ADS1299EEGFE-PDK: Spike in the digital output when there is a step input out of band signal at the ADS1299 input.

Part Number: ADS1299EEGFE-PDK
Other Parts Discussed in Thread: ADS1299

Tool/software:

Hi,

We are using ADS1299 EVM in ECG measurement application. During the measurement we are applying 50kHz sine wave at every 150mS from a different device for the duration of 1mS.

We see a spike in the output at the beginning and end of the 1mS burst signal. We did measure the analog input at the ADC after the RC filter in the EVM. We could not see any spikes.

ADS all inputs negative channels are connected to a single reference input.
In the channels where the 50kHz signals are not fed, there are no spikes in those channels.

1. What could be the probable reasons for this spikes ?

We thought this could be because of the sinc3 filter settling time?

We did one experiment to see our understanding, we tried to capture the same waveform at different data rate  - from 250SPS to 16kSPS.

We see at each data rate the spike remains 3 or 4 samples duration (so time is varying). If it is due to an analog spike /noise noise duration would have been fixed.

Is our understanding correct ? are there any fixes to this problem ? 

I have attached the data collected from EVM and plotted in an excel. Please check the summary sheet for the sample number observed. and let us know your thoughts ?

ADS1299EVM Spike data_TI.xlsx

  • Hi Chandann,

    Thank you for your post. What is the amplitude and common-mode voltage of the 50kHz sine wave? And what is the reference voltage connected to INxN pins?

    The output waveform data looks mostly correct, with some obvious power line noise coupling. I do see some random "spikes" in the waveforms, which become more frequent at higher data rates. Is this the behavior you are asking about? Can you just share a singular plot which illustrates the "spike" and settling time?

    My assumption is that this 50kHz noise is aliasing into lower frequency bands. The nature of the aliased noise will depend on how much of the signal is attenuated in analog/digital and the frequency to which it aliases. Any sinc3 digital filter will require 3 samples to converge to a new settled value, so it is possible that the input signal is large enough and fast enough to cause such behavior. The only solution would be improved filtering of the unwanted signal in analog.

    Regards,
    Ryan

  • Hi Ryan,

    Thanks for the quick response.

    - 50kHz sine wave is AC coupled. We are operating ADS EVM in bipolar supply. Expect common mode voltage to be 0V.
    - All INxN pins of ADC is connected to RA electrode of ECG.

    - Yes. We are discussing about the same spikes.

    - I have attached the extract of the noise spike images at 2kSPS, 4kSPS and 8kSPS from the excel sheet.

    - We increased the burst duration from 1mS to 100mS and 1 Sec. The spike always appears at the start and end of the burst signal irrespective of the burst width. This behaviour I could not understand completely. I could only come-up with the explanation as mentioned in the query (first message). If it is aliasing noise, we would expect noise across the low frequency pass band isn't it ?

    - I have attached the 50kHz signal seen at the ADC input after RC filter (4.99k and 4.7nF) with respect to SRB1.

    Please let me know if you need any more information.

    Spike Seen with 8kSPS

    Spike seen with 2kSPS

    Spike seen with 4kSPS

    50kHZ single ended at the CH3 Input with respect to SRB1

    At the ECG simulator (Before EVM) - Differential 50kHZ Between CH3 and CH5