• State
  • Locked Locked
  • Replies 1 reply
  • Subscribers 45 subscribers
  • Views 222 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.

ADS131M03: Resolution OSR

Chris Willis
Intellectual 290 points
Part Number: ADS131M03

I have a question regarding the clock frequency of the ADS131:

We will use ‘High resolution mode’ and the maximum oversampling rate (16384) because that gives the best quality results.

We only need to collect one reading per second.

Is there any difference in terms of accuracy or noise between :

  1. Using an 8MHz clock, getting 250 samples per second and averaging those 250 samples
  2. Using a 0.3MHz clock, getting 9 samples per second and averaging those 9 samples
  3. (or anything in between)

Obviously with larger number of samples we can do a more sophisticated averaging technique in our firmware such as median filtering, but assuming we just do a simple average – do we gain any benefit in terms of noise or accuracy by running at the higher clock frequency? 

I know we can potentially get more effective bits of resolution by averaging more samples but anything over 16 bits is more than adequate for us anyway.

Best regards

  • 0
    TI__Mastermind 38705 points

    Hi Chris,

    I would update my comments to avoid any confusion to you, please let me know if you have any further questions:

    1. Scaling the clock doesn’t change the ADC noise because the oversampling ratio is the same in both cases.
    2. It might be true that a higher-frequency clock could cause clocking issues / artifacts that might lead to other issues, though probably not at these frequencies e.g. 8 MHz. But purely in terms of ADC + PGA noise, slowing down the clock does not provide any benefit if the customer uses the same OSR in both cases (and the same gain of course). 
    3. Also, it is theoretically possible to lower the noise by averaging the results of multiple samples (this is fundamentally how a delta-sigma ADC operates). So if the customer takes 250 samples and averages them all together, the noise would decrease by a factor of √N, where N = # of samples. In this case, √250 = ~16. Note that this only works if the noise has a purely Gaussian distribution. 
    4. Therefore, if the noise is the same in both circumstances the customer identified (#1 and #2), then #1 is actually the better choice because they can average more samples (250 vs 9) and theoretically lower the noise more.

    Thanks&regards,

    Dale