• State Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 2 answers
  • Subscribers 47 subscribers
  • Views 804 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.

Original question:

ADS1261: Offset and Gain Calibration

ADS1261: Full-Scale Calibration with RefOut

Drew Nakamura
Prodigy 130 points

Part Number: ADS1261

I mentioned in a different thread (link for posterity) that it may be difficult to do a gain calibration since I'd have to find a way to apply the full-scale input voltage to the ADC, but the original thread made me realize that it may be easier than I thought.

Since I'm planning on using the RefOut voltage to bias my thermistor network, I should be able to apply the positive full scale input voltage needed for calibration if I short across RBIAS right?

Thanks,
Drew

  • 0
    TI__Guru 58971 points

    Hi Drew Nakamura,

    When your input voltage (VIN) is equal to your reference voltage (VREF), you should get a positive full-scale code because VIN / VREF = 1. So what happens if your signal chain has a positive gain error, such that VIN is actually VIN + 0.01%? Now your input signal will be larger than VREF, but this will still output a positive full-scale code i.e. there won't be any difference between this case and the ideal case. So how can you reliably calibrate the system? I agree this should in principle work if you had a negative gain error.

    -Bryan

  • 0 in reply to Bryan Lizon86
    Prodigy 130 points

    Hi Bryan,

    The situation you're describing is for when the system is actually in use, correct? In practice, we'd always be measuring across a divided down version of RefOut, <1.2V, so I don't think this would cause any problems for us.

    -Drew

  • 0 in reply to Drew Nakamura
    TI__Guru 58971 points

    Hi Drew Nakamura,

    I assumed you were going to tie REFOUT to AIN0 and then ground AIN1 for example, which provides a differential input signal of 2.5V. This would be your calibration signal. Then, you would use this full-scale voltage as the source to eliminate the ADC+PGA gain error. If so, then as mentioned this would not account for a positive gain error, so you would not be able to accurately calibrate your system

    If I misunderstood, let me know what you meant by "shorting across RBIAS"

    -Bryan

  • 0 in reply to Bryan Lizon86
    Prodigy 130 points

    Hi Bryan,

    Thanks for clarification. What you've stated aligns with what I was envisioning. Would the following procedure successfully account for positive gain error as well?

    1. Short across RBIAS as before (REFOUT on AIN0 and AIN1 to ground)
    2. Check to see if the output codes are maxed out
    3. If they are, write 0.95 gain factor to FSCAL[2:0] register
    4. Send GANCAL command 

    -Drew

  • 0 in reply to Drew Nakamura
    TI__Guru 58971 points

    Hi Drew Nakamura,

    Yes, this technically should work. However, there is another issue with this method: when you have a ground-referenced input, which would be the case for the calibration, you need to disable the PGA. Therefore, you wouldn't then be able to enable the PGA for the actual thermistor measurements. So you would not get much benefit from this method unless you kept the PGA bypassed at all times

    One other question: I looked back through your spreadsheet and I saw that the thermistor error is ~3%. And in a previous thread you mentioned that you were not going to calibrate the sensor. So why would you be concerned with the ADC gain error (0.05% typical) if the sensor error is so much larger? I don't see this is a meaningful effort

    -Bryan

  • 0 in reply to Bryan Lizon86
    Prodigy 130 points

    Hi Bryan,

    Does the PGA need to be disabled because the PGA output cannot be less than AVSS + 0.3V = 0.3V? If I wanted to avoid this problem, would it make more sense to try and apply a full-scale positive voltage across the inputs that's offset from ground (e.g. 3.5V and 1V)?

    Just to make sure I'm understanding correctly, calibrating the sensor would involve sweeping the system over a temperature range and recording a voltage/resistance vs temperature curve to use as either a lookup table or for equation fitting, right? I can't remember exactly what I thought calibration meant before, but we do want to calibrate the sensor to remove that initial 3% error as much as we can.

    -Drew

  • +1 in reply to Drew Nakamura
    TI__Guru 58971 points

    Hi Drew Nakamura,

    Drew Nakamura said:
    Does the PGA need to be disabled because the PGA output cannot be less than AVSS + 0.3V = 0.3V? If I wanted to avoid this problem, would it make more sense to try and apply a full-scale positive voltage across the inputs that's offset from ground (e.g. 3.5V and 1V)?

    Yes, this is correct. But how would you do this practically? The ADC REFOUT voltage is always measured with respect to AVSS, so how do you shift this voltage off of AVSS / GND?

    Drew Nakamura said:
    Just to make sure I'm understanding correctly, calibrating the sensor would involve sweeping the system over a temperature range and recording a voltage/resistance vs temperature curve to use as either a lookup table or for equation fitting, right? I can't remember exactly what I thought calibration meant before, but we do want to calibrate the sensor to remove that initial 3% error as much as we can.

    To some degree, yes. You would need to place the thermistor in a thermal chamber so you can be sure you are measuring an accurate temperature. Then you would take measurements at the low and high end of your measurement range while the system is at room temperature. You could then correlate this to your thermistor equation or look-up table that describes its behavior. Assuming the system error is linear, this should remove most of the error to the level of the system noise. You would have to repeat this process across system temperature if you also wanted to calibrate the over-temperature error

    We have a TI Precision Labs video that describes how this works for RTDs, but the same general principles can be applied to thermistors.

    -Bryan

  • 0 in reply to Bryan Lizon86
    Prodigy 130 points

    Hi Bryan,

    Thank you for the calibration resource. I will pass the information along to our team.

    Realistically I don't think it's feasible to apply an offset full-scale voltage. I was mostly asking to make sure I understand why the ground-referenced approach wouldn't work.

    I believe I'm good on this topic for now. Thank you.

    -Drew