AnsweredAssumed Answered

AD4661 output distored when cap coupled to gnd

Question asked by m3atwad on May 15, 2015
Latest reply on May 29, 2015 by harryh

Hello,

 

I have an AD4661 IC in a design right now (dual package op amp).  I had this working fine, but was required to make a small modification that could be causing an issue I don't totally understand.  The input to this amp is ~50mV sine wave.  The feedback path has a voltage divider consisting of a 2k resistor and a 20k pot (AD5272-20).  When the 5272 is configured with it's minimum impedance I would theoretically pass the input sine wave through with very little gain applied.  This worked when I DID NOT have the cap between the 2k resistor and ground. When I added this cap my signal got fuzzy.  You can see below in the screen shot from the scope.

 

Channel 1 (yellow) is the output of the 4661.  Channel 2 (blue) is the input of the 4661.

tek00000.png

 

I've attached a pdf of the schematic.

 

The problem occurs when the AD5272-20 is set to it's minimum resistance (it's a pot) allowing for the least amount of gain.  You can see the signal is very "fuzzy" on the output compared to the input.  When i record the data with the ADC in the product this is being developed for I get very fuzzy data.  I've also attached a screen shot of that.

 

screen of fuzzy data.jpg

 

 

So I'm trying to figure out how to clean this output up if possible. When I increase the resistance of the ad5272-20 the gain increases and this goes away.  I'm worried that there is a design flaw here now since I added that capacitor.  I need to decouple this amp from ground because it is applying a gain to ONLY the AC part of an input signal.  The input of this amp normally sits at 1.65V.  I only want to amplify the AC signal on the input which is why I added this cap.

 

 

Here is an example of what I have gotten in the past and what I would expect.  I apply a gain of 5 to the same signal and out put looks great.  This is data recorded with my ADC and plotted in the same exact manner as the distorted data with no gain applied.

 

gain of 5.jpg

 

 

 

Thanks

Rob

Outcomes