THS4567: Opamp oscillations root-cause debug- follow-up question to previous thread

Hello Shubhangi,

Were you able to test the use of a small isolation resistor (sometimes seen as Rout, Rseries, Riso) at each of the outputs of the FDA?  I also notice your gain-setting resistor (R125 perhaps?) is not located between the feedback loop and either the input signal or a node potential, but between the output and the feedback loop.  This is not strictly correct; have you used either TINA-TI or PSPICE-for-TI (or PSPICE in general) to simulate your circuit as you have it connected?  I would like you to confirm known good behavior of your circuit in simulation so we can focus on debugging the PCB hardware itself.  

On the PCB itself, what is the layer stack-up?  Are the traces impedance-matched?  Are the traces long enough to be antennas and couple-in high-frequency signals?

Do you have any interference sources in your lab?  

Have you tried using a different THS4567 IC on the board?  Was the soldering and board reflow done with all connections being good?

Best,

Alec

Hello Alec,

Thanks for your response.

On the PCB, I don't have a provision to test Riso.

TINA model didn't show oscillations (which I was using); p-spice model (that the designer was using) shows the oscillations.

I have already requested the designer to try with Riso in P-spice simulation.

R125 is not feedback resistor. As can be seen in the image it has been removed, (it was for the second stage opamp).

R123 and R128, R132 and R131 are our gain setting resistors.

The oscillations are reproducible on multiple PCBs. The PCB is assembled by reflow from a PCB Vendor. So no issues with soldering etc.

PCB has 4 layers. Traces for ompap are short with good solid GND reference in layer 2. Traces are symmetric and length matched. 

Thanks,
Shubhangi

Hello Shubhangi,

Would you be open to sharing the TINA-TI and PSPICE model simulations with me?  We can do this offline over email if the information is sensitive to share on the forum.  I want to see the oscillations in simulation, and this would be easier than me examining one of your PCBs.  You are certainly more than welcome to send a board to my office if you would be interested in doing so, but this would be as a courtesy and not a part of any official return or mediation process, as I am sure you understand.

My job is to help assist you with the understanding and fixes for these oscillations.  I have sent a friend request to you via e2e.  

Can you confirm the TINA-TI model and PSPICE model are the same?  You can view the netlists and compare; I can also do this if you share the model designs with me.

Thank you for the clarification on R125 and the gain-setting resistors.  Could you use the R125 pad to test Riso on the production PCB, or was R125 removed prior to the PCB fabrication?  For Riso you need a series resistor (or other passive pad) at the output of the FDA (one Riso for each output).

Best,

Alec