• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 46 subscribers
  • Views 114 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.

ADS1298: Delta-Sigma delay

Ignasi Tur
Prodigy 10 points
Part Number: ADS1298

Tool/software:

We are currently using an ADS1298 for an EEG application, and we are worried about the delay in the output sampling due to the delta-sigma modulator. How can we estimate it with the configuration registers? Or is it fixed?

For example, if we are measuring at 1kbps, what it would be the sampling delay?

Kind regards,

Ignasi

  • +1
    TI__Guru 64236 points

    Hola Ignasi,

    Thank you for your post.

    The ADS1298 uses a sinc-3 digital decimation filter. This filter is a low-latency filter with linear phase response, meaning the group delay is constant across input frequencies within the filter passband. A step-response to a sinc-3 filter will require 3 output conversions for the data to fully settle. The output latency is not dependent on the decimation ratio configured within the device register settings. Group delay will have an insignificant dependency on OSR and is not necessary to consider as the differences are much less than one output conversion period.

    May I ask why you are concerned with the delay in output sampling? All ADC channels in the ADS1298 will synchronize their output conversions together such that they are truly simultaneous. However, due to the nature of delta-sigma ADCs and their inherent oversampling topology, it is not feasible to pin-point the exact sampling instance at the device input. In most bio-potential measurement applications, such absolute timing is not usually necessary.

    We have a detailed application note about sources of delay in Delta-Sigma ADCs which you might find interesting. Please enjoy!:

    https://www.ti.com/lit/wp/slyy095a/slyy095a.pdf

    Regards,

    Ryan

  • 0 in reply to Ryan Andrews
    Prodigy 10 points

    Hi Ryan,

    thanks for the response.

    the thing is we are trying to Measure the N20 signal, dependent on the stimulation of the medial nerve, and a very important information of this signal is to determine its latency (how long it takes to propagate from the hand to the brain). This being typically at 20ms since the stimulation. 

    Hence, when acquiring the signal at 8kbps, 3 samples is not that significant. However, 3 samples at 1kbps means 3ms delay to take into account in the measurement, if I’m not mistaken. Is it like that? And 3ms more or less it is a significant difference for our case.

    Looking forward for your reply.

    kind regards,

    ignasi

  • 0 in reply to Ignasi Tur
    TI__Guru 64236 points

    Hi Ignasi,

    I understand your point. However, keep in mind that 3 samples is for what would be considered "fully settled" data. This is known as the settling time of the filter. Group delay, on the other hand, is the time is takes for an input-step to propagate to the filter output. This time is approximately one-half the settling time. This is described in Figure 4 and Table 1 of the white paper I shared in my previous post.

    I found another application note by the same author which contains exactly the step-response I wanted to show you (https://www.ti.com/lit/an/sbaa230a/sbaa230a.pdf). Here, you can see that the output of a sinc3 filter begins to transition approximately 1.5 samples after an input step. For your application, would this be sufficient to quantify the latency between nerve stimulation and brain response?

    Regards,

    Ryan