Filter design / signal conditioning for a full-bridge sensor amplifier

Discussion in 'Analog & Mixed-Signal Design' started by Juha Haaja, Oct 3, 2016.

  1. Juha Haaja

    Thread Starter New Member

    Oct 3, 2016
    4
    0
    Hello,

    I am trying to build an amplifier circuit to use with a full-bridge extensometer. I have managed to get the amplifier to work but am now facing a problem with EMC issues. I need to use the extensometer in a AC magnetic field which has a frequency of approximately 230 kHz. The magnetic field couples to the sensor nicely even though the cable is shielded and the shield is grounded.

    I have tried low pass filters on the inputs to the differential amplifier with little success. The amplifier I am using is Texas Instruments INA122PA. I am using high amplification (~500) to reduce the effect of the noise from the low quality ADC used to capture the signal to a computer.

    I'm not very experienced in this stuff so any help would be appreciated.
     
  2. ci139

    Member

    Jul 11, 2016
    341
    38
    the instrumentation amps are usually precise but slow - your's uses low power - and you have a huge gain (to be followed precisely ? by a single op amp)
    • usually you need your output to follow the input fast -- requires -- fast amplifier -- usually requires moderate to high power to drive the output "reliably"
    • the low noise amplifier does not mean low power
    the schematic shows capacitors from inputs of your gain 500 to "common rail" or "signal ground" what if they are not homogeneous throughout the DC voltage scale and add extra surprises at frequency scale
    so you take it as a part of setup ? substract it from your signal at some point - if that 230kHz may vary a lot i don't want to be in your shoes

    (PS! i have little to null practice in this field - analog signal processing - though i might've read some docs)
     
  3. Juha Haaja

    Thread Starter New Member

    Oct 3, 2016
    4
    0
    Thanks for the reply. I could have explained my application a bit better. Basically the problem is that I have a differential voltage signal that goes from -12 mV to approx +18 mV and I have to amplify this to around 10 V. The reason for such a high amplification is that the ready made ADC's that I have found in general have a noise of +/- 30 mV.

    I am actually measuring strain using an extensometer and in that sense I do not need a super fast following of the inputs. The problem is that I have to do this strain measurement in an environment that has the 230 kHz magnetic field present to mess with things. At the moment really the problem is to reliably filter out the disturbances caused by this magnetic field. These disturbances are so high that they completely mask the differential signal coming to the amplifier. By the addition of those 10 uF capacitors I was able to lower the disturbance caused by the magnetic field 40 fold.
     
  4. ci139

    Member

    Jul 11, 2016
    341
    38
    how about you run no signal extra pair inside the same shield to capture the noise e.c.
    there are also quite powerful "digitalized analog" signal processors -- but i don't remember them extracting heavy noise off the faint signal . . .
     
  5. Juha Haaja

    Thread Starter New Member

    Oct 3, 2016
    4
    0
    Now that I have done further testing on the system I have so far the biggest problem is a dc-offset that is caused when the magnetic field turns on. Of course I guess it is somewhat normal that the disturbance represents itself as a dc-offset after the amplifier.
     
  6. ci139

    Member

    Jul 11, 2016
    341
    38
  7. HW-nut

    Member

    May 12, 2016
    45
    11
    You may want to use a syncronious modulator/ demodulator to recover the signal. I have used this approach to recover uV signals from 60 Hz noise.
     
  8. Juha Haaja

    Thread Starter New Member

    Oct 3, 2016
    4
    0
    Im not sure what you mean HW-nut. Care to elaborate a bit?
     
  9. HW-nut

    Member

    May 12, 2016
    45
    11
    You can google synchronous demodulator for more information. But in general, you apply an ac excitation signal to the bridge and demodulate the output sycroniously with the excitation signal. It sounds complex but it is not. I have used this approach to measure signals in the uV range with several volts of 60 Hz noise. Check out the AD630 from analog devices. For best rejection, keep the modulation frequency away from harmonics of any interfering signals. Linear technology also has some application notes showing examples using analog switches.
     
Loading...