Because of the holidays, TI E2E™ design support forum responses will be delayed from Dec. 25 through Jan. 2. Thank you for your patience.

  • State Resolved
  • Locked Locked
  • Replies 11 replies
  • Answers 2 answers
  • Subscribers 48 subscribers
  • Views 585 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

AMC1311: Output of isolation amplifier, noise in low frequency range

Evgeniy Zubkov
Prodigy 20 points
Part Number: AMC1311
Other Parts Discussed in Thread: INA2143, , AMC1350, AMC3330

Dear Sir and Madam,

I am currently designing an isolated amplifier circuit, that is capable of measuring a high voltage signals up to 1500 V with at least of 20 kHz bandwidth. The high voltage is divided by factor of 1001:1 (voltage divider with 1Meg and 1 kOhm) and resulting signal get feed to the input of the AMC1311B. The differential output signal of AMC1311B get feed to one half of INA2143 amplifier, to get 10x amplified single side output signal, which I have finally connected to scope input, using the shielded BNC cable. After the prototype board was build, I have measured the noise performance of the whole circuit. I have found in the output signal the high frequency broadband noise, starting from about 50 kHz, with a broad peak at about 300 kHz and a high sharp peak at about 623 kHz, please see the attached picture. The signal output is bandwidth limited to 150 kHz due to INA2143 bandwidth.

FFT spectrum scope

The blue signal is the output from INA2143, the red signal is the +5 Volt VDD2 supply of AMC1311, measured at 100 nF blocking ceramic capacitor, both signals were AC-coupled. The VDD1 was supplied from 3 alkaline batteries (4,6V), because I was thinking the VDD1 supply was producing additional noise (I have disconnected the VDD1 DC/DC converter and unpowered it completely). It's a pity, that the AMC1311 datasheet does not show the whole noise picture up to 1 MHz in the Fig. 7-23. To support the circuit designer, the datasheet should be improved, in order to show the whole noise picture.

To get reed of the high frequency noise above 100 kHz I will use a low pass RC-filter, with a filtering frequency, starting from 20 kHz.

After I have changed the scope resolution, I have found some strange spikes in the INA2143 output signal. The next FFT spectrum shows two peaks at 2,64 kHz and 3,41 kHz, which are present at lower frequencies and limiting the AMC1311B noise performance. The peaks also can be seen in previous FFT spectrum.

FFT spectra

The input signals are the same, as described in FFT-spectra above.

Are the peaks at 2,64 kHz and 3,41 kHz typical for the AMC1311B output signal?

In the Fig. 7-23 of datasheet, the both peaks can be hardly recognized, because of the logarithmic scaling of frequency axis and not enough averaging/signal accumulation time.

Are the same noise features also available in ISO224B, AMC1350 and AMC3330 isolation amplifiers?

If the peaks are typical, how one can reduce them?

Thank you in advance,

Dr. Evgeniy Zubkov

  • 0
    TI__Expert 5765 points

    Hi Dr. Evgeniy Zubkov,

    Thank you for describing your results. There is a chopper on the AMC1311 you are seeing in your output. Our recommendation would be to add a low pass filter to mitigate this. 

    Best regards,

    Eva

  • 0 in reply to Eva Mason
    Prodigy 20 points

    Hi Eva,

    As I mentioned before, the designed bandwidth of the whole signal path must be at least 20 kHz. If I use a low pass filter with 20 kHz cut-off, will it remove the signal artifacts at 2,64 kHz 3,41 kHz as well?

    Best regards,

    Evgeniy 

  • 0 in reply to Evgeniy Zubkov
    TI__Expert 5765 points

    Hi Evgeniy,

    No, if you're still seeing these signals below the cutoff frequency the low pass filter won't remove these. I'm moving this question over to someone who can assist with the INA2143.

    Best regards,

    Eva

  • 0 in reply to Eva Mason
    TI__Genius 12415 points

    Hi Dr. Zubkov,

    I don't think there's any reason these spikes would come from the INA2143, it is a linear amplifier without any switching or clock generation.  The only issue may be stability; can you post the circuit you used so we can check the stability?  The main thing we would look for there is any significant output capacitance.

    Also, if you could check the spectrum of the INA2143 by removing the AMC1311 from the circuit, that would help us determine where to focus.

    Regards,
    Mike

  • 0 in reply to Evgeniy Zubkov
    TI__Mastermind 23210 points

    Hello Dr. Evgeniy Zubkov, 

    Do you have a schematic you can share for analysis? Have you tested the INA2143 half of the circuit stand alone in order to observe if the signal artifacts are still present? 

    Best Regards, 

    Chris Featherstone

  • 0 in reply to Chris Featherstone
    Prodigy 20 points

    Hi everyone,

    below you will find the used circuit

    original circuit

    I have tried to insert the RC-filter with 20 kHz bandwidth (2x40,2 Ohm and 100 nF) between AMC1311 and INA2143. The resulting FFT-spectra you can see below. 

    Rc filtered

    The 623 kHz peak is now much much lower. The low frequency artifacts are still there.

    Then I have tried to put another RC-filter (now second order), also at 20 kHz cut-off.

    The resulting FFT-spectra you can see below.

    2 times filtered signal

    The low frequency artifacts are still there. What is strange, that the 623 kHz peak was not reduced so much, as I have expected. Probably some EMI-stuff...

    Below you can find the lower frequency FFT-spectra with artifacts.

    2 times filtered signal

    The peaks are at (from left to right) 616 Hz, 1.24 kHz, 1.86 kHz, 2.48 kHz, 3.1 kHz, 3.45 kHz, 4.96 kHz. These peaks are now higher, than 623 kHz chopper peak.

    The output of the INA2143 is connected to BNC-cable, 1 m long. I have checked the BNC-cable specs, it has 101 pF/m. I have also checked the input specs of the scope1 MΩ ±1%, in parallel with 14 pF ±1 pF. May be the INA2143 has with total capacity of 115 pF an issue.

  • 0 in reply to Evgeniy Zubkov
    TI__Mastermind 23210 points

    Hello Dr. Evgeniy Zubkov,

    I will take a look over the schematic and the information provided and respond by the end of business friday. 

    Best Regards, 

    Chris Featherstone

  • 0 in reply to Michael Hartshorne
    Prodigy 20 points

    Hi Michael,

    I have removed AMC1311 from the PCB and measured the FFT-spectra of the INA2143UA output signal. The result you can see below.

    INA2143 output spectrum

    Best regards,

    Evgeniy

  • 0 in reply to Evgeniy Zubkov
    TI__Genius 12415 points

    Hi Evgeniy,

    OK, thanks.  I'm going to connect the AMC team again for support on this.

    Regards,
    Mike

  • 0 in reply to Michael Hartshorne
    TI__Expert 5765 points

    Hi Evgeniy,

    Thank you for getting these FFTs. These are probably low frequency harmonics from the AMC1311's chopper frequency. You could try adding a capacitor in parallel with your 1kOhm sensing resistor. The capacitor should be about 12pF. Please let me know if this helps!

    Best regards,

    Eva

  • 0 in reply to Eva Mason
    Prodigy 20 points

    Hi Eva,

    I have added a 12 pF capacitor in parallel to 1 kOhm sensing resistor. In the following image, you can find the FFT spectrum of the output noise of the current setup. I have noticed, just by the eye, that the low frequency peaks has moved to the lower frequency range. The INA2143 output noise level, measured just on the scope trace (peak-to-peak value), was reduced by about 10% and is 7,2 mV now.

    with 12 pF capacitor

    The bandwidth of the current circuit is about 8 kHz and is definitely too low. I will stay with the second order 20 kHz RC-filter solution on the output of the AMC1311, it really helps a lot. Thank you for support and useful discussion.