This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

OPA657: Power Spectral Density of Transimpedance Amplifier

Part Number: OPA657
Other Parts Discussed in Thread: LM7171, , LMH6622, OPA838

Hi

I have designed a TIA with a gain of 500k at a frequency of 451Khz using OPA657. I have connected a voltage amplifier stage using LM7171 to further increase a gain of 10. I wanted to analyze the noise power spectral density of the whole circuit. But my TINA simulation power spectral density and the results obtained through PCB are not matched. In my simulation the power spectral density increases upto 731Khz and after that it begins to fall. But for my PCB when I took my data using an oscilloscope I found the power spectral density   increases only upto 200KHz after that it begins to fall. I have decreased my feedback resistor from 500k to 100k of TIA but the power spectral density   in the PCB is still the same.

I have used Judson Photo diode having 1.2nV terminal capacitance.  I have taken my PCB observation by placing the photo diode in a closed box where no light falls on it.

Please help me to solve this problem.

I have attached my Tina file and measured results that I have processed in Excel.

 7455.Noise_Power_spectral.TSC

  • Oh I see now you updated with spectrum analyzer plot - yea that is definitely peaking to a lower peak at lower F than the simulation. Here is your plot with the markers for the OPA657 ouput noise - take 10X for your measured data

    Those 2nd 2 stages must be bandlimiting - they simulate fine of course. Not sure why you are using a +/-15V part there, might look at the LMH6622 

  • Hi Michael

    Thanks for your quick reply.

    I have already tested by removing  both voltage amplifier stages from PCB and I am getting 10 times lower noise but my peak is always present at around 200KHz and after that noise spectral density  decreases. Any further suggestions?

    Thanks Again

  • Well I would get more detailed on what your source impedance looks like including wiring parasitics - you say the diode is in a box, is the amplifier? 

  • Hi Michael

    I have not connected any source for noise analysis. I have just giving the biasing supply  and placing the whole circuit ( My amplifier) inside a metal box. No light enters the box.

    Thank you

  • Well you need the diode capacitance in place for the expected noise profile

  • Thank you, You mean  I should remove the photodiode from the PCB and replace it with a capacitance of 1.2nF as I did in my simulation?

  • I would try that, if it pops into place then work on your diode capacitance model

  • Thank you. I will try that and let you know.

  • Hi Parveen,

    what noise figure do you measure with a simple resistor in the metall box?

    By the way, have you connected the metal box to signal ground of your circuit?

    Kai

  • Hi Michael

    As per your suggestion I have changed my source impedance. That means I have connected my photo diode to zero bias. After that my power spectral curve is approximately the same as the simulated curve. I first tested only the TIA stage and it works well. After that when I connected the second stage (voltage amplifier), the oscillations developed in the noise (when I zoomed into the noise on an oscilloscope, it shows oscillation). I have tested the individual stage phase margin using TINA ,and they are good. But when I tested the whole circuit on PCB the noise shows oscillations . I have no idea how to check the stability of the whole circuit.

  • Hi Kai,

    can you further explain your first question?

    Yes, I have connected the metal box to signal ground of my circuit.

  • Hello Parveen, 

    I think you are saying the OPA657 stage seems to match simulation now - does that mean you replaced the detector with a 1.2nF cap (which seems very high) or that removing the detector bias might have made it look like 1.2nF?

    So that stage works, is this still what those 2nd two stages look like. If the sims look good but the board oscillates we are missing something in the sim files (board parastics, op amp model - this one is really old, might not have all the I/O impedance correct in the model). Also, what is your output load - 50ohm series to 50ohm spectrum analyzer is what we would expect through a 50ohm cable. 

  • Hi Michael,

    I have replaced the detector bias on pcb with zero. This means I have connected anode of my photo diode to ground potential. I have connected 50 ohm series to 50ohm spectrum analyzer. But as I connected voltage amplifier stage of gain 10, its behaviour changed and now the noise has oscillations. I have not connected the unity gain voltage amplifier stage (Stage 3). So what should I do next.

  • Normally, you do not say the noise has oscillations - looking into a spectrum analyzer not sure what the means. A time domain oscillation will show up on a spectrum analyzer as a spike - have you tried looking at the output of the gain of 10 stage on a scope? 

    The circuit looks like it should be stable. Not sure what to tell you, but if it was me I would move away from that very old LM7171 - excessively fast and high noise if your 1st stage is sub 1MHz. 

    1. why +/-15V parts

    2. Why slew enhanced input VFA, increases the noise - do you need the slew rate

    3. If you need a gain of 10 in that first stage, look at the new OPA838.- be careful with feedback C in that stage if using decomp device. Can bandlimit in the last stage if needed. 

    The other advantage to moving to parts done in the last 10years is much better modeling. not sure how good the LM7171 model is and I doubt anyone wants to work on that question.

  • incidentally, if you are seeing spikes in your spectrum, try turning the power off on the amplifiers - if they stay there, you are picking up something in the ambient. 

  • Dear Michael,

    Thank you for such a comprehensive and detailed reply. What you suggested is beginning to make sense to me. I will incorporate some of your suggestion in my circuit and see how it goes.

    Thanks again for your time.