• State
  • Locked Locked
  • Replies 7 replies
  • Subscribers 47 subscribers
  • Views 13643 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.

Microvolt Signal Amplification

Shardul Upadhyay
Prodigy 30 points
Other Parts Discussed in Thread: OPA333, OPA2333, INA333, TINA-TI

Hi,

I need to amplify a bipolar signal in the ten's of microvolt range (lets say 50 microVolts) for measuring strain (in a wheatstone bridge configuration) to a unipolar 0-5 V for AD conversion through a 12-bit ADC.

That means I will need a gain of approximately 50 000. I am thinking of having a 2-stage amplification circuit with an instrumentation amplifier (with a reference level of 2.5 V) and then a non-inverting amplifier but I am worried that the signal may be too small to amplify without significant noise, considering a single bit change on the ADC will be equivalent to ~12.2 nV.

Furthermore, would a 3-stage amplification circuit be more ideal or will a 2-stage circuit suffice?

What components are recommended for this situation?

Thank you.

  • 0
    TI__Intellectual 920 points

    Shardul,

    Thank you for your inquiry. What project or end solution is this sensing application for? Could you please provide us with the full expected range of your bipolar input voltage? Do you plan on calibrating this sensing circuit in your post-processing, ADC, uC, etc?

    If you plan on using a full-scale range of 5V in a 12-bit ADC, I am calculating a single bit change to be ~1.22mV.

  • 0 in reply to Michael Mock
    Prodigy 30 points

    Hi Michael,

    The application is for force measurement using strain gauges in a wheatstone bridge configuration, and as I wrote the output is expected to be 50 microvolts bipolar (100 microvolts peak-to-peak).

    I do expect the need to calibrate the output in the microcontroller and that is not a problem as long as the final result is an accurate measurement of the force.

    And yes, you are correct that a single bit change at the ADC will be ~1.22 mV but after the gain it translates to approximately nV changes before the instrumentation amplifier gain, although I am not sure if this matters.

  • 0 in reply to Shardul Upadhyay
    TI__Intellectual 920 points

    Shardul,

    I have attached a proposed a solution that uses the INA333 instrumentation amplifier as well as two OPA333 (OPA2333 for PCB) . One is used to drive the mid-supply voltage and the other is used as the second gain stage in the front-end of the system. Without knowing more of your design constraints (power suppply, specific ADC, etc) I went ahead and designed around a single 5V supply.

    Take a look at the simulation I have included and see if it meets some of your design ideas you have stated. I have simulated the front end inputs as a 100uVpp LF input signal that is centered between the bridges reference voltage (5V). The output is a larger sine wave that ranges between 0-5V. The gain can be adjusted to allow maximum/conservative output range without approaching non-linear regions of the devices. 

    Let me know if you have any questions about the circuit or design or let me know if it meets your needs. Feel free  to change anything in the TINA-TI simulation to adjust circuit parameters (gain, filtering, biasing, etc). Thanks.

    5852.Shardul_Bridge_Sensing_10.4.13.TSC

  • 0 in reply to Michael Mock
    Expert 1095 points

    Hi Michael Mock,

      How is it  approximately 100Hz low pass filter? Can you please explain

  • 0 in reply to Deepak%20murali
    TI__Intellectual 920 points

    Deepak,

    That RC filter arrangement has a common mode and differential mode corner frequency. Using the component values I have chosen (which are just common value R's and C's that can be purchased), the differential-mode corner frequency of the input low pass filter is approximately 100 Hz. It is actually closer to 101 Hz. This front-end amplifier circuit can function, in theory, without the RC input filter, but my thinking was that it would reduce the bandwidth/noise of the differential/common mode signals from Shardul's bridge force sensing input and other external EMI/RFI sources.

    Take a look at these two articles if you have more questions or feel free to ask.

    http://www.analog-eetimes.com/en/signal-chain-basics-53-properly-scaled-filter-components-improve-noise-attenuation.html?cmp_id=71&news_id=222901977

    http://electronicdesign.com/analog/improve-noise-immunity-rtd-ratiometric-measurements

  • 0 in reply to Michael Mock
    Prodigy 30 points

    Thanks for your help Michael. I ordered the parts and am looking forward to testing out how accurate these in-amps can be.

    I was in particular concerned about noise before the in-amp stage as the bridge output is very small so the signal-to-noise ratio might be very small as well, making it difficult to get good output even with filtering. This circuit is not being used in a particularly noisy environment by any means but even common noise might over power the small signal. What is your opinion on that problem?

    Thanks,

    Shardul

  • 0 in reply to Shardul Upadhyay
    TI__Intellectual 920 points

    Shardul,

    I would agree that sensing signals at low amplitudes can be challenging when dealing with a signal to noise ratios, however I think the CMRR of the instrumentation amplifier (100-110dB for INA333) is one of the most important aspects to ensure that the common-mode noise at the input stage is not amplified to the output. The noise of the device should also be noted in the design process. RC filtering on the inputs will be more effective for differential signals so depending on your application or the frequencies you are sensing, it may or may not be effective. With poorly matched R and C components it can even reduce the CMRR. There are many other things to consider such as power supply noise/ripple or your reference voltage noise, capacitive coupling noise with poor layout techniques, etc. I think that building/prototyping the front-end system and testing the performance is a great step in the right direction to confirm/compare with theoretical results.