• State TI Thinks Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 2 answers
  • Subscribers 46 subscribers
  • Views 994 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

multiplexed unipolar transimpedance amp

TC55946
Prodigy 120 points
Other Parts Discussed in Thread: ADS1158

I would like to use the ADS1158 16-channel multiplexed ADC IC as a high precision (i.e. ideally full 16 bits) for measuring 16 variable resistances using a single-ended supply. Ideally I would like to use a transimpedance amp for measuring the resistance to maintain a constant voltage across the resistance.

The datasheets give a couple examples of circuits on page 42 & 43: http://www.ti.com/lit/ds/symlink/ads1158.pdf

One is for resistance measurement using a bridge, but there are two limitations for my applicatoin: (a) it is a bipolar supply, and (b) it only allows measurement of 8 resistances (since two inputs are used per bridge).

The second circuit is for a unipolar supply, but is for measuring voltages only and seems to use an instrumentation amplified type of configuration between MUXout and ADCin.

Can anyone suggest the best way to modify the second circuit using a transimpedance amp on the MUXout and yet use as much of the ADC's range as possible?

Thanks

  • 0
    Prodigy 120 points
    To refine or perhaps add to my question, and considering the use of a uipolar supply...

    A) Is it possible to use a transimpedance amplifier to measure 16 resistances using this chip? (i.e. something like the following circuit, but with the ADS1158's MUX in between the diode and the op-amp)?

    B) Other than some type of voltage divider (which I am trying to avoid since I would like to maintain a fixed voltage across the resistances being measured), is there any way to use the ADS1158 to measure 16 different resistances (i.e. not using the Wheatstone bridge measurement method that uses two inputs per resistor being measured and cuts down the number of resistances that can be measured to 8)?

    A) Is there any difference between the AINCOM and MUXOUTN signals if AINCOM is set to a fixed voltage (such as GND) that is always below AIN<x>?
  • 0 in reply to TC55946
    Prodigy 120 points
    forgot to post this circuit link for (A): circuits.datasheetdir.com/.../OPA380-circuits.jpg
  • 0 in reply to TC55946
    Prodigy 120 points
    To follow up on the 3rd of the above 3 questions, and sorry for the confusing lettering since that should read question (C) and not (A), I am not clear from figure 22 in the datasheet exactly what MUXOUTN corresponds with or is connected to. If I am reading the diagram correctly it sort of seems that the MUXOUTN is connected to AINCOM regardless of which external input is selected, but then has an internal option to be connected to something other than AINCOM if measuring the internal offset monitor. Sorry I find the diagram confusing. Hopefully someone can please clarify.
  • 0
    TI__Mastermind 44720 points

    Hi TC,

    What range of resistances do you need to measure?
    This will certainly affect the way you can configure the circuit.

    I had to think about the transimpedance amplifier implementation a bit. Originally, I thought about placing the resistor you were trying to measure in the feedback loop of the amplifier, but this didn't make much sense because this would still require two inputs per resistor. However, if you used the following configuration with a fixed gain resistor and a bias voltage, you would be able to measure sixteen resistors...

      Here are a few pros and cons to this configuration...

      PROS:

      • This configuration provides the 16 resistors inputs and only requires a single transimpedance amplifier (TIA).
         
      • Vbias sets the voltage across the resistor you're trying to measure...this creates a current through both the resistor you're measuring and the gain resistor, which in turn produces the voltage sampled by the ADC. If Vbias changes, it will affect the voltage output of the TIA. However, by measuring the differential signal (the output of the TIA versus the bias voltage), the differential signal will not be affected by changes in Vbias.
         

      CONS:

      • This configuration includes the MUX's on-resistance in the measurement. This will cause an offset error in the measurement. While the initial offset error can be removed by calibration, there will still be a residual offset error that changes over temperature.
      • The above schematic implies a perfect ground, but in reality, ground may be noisy and it will have an impedance. Therefore, ground plane noise and ground plane impedance will also be included in the measurement results.

        

      Hopefully, that answered most of your questions... Regarding MUXOUTN, that is another MUX output pin (both the MUX output and ADC inputs are differential), but in the case of the above circuit, you would not use MUXOUTN.


      Best Regards,
      Chris

    • 0 in reply to Christopher Hall
      Prodigy 120 points

      Thanks for the reply. This is how I had it configured. Resistance will be on the order of 10k+, so multiplexor resistance should not affect that much. The remaining question is how to optimize the circuit and the other inputs on the ADS1158 (VREFP, VREFN) in conjunction with Vbias such that I can use the entire range of the ADC.

      There is a suggestion at the bottom of this post that encourages using 2.5V as VREFP, but it is not quite the same configuration since that post also suggests using 2.5v as ADCINN:

      http://e2e.ti.com/support/data_converters/precision_data_converters/f/73/t/388596

      Is is possible with the transimpedance circuit you posted to use the full ADC range by adjusting Vbias and VREFP & VREFN, and if so, what would you suggest?

      Thanks

    • 0 in reply to TC55946
      TI__Mastermind 44720 points

      Hi TC,

      The only way I see of utilizing the full ADC input range is to make Vbias as close to 0V as possible (because the TIA output voltage will always be greater than Vbias).

      Therefore, you may need to consider using a negative "common voltage" and connect all your resistors to this negative potential, instead of ground. This way you can still generate a current across the test resistors. (Note that the ADC's reference voltage could still be 2.5V, referenced to ground.)

      Also, the TIA output will not be linear near its supply rails, so you might also be able to take advantage of the negative supply voltage so that the TIA provides a wider linear range on the low-end (the high-end would still be limited unless you provided a supply voltage greater than 5V).

      To really optimize/match your TIA circuit to the ADC's input range, you'll need to know the precise range of resistances that you plan to measure. From there you can adjust the test resistor "common" voltage, Vbias, and the TIA gain to provide the most sensitivity across your particular resistance range.


      Best Regards,
      Chris

    • 0 in reply to Christopher Hall
      Prodigy 120 points
      Regarding the diagram you posted, just wanted to check if the ADCINN and ADCINP pins should be reversed since with the resistor being measured connected to ground (or lower), the TIA's output will always be above Vbias, won't it?