. The gain of the other sections (2, 3 etc.) at fogp is determined the same way. The overall gain is: H(fope) = Hy(fope) * Halfogp) * ete. For cascaded filters with zeros (fz) such as elliptics, the gain adjustment factor for each stage is: _ QyLF zi? - F 2] [(F,2 - 1)? + (F,/Q,)2] Fy?(Fz4? ~ 1) where Fz, = {z,/fogp, and F, is the same as above. Gy G, 4TH-ORDER MULTIPLE (A VALUES IN ( ) ARE FOR Re MAL ? ! (312k) ! | Re bee ow tt oe oe Ro (10k) 2 IN Ww #1- IN BP iN BP 2ND-ORDER 2ND-ORDER ~ (227kQ) I, FILTER A FILTER B OUT ra 2 C2-Ca: 2.5pF TO 10pF FEEDBACK CAPACITORS ARE SOMETIMES NEEDED TO PREVENT RESPONSE FEEDBACK BANDPASS DESIGN EXAMPLE IN TEXT) PEAKING. + ANAY + : Ra r u---4 +-- wA R3 Cy + NAr sd Ra cs | Low. = > db R i C2 : Je L---4p---4 WA 40kKO Ro IN AA AO 2ND-ORDER 2ND-ORDER 2ND-ORDER 2ND-ORDER __ BP + FILTER A = [77]_~sFILTER B FILTER A FILTER B * out = #1 #2 #3 #4 Ic #1 iC #2 8TH-ORDER MULTIPLE FEEDBACK BANDPASS NOTE: IN MULTIPLE FEEDBACK FILTERS, ALL 2ND-ORDER SECTIONS USE THE SAME feix, icux/te RATIO, AND Q. Figure 13. Multiple Feedback Bandpass Biock Diagram (See Text for R Values) 16 MA AXIMPin Programmable Universal and Bandpass Filters _._____... Application Hints Power Supplies The MAX263/64/67/68 can be operated with a variety of power supply configurations including +5V to +12V single supply, or 2.5V to 6V dual supplies. When a single supply is used, V is connected to system ground and the filter's GND pin should be biased at v*/2. The input signal is then either capacitively coupled to the filter input or biased to V"/2. Figure 14 shows circuit connections for single supply operation. Power consumption at +5V is reduced if CLK, and CLKg are driven with +5V, rather than TTL or 0 to 5V levels. Operation with +5V or +2.5V power lowers power consumption but also reduces bandwidth by approximately 25% compared to +12V or <:5V supplies. Best performance is achieved if V* and V" are bypassed to ground with 4.7yF electrolytic (Tantalum is pre- ferred.) and O.1uF ceramic capacitors. These should be located as close to the supply pins as possible. The lead length of the bypass capacitors should be shortest at the V and V pins. When using a single supply V* and GND should be bypassed to V" as shown in Figure 14. Output Swing and Clipping MAX26X outputs are designed to swing to within 0.15V of each supply rail with a 10kQ load. To ensure that the outputs are not driven beyond their maximum range (output clipping), the peak amplitude response, individual section gains (Hogp, Hoip, Hoxp). input signal level, and filter offset voliages must be carefully considered. It is especially important to check UNUSED outputs for clipping (i.e. the lowpass output in a bandpass hookup) because overload at ANY filter stage severely distorts the overall response. The maximum signal swing with +4.75V supplies and a 1.0V filter offset is approximately +3.5V. For example let's assume a fourth-order lowpass filter is being implemented with a Q of 2 using Mode 1. With a single 5V supply (i.e. +2.5V with respect to chip GND) the maximum output signal is +2V (w.r.t. GND). Since in Mode 1 the maximum signal is Q times the input signal, the input should not exceed +(2/Q)V, or +1V in this case. 40k 10k0 ry 4 Vin ov ~~~ VO NOTE . 2.5kQ NS wl 75KO = / TO vt lov bRe_~< vy Sor anyoc / a S INp < TO GND PIN iNB ~ -8V _- iin Nu MAXIM ov MAX263 v* at +5V MAX264 MAX267 MAX268 GND | O4yF AS 4.7uf TT OAuF ve NOTE: OP AMP LEVEL SHIFT CIRCUIT WILL HAVE A GAIN OF 0.5 FROM V*. Figure 14. Power Supply and Input Connections for Single MAXIM Supply Operation 17 897CXVW/LOSXVW/P9OSCXVW/ESSXVINMAX263/MAX264/MAX 26 7/MAX268 Pin Programmable Universal and Bandpass Filters A IWODIV. ov B 5mV/DIV. ov C 5mwDW. ov 1yS/DIV, Cae r-] Na BPA TRACE C 7 R, 10kQ I C, 1000pF TRACE A ~{ CLK, * 500kHz TTL MAX263 Figure 15. MAX263 Bandpass Output Clock Noise Clock Feedthrough and Noise Typical wideband noise for MAX26X series devices is O.5MVpp from DC to 100kHz. The noise is virtually independent of clock frequency. In multistage filters, the section with the highest Q should be placed first for lower output noise. The output waveform of the MAX26X series and other switched capacitor filters appears as a sampled signal with stepping or staircasing of the output waveform occurring at the internal sample rate (fo,,/2). This stepping, if objectionable, can be removed by adding a single pole RC filter. With no input signal, clock related feedthrough is approximately 8mV,,,. This can also be attenuated with an RC smoothing filter as shown with the MAX263 in Figure 15. Input impedance The filter input model is shown in Figure 16. Input capacitor Ca, is shunted by Cg which is switched at one half the input clock frequency (Fe, ,/2). The input impedance is described by: Ryy = ane X fer). There is also a fixed stray capacitance of about 5pF to ground. Digital Inputs Filter programming is accomplished by tying input pins MO, Mi, FO-F4, and Q0-Q6 to high or low voltage levels, typically V* and V~. Inputs are not internally pulled up or down, so these inputs must not be left unconnected. Input thresholds are guaranteed to be no higher than V' 0.5V and no lower than V- +0.5V. When driving the cigital inputs (i.e., the digital inputs 18 198 Lf e-,Do VIN TT -sor] = {Ca 12pF = 4 a ae IN 750 Ca fork Figure 16. MAX263/64/67/68 Input Model are tied to microprocessor |/O lines), additional protec- tion is provided by placing a 1k resistor in series with ihe programming pins. If pull-up resistors are used with switches at the programming inputs, as might be the case in prototype breadboards, the pull-up resistors should be no more than 3.3kQ2. fg and @ at Low Sample Rates When low fe_k/fp ratios and low Q settings are select- ed, deviation from ideal continuous filter response may be noticeable in some designs. This is due to interac- tion between Q, and fo at low fc. /fp ratios and Qs. The data in Figure 17 quantifies these differences. Since the errors are predictable, the graphs can be used to cor- rect the selected fp and Q so that the actual realized parameters are on target. These predicted errors are not unique to MAX26X series devices and in fact occur with all sampled filters. Consequently, these corrections can be applied to other switched-capacitor filters. In the majority of cases, the errors are not significant, i.e. less than 1%, and correction is not needed. However, the MAX264/68 does employ a lower range of fc, j/fp ratios than the MAX263/67 and is more prone to sampling errors as the tables show. Maxims filter design software applies the previous cor- rections automatically as a function of desired fo_K/fo, and Q. Therefore, Figure 17 should NOT be used when Maxims software determines fg and Q. This results in overcompensation of the sampling errors since the cor- rection factors are then counted twice. The data plotted in Figure 17 applies for Modes 1 and 3. When using Figure 17 for Mode 4, the fo error obtained from the graph should be multiplied by 1.5 and the Q error should be multiplied by 3.0. In Mode 2 the value of ferK/fg should be multiplied by v2 and the programmed Q should be divided by v2 before using the graphs. MAKINPin Programmable Universal and Bandpass Filters fg ERROR vs fo, x/fp RATIO fy error is plotted for Modes 1 and 3 MODE 2: Multiply foux/ty by V2and divide Q by ./2 before using graph MODE 4: Multiply (, error by 1.5 Q= 0.512: = = 0.69 x = 5 i = Q=121 a + Q=229 Q= 4.94 40 60 80 100 120 140 160 180 200 foix/a RATIO Q ERROR vs fo. ,/ty RATIO =F t t T T T Q error is plotted for Modes 1 and 3 6 \ MODE 2: Multiply foux/fy by V Zand divide Q by \/2 before using graph \ : MODE 4: Multiply Q error by 3.0 5 i ~ (Q=05 2 Ne F 4 Ss = 0,6 bmn bee < = & 3 \ X\ Q = 0.83 = 121 o ASA Q ; -2 NX my = OQ = 3.05 Q= 711 -t RSSS 0 40 66 88 100 120 140 160 180 200 ferx/fp RATIO Figure 17 Sampling Errors in to, x/fy and Q at Low fg, /fo and Q Settings Aliasing As with all sampled systems, frequency components of the input signal above one haif the sampling rate will be allased. In particular, input signal components near the sampling rate generate difference frequencies that often fall within the passband of the filter. Such aliased signals, when they appear at the output, are indistinguishable from real input information. For example, the aliased output signal generated when a 99kHz waveform is applied to a filter sampling at 1O0kKHzZ, (fork = 200KHZz) is 1kHz. This waveform is an attenuated version of the output that would result from a true 1kHz input. Remember that with the MAX26X series filters, the nyquist rate (one half the sample rate) is in fact fo.,/4 because fo. is internally divided by two. MAKIM A simple passive RC lowpass input filter is usually sufficient to remove input frequencies that can cause aliasing. In many cases the input signal itself may be band limited and require no special anti-alias filtering. The wideband MAX264/68 uses lower fo. ,/fp ratios than the MAX263/67 and for this reason is more likely to require input filtering than the MAX263 or MAX267. Trimming DC Offset The DC offset voltage at the LP or Notch output can be adjusted with the circuit in Figure 18. This circuit also uses the input op-amp to implement a single pole anti-alias filter. Note that the total offset wiil generally be less in multistage filters than when only one section is used since each offset is typically negative and each section inverts. When the HP or BP outputs are used, the offset can be removed with capacitor coupling. Cy tL Re 100kQ uN Ving A Rg 270k0 Ry 100k sv yp TO FILTER INPUT 400K OFFSET = TRIM GAIN = -Ry/Rg ~5V 1 LP BRC Figure 18. Circuit for DC Offset Adjustment Design Examples 4th-Order Multiple Feedback Bandpass--MAX 268 In Figure 19, a pin-programmed MAX268 operates as a 4th-order 50kHz Chebyshev bandpass. The specifi- cations are: Center frequency (fo) = 50kHz Pass bandwidth = 10kHz Max. passband ripple = 0.1dB Gain at center freq. = 1VW/V Two identical 2nd-order sections and the internal op amp are used with multiple feedback. The general form is as in Figure 13. Maxims design program, BP generates the programming codes and feedback resis- tor values. With a 2.5MHz crystal clock the realized parameters are: Center frequency = 50.305kHz Pass Bandwidth = 10.07kHz Programmed fo, x/fg ratio = 50.27 (N = 3) Programmed g = 4.27 (N = 113) (desired Q = 4.215) Actual Q (with error correction) = 4.21 Resistors: Ro = 131k, Rg = 75kQ, Re = 10k 19 89ZXVW/LOSXVIN/P9ICXVW/E9CXVINMAX263/MAX264/MAX267/MAX268 Pin Programmable Universal and Bandpass Filters 8 9 S -16 z=. < 24 ~32 -40 ~48 ~56 10k 20k 50k 100k FREQUENCY (Hz) C2 if it 2.5pF Re AW 131K Re AAA, e Vout 10kQ 315 |2 4/1 24 Ro OP OUT! INag BP, INg BPa 9 Viney 4 {OP INES F4 v- 7BKQ Oo . F3 2 21 n| = MAK F2 5 _e CLKa MAX268 FA 22 2.5MHz 1 CLKg FO I vt Q6 Q5 G4 Q3 Q2 Q@1 Qo [7 G [23 20 [19 [14 fs Vi o_e4 Vv Figure 19. 4th-Order 50kHz Chebyshev Bandpass Using Multiple Feedback Other ciock rates and fo, ,/fp ratios can be chosen to implement the same fifter, but larger foyx/fp ratios provide performance closer to the ideal. Capacitor Cy may be needed to prevent response peaking at the passband edge. In this example C, = 2.5pF. Multiple feedback can also be extended to 8th-order designs while still using one clock by adding a second MAX268 and 2 additional feedback resistors. These can also be calculated with the design program, BP. Note that for filter order above 4, the feedback signal from odd filter sections is inverted before it is summed (see Figure 13). 20 Vin Vour |s [2 1 | 1 IN, BP, ING BPp | , 1.89 CLKa F4 v- Miz | F3 10 om AVIA AL svi F2 21 = osc our MAX268 FA = 2.5MHz 1 CLKs FO vt Q6 Q5 Q4 G3 G2 a1 a0 7 {6 [23 20 l19 [14 {43 ve 44 v- Figure 20. 4th-Order 50kHz Chebyshev Bandpass Using No External Resistors 4th-Order Band; (No Multiple Feedback)MAX268 Without multiple feedback, the previous example can be implemented with no external components, how- ever separate clocks are required for CLK, and CLKs (Figure 20). The target specifications are the same as before. The realized parameters are now: CLK, = 1.89MHz, CLK, = 2.5MHz Center frequency = 50kHz Pass bandwidth = 10kHz Programmed fo. k/fg ratio = 43.98 (N = 1) Programmed G = 4.27 (N = 113) (desired Q = 4.215) Actual Q (with error correction) = 4.2 With the chosen fo, /fg ratio, a crystal may be used at CLK, while a divi ed system clock, if available (2.5, 5, 10, or 20MHz), drives CLKg. This is suggested because CLK, has internal circuitry to drive a crystal while CLKg does not. Other clock sources may be used with a different programmed fo.x/fp as long as the ratio between CLK, and CLK, remains the same as above. Another advantage of this circuit is that higher. center frequencies can be achieved relative to equivalent multiple feedback designs because lower Q sections are used compared to multiple feedback. MAXIMPin Programmable Universal and Bandpass Filters 8 F&F wi = ie S g > -180 a 2 6 % < -360 = 2 20 2K 20K FREQUENCY (Hz) FILTER 1 FILTER 1 IN | v- V- OuT v- vt v- y | ve INA MO M1 LPa Q6 V INn MO M1 LPa a6 Fa Qs Fa Q5 . F3 MAK vl Q4 F3 MAKLM 4 V F2 MAX263 a3 F2 MAX263 03 Fi a2 Fi a2 FO ai FO a1 INg CLKa CLKg LPg G0 INg CLKa CLKg LPg G0 ve i ve A J FILTER 2 FILTER 2 IN OuT CLK (400kHz) - CLKag |MODE| fox fon Q, Qs CLK, |MODE| fon fog Qa Qs 400kKHz 1 Newt|N= 111N = 79[N = 79 400kHz 1 N=12|N=12|N=2=71IN271 Figure 21. Dual Tracking 3kHz 4th-Order Lowpass Dual 4th-Order Tracking LowpassMAX263 In Figure 21, two Butterworth lowpass filters are set up to accurately track each other. By splitting two MAX263s only one clock is needed. The specifications are: Cutoff frequency = 3kHz fon = fog = 3kHz Gh = 2307, Qg = 0.541 These values can be programmed directly into the filter. However, since the Qs are low, sampling errors may be large enough to deserve attention. From Figure 17, if forx/fp is near 130 (for, iS 400KHZ), fon and fo will be about 4% and 1.5% high respectively. Qa and Qg, will be 1.2% and 0.5% low. These errors may not be large enough to worry about but are corrected here (within the programming resolution of the MAX263) MAXIMA by the filter design programs PZ and MPP fo, and f are programmed to different values (N, = 11 Ng = a for this reason. Mode 1, CLK, = CLKg = 400kHz fc ton = 13 08, N= 11 ttarget fo, = 2961Hz, actual = 3008Hz) fox/fon = 138.23, N = 12 Yfarget fog = 2894Hz, actual = 3015Hz) Q, = 1.31, N = 79 (actual Qy = 1.30) Qn = 0.547, N = 11 (actual Qg = 0.542) ai S9ZXVW/LOTXVIN/POSCXVWV/ESSCXVINMAX263/MAX264/MAX26 7/MAX268 Pin Programmable Universal and Bandpass Filters _._ Ordering Information (continued) PART TEMP. RANGE PACKAGE* ACCURACY MAX264AEPI -40C to +85C Plastic DIP 1% MAX264BEPI -40C to +85C Plastic DIP 2% MAX264ACWI 0C to +70C ~=- Wide SO 1% MAX264BCWI 0C to +70C = Wide SO 2% MAX264AMJI --55C to +125C ~=CERDIP 1% MAX264MBJI - -55C to +125C CERDIP 2% MAX267ACNG oC to +70C ~ Plastic DIP 1% MAX267BCNG 0C to +70C ~Plastic DIP 2% MAX267AENG -40C to +85C Plastic DIP 1% MAX267BENG -40C to +85C Plastic DIP 2% MAX267ACWG 0C to +70C = Wide SO 1% MAX267BCWG 0C to +70C = Wide SO 2% MAX267AMRG = -55C to +125C ~=CERDIP 1% MAX267BMRG_ -55C to +125C ~=CERDIP 2% MAX268ACNG oC to +70C ~Plastic DIP 1% MAX268BCNG 0C ta +70C Plastic DIP 2% MAX268AENG -40C to +85C Plastic DIP 1% MAX268BENG ~-40C to +85C Plastic DIP 2% MAX268ACWG oC to +70C ~=3= Wide SO 1% MAX268BCWG 0C to +70C = Wide SO 2% MAX268AMRG = -55C to +125C =CERDIP 1% MAX268BMRG_~-55C to +125C CERDIP 2% * MAX263/264 packages are 28-pin 0.6 wide DIP and 28-pin 0.3 wide SO (Smaii Outline). MAX267/268 packages are 24-pin 0.3" narrow DIP and 24-pin 0.3 wide SO (Smail Outline). 22 N.C.(OP_ IN) - M1 (N.C.) MO (N.C.) - Chip Topography 0.128 g__ (3.251mm) -. a N.C. (OP OUT) | HP HPa 8 | BPaALPa INBLPBBPB| gy FO F2 -Q3 #2 | 0.199" (5.055mm) | | | -OSC OUT | bGND i F4 F3. QO Q1 FI ClLKa CLKg v- NOTE: LABELS IN PARENTHESES ( ) ARE FOR MAX 267/268 ONLY MAXIMPin Programmable Universal and Bandpass Filters Package Information TED L De c BO NOT INCLUDE MOLD FLASH 2, MOLD FLAZH OF PROTPUC TON? NOT TO EXCEED Sm 1.006 3, CONTPOLLING DIMEN? TD MILLIMETEP 4, MEETC JEDEC MiOl-x% at CHOWN IN ABOVE TABLE 5. SIMILIAP TO JEDEC MO-d53AB 5. N = NUMBEP OF PINZ AVIAAIAVA Pace GE FeMiLy OUTLINE: PDIP 300" | [ot a a a i a ON 1 fat (43 4 L_ LA rime corn. ner_res_! NOTES: 1, De c DBO NOT INCLUDE MOLD FLATH MOLD FLAZH OF PROTPL: TON? MOT TO EXCEED Sturm 1006 LEADS TO BE COPLAMAP WITHIN ern 004 CORTPOLLING DIMEN TOM MILLIMETER CIMILIAP TO JEDEC MOO1S wh IM ABOVE TABLE Mo= NUMBEF OF PINT Ww fur a sor fa 7oo ch | alan WE ou44 Ay MAAN 23 B9ICXVW/LICXVW/PICXVWEICXVWNMAX263/MAX264/MAX267/MAX268 Pin Programmable Universal and Bandpass Filters Package Information (continued) cas [oi CO tobe [0 Ch mols [o.3 Coos pots [es ps2 iRebatat [As EO Sa O34 0.419 [o.oo] i.e UCL | OoSo | es | 7S MOLE [VCS | Ott a7 TED: . Dee BO NOT INCLUDE MOLD FLASH 2. MOLD FLATH OP PROTRUSION? NOT TO EXCEED JSmm +006"! 3, LEADS TO BE COPLANAP WITHIN Adzrm 1.0047" 4, CONTPOLLING DIMENSION: MILLIMETER 5. rls SHOWN 6, MEETS JEDEC Mi0L3-s% AT IN ABOVE TABLE 5. M = MUMBEP OF PINS | MAXIM |Pock 0c FAMILY QUTLINE: SOIC 300" a [el-ul4e A | SONIA CONPOL HOE _ Ew 2! 2 TiiAihiae +| be a7 (La75 99 {i600 NOTES: MLO on 1, CONTPOLLING DIMEMZION! INCH C.2n j 2 MEET? 1835 CATE OUTLINE CONFIGUPATION #1 ss 4&2 SHOWN IM ABOVE TABLE aor act ae 3 N= NUMBEP OF PIN COS a O..or a.m Maxim cannot assume responsibility for use of any circuitry other than circuitry entirely embodied in a Maxim product. No circuit patent licenses are implied. Maxim reserves the right to change the circuitry and specifications without notice at any time. 24 Maxim Integrated Products, 120 San Gabriel Drive, Sunnyvale, CA 94086 408-737-7600 1998 Maxim Integrated Products Printed USA MAXIM is a registered trademark of Maxim Integrated Products.