THS4021 noise

Kenneth,
I see you are using a very high feedback resistance. In such cases a FET input amplifier like the OPA657 will probably work best. Also, unless the final closed loop BW of your TIA is < 10kHz the flicker noise will not be the dominant factor contributing to total output noise. In order for me to verify this and suggest the best possible amplifier can you please let me know the BW requirement for your system, the input capacitance of the photodiode you plan to use and finally the desired transimpedance gain.

-Samir

Hi Samir,

Thank you for responding.

I was mistaken, I am not concerned with noise below 10 kHz. You are right, with the 1 MOhm resistor, a FET-input op amp makes sense here.

The photodiode capacitance with the bias I am using is 24 pF.  The photodiode shunt resistance is 1 MOhm.  Desired bandwidth is 1.5 MHz.  DC is removed by integrator upstream from the transimpedance amplifier (TIA).  I saw the THS4631 and used that.  That, being a FET input op-amp, greatly reduced the integrated noise from 10 kHz to 1.5 MHz, I know get about 660 uVRMS over that BW.

What's the best way to optimize the SNR based on the Rf value?  Could I simply take the partial derivative of the SNR with respect to Rf , set the derivative to zero and sove for Rf?

Ken

Ken,

The OPA657 should work well for your application. It does have more bandwidth than you need but has lower noise than the OPA656 and THS4631.

Its a little complicated to determine SNR vs Rf value because most of the noise in the application will come from the high-frequency peaking due to noise gain. Please see this article for an explanation of what I mean by noise gain peaking

I have some theoretical calculators to determine SNR and for the case of JFET/CMOS input amps., the higher the Rf the better the SNR....so theoretically you can take this to the limit of achieving the maximum possible gain while maintaining your bandwidth requirements. The OPA657 with its higher than needed gain bandwidth product (GBP) will be beneficial here. Remember however that as the feedback resistance increases the value of feedback capacitance needed to stabilize the TIA starts to get very small. So this sets a limit as well. The link below is a blog article I wrote about stabilizing transimpedance amplifiers and also contains a link to a calculator to determine the value of feedback capacitance needed to stabilize the TIA. Looks like around 1.5MOhm of feedback resistance will require a 0.1pF feedback capacitor which is probably what you are going to get from the parasitics on the PCB and the Rf itself, so no extra physical Cf is needed in the circuit.

Let me know if you need further help Ken.

-Samir

Thanks Samir,
The OPA657 is good, but I would like the TIA to run from +/-15V.

Ken

Ken,
The highest bandwidth FET amplifier we have is the THS4631 with 210MHz of GBW. Its going to be challenging to find much choice with high voltage supplies and large BW.
-Samir