• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 46 subscribers
  • Views 260 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.

ADS131E04: 24bit resolution not fully usable

Simon Blum
Prodigy 10 points
Part Number: ADS131E04
Other Parts Discussed in Thread: ADS131E08

We are using a ADS131E04 in 24bit 16ksps mode. The values we get seem to be valid on all channels apart from the occasional extreme outlier, but we can deal with those. But if we look at the 24bit integer values in decimal form we see that ALL the values are divisible by 10.

We tried deceasing the sample rate to 8ksps and now the decimal form of the values NEVER ends in digits 2 or 7. After further deceasing the sample rate to 4ksps values ending in 0, 2, 5 or 7 only appear about half as often as values ending in 1, 3, 4, 6, 8 or 9.

We see this behaviour on all channels and on all channel multiplexer settings. In 16bit mode we were not able to see similar strange behaviour. We currently have two boards with the ADS131E04 in our lab and we witnessed this behaviour without fail on both of them every time we checked.

We also were unable to find anything in the datasheet that would explain such behaviour. It seems especially strange to us that such patterns appear in decimal representation of data that comes from a binary device.

As our application requires at least 22 bit resolution at 16ksps this is a massive problem for us and at the moment wo have run out things to try to resolve this issue and we hope someone here can help us.

  • +1
    TI__Guru 63216 points

    Hi Simon,

    Thank you for your post and welcome to our forum.

    Most 24-bit ADCs will be limited in resolution by thermal noise or other extrinsic noise sources. Their LSB size (voltage weight per code) is almost always much smaller than this noise level, sometimes by more than one order of magnitude. As you can see in Tables 1 and 2, the effective number of bits (ENOB) is only about 17 bits at best when using a data rate of 16 kSPS. 

    With a 2.4-V reference and Gain = 1, the ADS131E08 has a dynamic range of about 103 dB at 16 kSPS, which equates to about 12 uVrms or 34uVpp. The LSB size of the ADS131E08 is only 286 nV.

    I'm not sure exactly how to explain why the least significant digits of the decimal codes exhibits these patterns that you've observed, but I believe it's simply an artifact of the digital decimation filters and how they calculate the 24-bit outputs for different oversampling ratios (OSRs), which in turn produce the different data rates. This is not an issue which is specific to this product or the the devices on your boards. However, it's really irrelevant given that the missing codes would be hidden by the ADC noise anyway, let alone the additional extrinsic noise that may enter from the signal source, supplies, or elsewhere in the system.

    Best regards,

    Ryan

  • 0 in reply to Ryan Andrews
    Prodigy 10 points

    Ok, but please consider to clarify this in the next revision of the datasheet. If I hadn't tested two boards those filter artifacts would have made me believe the device is broken.

  • 0 in reply to Simon Blum
    TI__Guru 63216 points

    Hi Simon,

    Thank you for your feedback. We will consider how to better explain this to customers in the future as it applies to many of our products.

    Regards,

    Ryan