high frequency ripple distortion seen at output of OPA381

Hi Tom,

I suspect the 500 kHz signal is actually a low-level oscillation. If the capacitance of photodiode is too high, for the compensation that is used, there can be enough phase shift introduced into the feedback loop to cause instability. Increasing the compensation capacitance stabilizes the loop, but reduces the bandwidth as you have observed. The OPA381 transimpedance bandwidth is discussed on page 10 in the data sheet.

Is there any possibility of using a photodiode having lower capacitance? Also, you might be able to reduce the photodiode junction capacitance by increasing the reverse bias on the diode. Your circuit already has the R13, R14 divder on the non-inverting input. Increasing that divider voltage increases the CMV which is the voltage that will be mirrored at the summing node (inverting input).

Regards, Thomas

PA - Linear Applications Engineering

Thanks Thomas for the quick reply. I'm a little confused why there would be a loop instability tho. my feedback resistance, Rf,  is ~100k and the diode capacitance is only 22pf. Looking at page 6 on the datasheet, the gain plot for diode capacitance of 20pf and rf=100k curve (which is essentially my circuit), I should have flat response out to 1MHz.

raising the DC bias will unfortunately significantly reduce my gain at this stage. could the OPA380 with higher bandwidth help resolve this issue?

Hi Tom,

The OPA381 curve that you mention is flat when a small feedback capacitor, about 2 pF, is included. Otherwise, there is substantial peaking which is an indicator of reduced phase margin. You indicated that adding a feedback capacitor eliminates the oscillation, but the resulting bandwidth is too low. Did you use a small capacitance like 2 pF, or was something larger required to eliminate the oscillation?

The OPA380 has wider bandwidth than the OPA381. You still may have to add a small feedback capacitor to keep the circuit stable, but the overall bandwidth might meet your application requirements once the required capacitance is determined.

Regards, Thomas

PA - Linear Applications Engineering

As per my original pdf attachment, I had to add a 12pf feedback cap to sufficiently suppress the oscillation at the expense of needed bandwidth. my thoughts were if the oscillation were pushed out further to a higher frequency with the OPA381 then I may be able to roll it off with a lower value cap. otherwise, I'm not sure how else I can remove the oscillation.