THS4531A: ~14MHz Instability on output

Hello Evan,

  Thank you for sharing this information with us, and the detailed debugging. It does seem like the FDA is oscillating. As a test, would it be possible to lower the feedback capacitors C152 and C166 to <10pF?

Thank you,

Sima 

Sima,

I changed both C152 and C166 to 10pF, no appreciable difference in circuit behavior. 

Thanks,
Evan

Hello Evan,

  Thanks for trying out the suggestion. Depending on parasitic capacitance, this issue might have been mitigated by shaping noise gain with lower feedback caps.

  I ran a stability check, and it does seem unstable with a phase margin of 14 degrees:

   We most likely have to compensate the circuit in other ways that would be hard to modify on this board. As a quick test again, keep the 10pF feedback capacitors and would you be able to change out the resistors to a higher gain? This will increase the noise gain and intersect with AOL curve at a lower frequency, thus making the circuit more stable, but limiting overall bandwidth.

Thanks,

Sima

Sima,

I have just tried this with the values above (392R). The waveforms are still oscillating but do look slightly different.

R219/C152 node:

At ADC_AIN+ node:

Obviously signal amplitude a lot higher on ADC_AIN+ node now since we increased the amp gain. But the ~14MHz is still there. 

Hi Evan,

you should not probe directly at the input pins of THS4531. The probe capacitance can destabilize the OPAmp. Connect a suited isolation resistor to the input of THS4531 first and probe at the cold end of this resistor. By this the probe capacitance is isolated from the input of THS4531. Or better probe at the output of THS4531. And don't forget the isolation resistor.

Kai

Thanks for the tip. However, the oscillations are still present at the output of the amp, with no probe connected at the input. 

And what isolation resistor do you use at the output?

Kai

None. Is 3.9pF of probe load capacitance really going to cause that instability ringing?

Hello Evan,

  Sorry for the long delay between replies. I ran a couple more stability variations, and it seems like your output RC stage to the ADC is mainly helping with overall stability. 3.9pF load capacitance might cause trouble at unity gain, but with the increased gain it shouldn't still be oscillating. 

  One more thing you can try. Can you keep the increased gain (2k  RF, 392 RG), but now remove both feedback 10pF capacitors? 

  If that doesn't work, the only other variabilities would be layout and the shared power supply. Would you be able to send us a pdf or snippet of the layout section near power supplies + THS4531A? If you would like to share these privately, I have sent you a friend request.

Thank you,
Sima 

Hi Evan,

the THS4531 would not only see the 3.9pF probe capacitance but the complex load of the whole probe cable. You can easily check whether this is an issue here by adding an isolation resistor.

Kai

Removing 10pf caps didn't change behavior. I sent you a private message.

Using 120ohm isolation resistor didn't change anything on the measurement.

Hi Evan,

thanks for the feedback 

It's rather weird what is going on in your application. I carried out some phase stability analysis and found that your THS4531A circuit (including R225 and R228) is stable. And the OPA320A circuit is stable as well.

I would try to disconnect and/or deactivate all the sections surrounding the THS4531A - one after the other - until the 14MHz oscillation disappears. Start with the input resistance switch, proceed with the ADC and then the OPA320. Even replace the +3.3V supply voltage by another power supply. If all this doesn't help, the THS4531A itself must be the wrongdoer. Replace the THS4531A by a fresh one and check again. Or you might need to change the layout. Or add shielding, if the 14MHz comes from outside. By the way, are you sure that all components values are correct? Sometimes caps are wrong by a factor of 10, 100 or else. Then the circuit might no longer be stable.

Somewhere must be the cause of this weird 14MHz oscillation.

Kai