• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 46 subscribers
  • Views 340 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:

ADS131M08: Example code

ADS131M04: Max INL estimation

Shota Mago
Expert 8310 points
Part Number: ADS131M04


Hi Team,

One of my customer considers to use ADS131M04 and provided some questions as follows.

  1. ADS131M04's INL only shows typ(6ppm) value.
    Is assuming 6ppm typical value as 1σ and multiplying 6 feasible to estimate max INL?
  2. According to datasheet, Vrms is around 2.39uV in the case of OSR=8192. 
    When using ADS131M04 as single-ended input, approximate Noise Free Bit can be calculated as below.
    ln(1.2V/6*2.39uV)/ln2=16.35bit
    Is this understanding correct?
  3. After offset/gain calibration, how can we estimate worst total error considering INL/NFB?
    When assuming NFB=16.35bit and max INL=36ppm, worst total error will be √(NFB^2+INL^2)=√(12ppm^2+36ppm^2)=37.9ppm
    Is this understanding correct?
    Or we should not mix up DC spec and AC spec?

I'm looking forward to hearing back from you.

Best regards,

Shota Mago

  • 0
    TI__Guru* 98255 points

    Mago-san,


    I'll respond to your questions in the order they were asked.

    1. As you mentioned, the INL for the ADS131M04 is listed as 6ppm typical value. It would be reasonable to use 6σ for the maximum INL, which does push the value to 36ppm. I would note that on many ADCs, the INL is determined by the charge injection from the switches during the input sampling. Because of this, most INL is often constant (or very similar) from device to device. However, the INL may vary depending on gain and common-mode signal. For that reason, I would be conservative with the INL specification and still use the 6σ value.

    2. Normally, we use the full-scale range of the ADC for calculating for the noise-free bits, which would include ±1.2V (or really 2.4V). However, your calculation is basically correct. There's no difference between 16.35 bits noise free on 23 bits and 17.35 bits noise free on 24 bits with the change in range.

    3. For the lower speed ADCs (this would often be included as lower speed), the INL and the noise are often both considered as DC specs. The root-mean-square method of calculating the maximum error is reasonable, as we use this for the total unadjusted error (TUE). However, the TUE usually includes the offset and gain error, which often dominates the specification.


    Joseph Wu

  • 0 in reply to Joseph Wu
    TI__Expert 8310 points

    Hi Joseph-san

    Thanks for your kind explanation.

    Let me check one thing in terms of 2nd question.

    Although ADS131M04's full scale range is ±1.2V(or 2.4V) as it has differential input, in case of single-ended input usage, full scale range is just 1.2V.

    So if the customer use only single end input, actual NFB will not be 17.35bit but 16.35bit.

    Is this correct?

    Best regards,

    Shoat Mago 

  • 0 in reply to Shota Mago
    TI__Guru* 98255 points

    Mago-san,

    The noise free bits calculation depends on the noise and its relation to the total range of operation. If the input range is single ended, then your calculation is correct. If the input range used is bipolar, then the same noise value would be on the full range, which would add the extra bit to 17.35 bits.

    Because the customer uses only a single-ended input, the actual NFB would be 16.35 bits.


    Joseph Wu