ADS5562: Collected data have serrated wave

Hi,

Well, first off I notice that you have an amplifier powered from a 5V supply, but the VCM for the ADC is about 1.5V, so the amplifier is being asked to drive a signal that has very little headroom on the lower side of the signal but a lot of headroom on the upper side.  Since the full scale input of the ADC is almost 3.6V peak to peak differential, a full scale signal from the amp would have to swing from almost 0.9V below VCM to almost 0..9V above VCM.   You may be clipping the low side of the signal, and if so this would cause severe distortion and in addition to that the instantaneous VCM of the signal at the moment the low side is clipped and the high side isn't would be far from the desired VCM.  The output codes from the ADC should be centered around midscale, and the codes that you show do not - they sit at code zero for a while and then somewhat approximate part of the signal shape above that. 

What is the frequency of your input signal?  Is the frequency high enough that you could you temporarily insert AC coupling caps in the analog signal path between the amp and the ADC so that the ADC can bias the signal to the desired VCM while not asking the amp to bias the signal down that low?  (making sure to remove the VCM connection to the amp when you do so.)  this would allow you to determine if the only issue is that the VCM is too low for the amp, or if there are other issues as well.    One thing we have done when there is too large a mismatch between the ADC VCM and what the amp can support for VCM is we would not power the amp from +5V to ground.  We might give the amp a 4V supply on the positive supply and a -1V supply on the lower supply, so that the amp will naturally be most centered with its output VCM around 1.5V.      Also, this ADC you are using has a much larger full scale range than many of our other ADCs, so even with the common mode well centered for the amp midway between the supply rails there may still be a little distortion at the peaks when you try to get a full scale output - compared to using an ADC with a full scale range of 2V peak to peak differential.    But at present, it looks like you are having headroom problems on the low side of the signal due to the ADC VCM being too low for the amp.

Regards,

Richard P.