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.

OPA347: BandPass Filter Output non-linear

Part Number: OPA347
Other Parts Discussed in Thread: OPA4347

I have designed and simulated a low-voltage sensor acquisition and amplifier circuit to amplify low voltage differential input signals to give a DC output voltage.

After the In-Amp I have a Sallen-Key bandpass filter that gives me an output voltage that decreases with input frequency but the simulated results do not do this and are almost linear, the output voltage difference at opposite ends of the bandpass filter are about 30mV which is fine.

The passive High-Pass filter (before the In Amp) was removed from my dev. board as it reduced the input signal by more than 50%.

Is there any logical reason why the actual results differ so much from the simulated results? Could this be a hardware issue?

Pre-Amp Sim.TSC

  • This is the output I am getting.

  • Former Member
    0 Former Member

    Hi David,

    Thanks for your question.  I do have a follow-up question, if you don't mind me asking.

    In your first sentence, you state "I have designed and simulated a low-voltage sensor acquisition and amplifier circuit to amplify low voltage differential input signals to give a DC output voltage."  However, you are concerned with the output voltage across frequency.

    Can you clarify your desired/expected output signal?  Is the measured output voltage across frequency as desired?

    Just want to make sure I am helping to solve the right problem.

    Regards,
    Daniel

  • Well David, the sim looks great, log scale here. I assume what you are talking about are those notches in the passband? 2 ideas - 

    1. Those might show up if you run a monte carlo with RC tolerances

    2. You might add self resonance L to all your caps and see if they show up in sim then. 

  • Hi and thank you for your replies.

    The output is supposed to be the same across the bandwidth (220Hz) with a fixed input voltage and so the gain is to be 10,000 (100 x 100) from 30Hz to 250Hz as is shown on the simulated frequency response. The response was done with a PicoScope Data Logger and was verified with an oscilloscope. Increase the frequency on my function generator and the output drops significantly even after a few Hz.

    There are two Fleige Notch Filters (50Hz & 60Hz cascaded) that I added after the amplification stage as there was significant 50Hz extrinsic noise (added 60Hz as I was experimenting with notches).

    The actual IC used for the Amplifier was the OPA4347 (4 Op Amps).

  • Former Member
    0 Former Member in reply to Dave

    Hi David,

    So as I understand, and please correct me if I am mistaken, the issue is with the bizarre measured response and not with the sim.  By the way, the INA in the sim is making 30dB of gain instead of 40dB.  I am not sure that was intentional.

    I would recommend debugging the hardware by first identifying whether the issue might be with the development board and its setup or with the measuring equipment setup.  Consider taking progressively more simple measurements until you get something that makes sense.  For example, you have that passive low pass filter at the input and you've taken out the high-pass filter.  Would it be possible to just measure the response of the filter?  The INA input is high impedance.  So, I think you could just check that assuming there is space on the board.

    If that does not work as expected, then there is some aspect of the measurement setup that needs to be adjusted since this is pretty much the most simple AC test you could run.  If the filter response looks correct, then move on to the amplification stage and see if you get the 40dB of gain across the passband.  Also, if the 50Hz and 60Hz noise is not overwhelming, you may consider removing those while you are just running these simple tests in order to eliminate as many variables as possible.  Again, I would first try to identify whether the development board or the test equipment is the source of the problem.

    Let us know how it goes.

    Regards,
    Daniel

  • Hi Daniel,

    The total system voltage gain (In Amp + BP Amp) is 80db (100 x 100). I need to modify the technical writing.

    I am thinking that this is hardware related as I have built another dev. board and I am getting the same thing. All component values have been checked and the tracks to them. I feel that this may be Inductive Reactance as the output voltage reduces as the frequency increases, this is from the output of the In Amp to the BP filter. I will have to do some tests.

    So the In-Amp part is fine and accurate.

    The PCB I made is FR4, 4 layer, 1oz copper.

    The equipment setup seems fine.

    The high-pass filter was removed as it was reducing the input signal by way too much.

    I will let you know how I get on.

    Regards

  • Former Member
    0 Former Member in reply to Dave

    Hi David,

    Understood.  If you want to go looking for parasitic inductance on the board, I would recommend looking at the following things.  You may likely be aware of all this already.  But, it doesn't hurt to point it out:

    1. Check to make sure that the decoupling capacitors for your amplifier are placed close to the supply pin of the op amp and that they go between the op amp and the supply source.  If you have long paths from the supply sources to your supply pins, then those could have inductive effects.  Bypass capacitors will help with this if they are properly placed.
    2. Verify that your grounding setup has at least one solid ground plane to avoid long return current paths.  Any sensitive ground nodes should be closely tied to the rest of GND without needing long trace.
    3. Make sure that any sensitive analog signals are not crammed by digital signals.  You especially don't want your sensitive analog traces running long paths in parallel with noisy digital signals or supplies.

    Let us know if we can offer any further help.

    Regards,
    Daniel

  • Solved. Got hold of a capacitance meter and checked each cap in the circuit. 2 were out of tolerance and another should of been 10nF but was 96nF. I put this value back into the simulation and got the same response as I was getting with the PicoScope. I did not expect this component to thrown the response out by so much as it was on the last amplifier in the chain.

    Anyway it usually turns out to be human error, thank you for your help.

  • Former Member
    0 Former Member in reply to Dave

    Hi David,

    Thanks for letting us know.  Yes, unfortunately these things happen and time is lost.  But, I am happy to hear you were able to isolate and resolve the issue.

    Regards,
    Daniel