ADS1292R: Using external clock for respiration modulator

Hello Ccent,

Thanks for the update. I'm still trying to confirm the limit on the tBLKDLY spec with a designer. I think in the end, we chose a limit which we thought gave the best results through empirical testing.

Intuitively, the reason we block part of the incoming modulated waveform is to reject the initial R-C time constant produced by the switching of the RESP_MODP and RESP_MODN waveforms. What we want to pass through is only the settled voltage that appears at the Channel 1 inputs. The output of the demodulation circuit will reflect how much of the incoming modulated waveform is blocked/passed through (hence the name). If tBLKDLY is too long, you will be passing through some of the initial unsettled waveform, then blocking some of it, then passing through the remaining settled portion.

The concept is much easier to understand with a plot. See Figure 9 taken from our application note on measuring respiration with impedance pneumography. Notice the difference in the black demodulated waveform - the less you block, the sooner the demodulated waveform follows the input signal. I imagine that if the blocking signal and modulation clock were not phase-aligned, then the resulting waveform will have some distorted result.

www.ti.com/lit/an/sbaa181/sbaa181.pdf

Best Regards,

This is helpful, but we still haven't been able to fully resolve the issue. Can you confirm that the internal master clock on the ADS1292R can be used when an external modulator clock is provided on the gpio? It seems that the ADS1298R requires an external master clock when an external modulator is used.

Thanks for the update. Does that mean that the external master clock and external modulator clock have to be synchronized?

Yes, the external master clock and external modulator clock must be synchronized.

Hi Ryan,

Thank you for the detailed response. We've configured the ADS1292R so that the clocks are as follows:

External modulation clock (GPIO1) = 64kHz

Blocking signal (GPIO2) = 128kHz (multiple duty cycles to test different phases)

Master clock (external) = 512kHz

All signals are synchronized within 5ns.

When measured on the RESP_MODP and RESP_MODN pins, we see the correct signals, however, the result that we receive from the ADC is not correct (when compared to using the internal modulation clock with the same settings). This leads me to believe something may be wrong with the blocking signal. Interestingly enough, if we run the modulation clock=32kHz and the blocking signal=64kHz, we do see the expected signal. Let me know if you have any thoughts on this.

Hello Ccent,

Thanks for the update.

How are the RESP1 and RESP2 registers configured during these tests? I believe the phase setting (RESP1[5:2]) and the respiration frequency setting (RESP2[2]) are ignored when external respiration signals are used (RESP1[0] = 1). In case that is not correct, can you try setting RESP2[2] = 1 when you run the modulation clock at 64 kHz and blocking signal at 128 kHz? Meanwhile, I'll try to confirm with a designer whether or not that's necessary.

Also, what is the phase of the blocking signal that you're trying to use?

Best Regards,

Hello Ccent,

That seems correct to me. If the blocking signal is 2x the modulation frequency and 50% duty cycle, you should be blocking for 90 degrees.

I'm wondering if there is too much of a delay in the signal path for 90 degrees of blocking to be effective. If your blocking signal is synchronous with the modulation and demodulation clock, but not in phase with modulated signal seen at the Channel 1 inputs, then it will have less of an effect. As I mentioned in my previous post, the blocking signal prevents you from demodulating the beginning of the modulated input signal, which contains an unwanted R-C rise component. Aligning the phase of the demodulation clock with the modulated input is not really feasible, so blocking is a simpler alternative. However, if the incoming modulated signal includes a significant delay, this blocking will be less effective, so more blocking time may be needed.

Try increasing the blocking to more than 90 degrees for 64 kHz modulation clock (it may improve your results with 32 kHz modulation as well). Page 59 of the datasheet gives a few recommended combinations for PGA gain, RESP_MOD frequency, and blocking phase which were empirically found to give the best noise performance.

Best Regards,

Hi Ryan,

Thank you for the reply. We're still trying to resolve this issue. We think we've narrowed it down to the blocking signal. The part operates fine with an external 512kHz master clock and we're seeing the correct output on the modulator as we change the modulator clock frequency. However, the demodulated signal does not look correct. It is very different from what we see when we run the part with internal modulation and blocking signals at the same frequency and duty cycle. I've included an image of our modulation (64kHz) and blocking (128kHz) signals. This example shows a 157 degrees phase for the blocking signal. We've tried other phases with no success.

Any additional insight would be greatly appreciated. Thanks!

Hi Ryan,

Thanks for looking into this - I look forward to seeing your findings. 

Hi Ccent,

Please excuse my delay.

The attached Excel sheet summarizes the results of the respiration measurements with the ADS1298R. Using the internal master clock, I first configured the device to use internal respiration and blocking signals. The duty cycle of the blocking signal was varied to produce different phase settings from 22.5 degrees to 157.5 degrees with RESP_MOD = 64 kHz. Of course, the master clock is guaranteed to be an integer multiple of the RESP_MOD and blocking signals and everything is synchronous in this scenario.

For comparison, I then provided the RESP_MOD clock and blocking signals externally from a dual-output function generator while still using the internal master clock. The results are very similar. In this case, not only is the internal master clock asynchronous to the external RESP_MOD and blocking signals, their frequencies are also no longer integer multiples of one another. I did not see a major difference in the results.

To confirm, I also tried using an external master clock (10-MHz reference shared with the dual-output function generator) and adjusted its phase relationship to the synchronous external RESP_MOD and blocking signals. I found little to no correlation there either. I was even able to match the datasheet noise spec for respiration impedance measurements.

What are you using as an input to the ADS1292R to evaluate the respiration performance? Can you also share a summary of the results you obtained with internal and external respiration signals for comparison?

At this point, perhaps a quick call would make sense. Let me know if you would like to arrange something for early next week and I can contact you offline.

ADS1298R_External RESP_MOD and Blocking.xlsx

Best Regards,

Hi Ryan,

Thanks for all of this data. Can we jump on a call this week to discuss further?