Advance Information WM2125
w
ww
w AI Rev 1.2 January 2003
17
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 _ LSB before the first code
transition. The full-scale point is defined as a level _ LSB beyond the last code transition. The
deviation is measured from the centre of each particular code to the true straight line between these
two endpoints.
2. Differential Nonlinearity (DNL) — An ideal ADC exhibits code transitions that are exactly 1LSB
apart. 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 test (i.e., (last transition level – first transition level)/(2n– 2)). Using this definition for DNL
separates the effects of gain and offset error. A minimum DNL better than –1LSB ensures no
missing codes.
3. Zero and Full-Scale Error — Zero error is defined as the difference in analogue 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 is determined by adding the voltage corresponding to _ LSB 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 analogue input voltage—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. Analogue Input Bandwidth — The analogue input bandwidth is defined as the max. frequency of a
1dBFS input sine that can be applied to the device for which a 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 t h(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 OEB signal is asynchronous. OEB timing tdis is measured from the VIH(MIN) level of OEB to the
high-impedance state of the output data. The digital output load is not higher than 10pF. OEB timing
ten is measured from the VIL(MAX) level of OEB 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 between conversion initiation on an input
sample and the corresponding output data being made available from the ADC pipeline. Once the
data pipelines 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 buffers. 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 COUTB 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 to 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