AD8216 input filter


I'm new to these forums, so I hope I could explain my problem well.

We have a design in which we are trying to measure current from a motor phase, in order to be able to limit the PWM signal. This signal has 0-28V range and its frequency it's 12kHz (the duty could be from 5% to 95%). In order to reject the common mode, we chose the AD8216 amplifier because its specifications were adjusted to our design, and its gain and ofsset could work with the isolated amplifier from the next stage. The LTSpice simulation worked, following the schematic below, but measuring in the real circuit, we see an unexpected voltage input of 2V approximately. This voltage is the differential voltage at input, and it is greater than the opamp maximum ratings. In our calculations, Rshunt value was chosen in order to have 230mV with 14A approximately, so this measure has no sense to us. Should we implement a filter input stage or have we another problem? The layout to the resistor is not perfect, so we will try to wire it directly in order to cast aside the possible parasitic resistance from layout.

The AD8216 is connected in-phase, so as it's indicated in Figure 31 from the datasheet, and we aim to measure the bidirectional current.

Thank you very much in advance.

  • 0
    •  Analog Employees 
    on May 6, 2020 4:37 AM 8 months ago

    Hi mdtb,

    The 2V diff input signal that you see across your 15mohm shunt resistor: Does it cause the AD8216 output to run into the rails and saturate? Is the problem that AD8216 output then takes long to recover from this transient?

    Do you happen to know if it is "out of band" (meaning its above your 12kHz frequency of interest)? If so, I'd think some sort of filtering might help you. If it is in-band, I'd think filtering would affect your measurement accuracy.

    You could possibly use a spectrum analyzer to see what frequency range you are dealing with.



  • Hi Hooman,

    the 2V input causes that the amplifier output goes to 1.01V, which is the offset programmed with the REF voltage. This output hasn't changed significantly with an increase in the intensity that goes through the phase, so I think it's failing somehow, I don't know if it's saturated.
    We'll try to see the frequency of the signal, thank you very much for the advice!

  • This are the laboratory results. We have implemented a filter yet, but without improvement.

    Yellow curve is differential input measured with an isolated probe between the shunt resistor terminals. Green signal is AD8216 output, and pink curve is the FFT of the input signal.

  • 0
    •  Analog Employees 
    on May 6, 2020 8:06 PM 8 months ago in reply to mdtb

    Hi mdtb,

    I can't quite make out the fft of the waveform you attached to tell whether the signal is beyond your 12kHz band of interest or not. Sounds like you implemented a filter but that did not help.

    What filter bandwidth did you design for? I'm assuming a Low Pass filter? Did you place it between the shunt resistor and the AD8216 inputs?

    Any ideas where this spurious 2V differential signal is originating from? With a low value shunt resistor, it would be hard to believe that this signal is "picked up" but instead is a true representation of transient current. Have you thought about adding a differential clamp across the shunt to limit the swing? I'm thinking a couple of schottky / fast diodes that limit the swing to ~ +/-0.3V?



  • Hi Hooman

    my apologies, the previous scopes were erroneous, now we could see the response below.

    The yellow curve is the differential input of the AD8216, and the green one is the output. I think that this is a more logical response, and the problem were two undesired capacitors connected to the motor connector, they were marked as not mounted but they were mounted.

    Although we maintain some problems of gain and noise, we think that we could work with this response. The additional filtering seems to be more perjudicial, so we'll keep it without it.

    Thank you very much for your help.