SBAS258A – JUNE 2002 – REVISED JULY 2002
www.ti.com
14
As for power supplies, separate analog and digital
supply pins are provided on the part (AVDD/DVDD). The
supply to the digital output drivers is kept separate as
well (DRVDD). Lowering the voltage on this supply to
3.0V in s t ead of the nominal 3.3V improves performance
because of the lower switching noise caused by the
output buffers.
Due to the high sampling rate and switched-capacitor
architecture, the ADS5203 generates transients on the
supply and reference lines. Proper decoupling of these
lines is, therefore, essential.
NOTES
1. Integral Nonlinearity (INL)—Integral nonlinearity
refers to the deviation of each individual code from a line
drawn from zero to full-scale. The point used as zero
occurs ½LSB before the first code transition. The
full-scale point is defined as a level ½LSB beyond the
last code transition. The deviation is measured from t h e
center of each particular code to the true straight line
between these two endpoints.
2. Differential Nonlinearity (DNL)—An ideal ADC
exhibits code t ransitions that a re e xactly 1LSB a part. DNL
is the deviation from this ideal value. Therefore, this
measure indicates how uniform the transfer function step
sizes are. The ideal step size is defined here as the step
size for the device under tes t (i.e. , (las t trans ition lev el –
first transition level)/(2n – 2 )). Using this d efinition for DNL
separates the effects of gain and offset err or.
A minimum DNL better than –1LSB ensures no missing
codes.
3. Zero and Full-Scale Error—Zero error is defined as
the difference in analog input voltage—between the
ideal voltage and the actual voltage—that will switch the
ADC output from code 0 to code 1. The ideal voltage
level i s d e t e r m i ned by adding the voltage corresponding
to ½LSB to the bottom reference level. The voltage
corresponding to 1LSB is found from the difference of
top and bottom references divided by the number of
ADC output levels (1024).
Full-scale error is defined as the difference in analog
input vo l t a g e —between the ideal voltage and the actual
voltage—that will switch the ADC output from code
1022 to code 1023. The ideal voltage level is
determined by subtracting the voltage corresponding to
1.5LSB from the top reference level. The voltage
corresponding to 1LSB is found from the difference of
top and bottom references divided by the number of
ADC output levels (1024).
4. Analog Input Bandwidth—The analog input
bandwidth i s defined as the max. frequency of a 1dBFS
input sine that can be applied to the device for which an
extra 3dB attenuation is observed in the reconstructed
output signal.
5. Output Timing—Output timing td(o) is measured
from the 1.5V level of the CLK input falling edge to the
10%/90% level of the digital output. The digital output
load is not higher than 10pF.
Output hold time th(o) is measured from the 1.5V level
of the COUT input rising edge to the 10%/90% level of
the digital output. The digital output is load is not less
than 2pF. Aperture delay td(A) is measured from the
1.5V level of the CLK input to the actual sampling
instant.
The OE signal is asynchronous. OE timing tdis is
measured from the VIH(MIN) level of OE to the
high-impedance state of the output data. The digital
output load is not higher than 10pF. OE timing ten is
measured from the VIL(MAX) level of OE to the instant
when the output data reaches VOH(min) or VOL(max)
output levels. The digital output load is not higher than
10pF.
6. Pipeline Delay (latency)—The number of clock
cycles be t ween conversion initiation on an input sample
and the corresponding output data being made
available from the ADC pipeline. Once the data pipeline
is full, new valid output data is provided on every clock
cycle. The first valid data is available on the output pins
after the latency time plus the output delay time td(o)
through the digital output buf fers. Note that a minimum
td(o) is not guaranteed because data can transition
before or after a CLK edge. It is possible to use CLK for
latching data, but at the risk of the prop delay varying
over temperature, causing data to transition one CLK
cycle earlier or later. The recommended method is to
use the latch signals COUT and COUT which are
designed to provide reliable setup and hold times with
respect to the data out.
7. Wake-Up Time—Wake-up time is from the
power-down state to accurate ADC samples being
taken, and is specified for external reference sources
applied t o the device and an 80MHz clock applied at the
time of release of STDBY. Cells that need to power up
are the bandgap, bias generator, SHAs, and ADCs.
8. Power-Up Time—Power-up time is from the
power-down state to accurate ADC samples being
taken with an 80MHz clock applied at the time of release
of STDBY. Cells that need to power up are the bandgap,
internal reference circuit, bias generator, SHAs, and
ADCs.