QWhat is the best configuration for the input of this device to measure a single
dc voltage? I have currently tied Vin- input to half AVDD and decoupled it with
100pF, and I am feeding in a DC signal into Vin+ via a 50Ohm resistor with
100pF of decoupling at the pin. Vref is set for 1V, AVDD and DRVDD are both 3V
and the clock is switching at 10Mhz and is buffered.
Is my input configuration correct, do I require further decoupling, should I
tie Vin- to Vref or to ground? I am using a Analog devices Eval. board thus the
problem cannot be layout related.
Examples of errors
1Vin Output logic LSB 0101000100 MSB
1.5Vin 1111000001 no real error
I am expecting some error as I am only using a pure DC signal at this time for
test purposes but I am seeing a great deal more error then I would expect at
some points for example at 1Vin and 2Vin.
AGrounding the inputs or connecting the inputs to a steady DC voltage and
looking at the code spread is a good way test a low bandwidth precision
converter such as the AD7705 sigma delta, here you limit the bandwidth to near
DC and therefore any noise is due to the converter itself. But this is very bad
way to test a high speed converter such as the AD9215 ADC.
The input bandwidth on the AD9215 is much larger than the Nyquist bandwidth
therefore it is not surprising that you see code flicker with the inputs
shorted. In fact, the only thing that such a noise measurement tells you is
that you have a wide input bandwidth on your converter, it tells you nothing
about the noise performance of the converter in the band of interest. You could
of course filter the output of the ADC in order to achieve a single output
code. To test a high speed converter, apply a pure sine wave to the input
capture a few thousand codes and perform an FFT. Now you can look at the
Spurious Free Dynamic Range, measure the Signal to Noise Ratio or Signal to
Noise + Distortion and calculate the Effective Number of bits over the band of