Noise on pleth signal

Ed Hepler

Expert 1940 points

Other Parts Discussed in Thread: AFE4490

I'm using the AFE9940 in push-pull mode with a reflective sensor.

The AFE9940 is programmed using the "example" settings as shown in the TI driver code. The Pulse Repetition Frequency is set, as in the example, to 100 Hz.

I'm getting a signal, but there is a 20 Hz noise (sine wave) on top of the pleth waveform.

If I postprocess the datapoints, the 20 Hz peak (in the spectrum analysis) is clearly seen. If I put the datapoints through a low-pass filter (again postprocessing in Matlab) to filter out the 20 Hz, the pleth waveform appears as expected.

The system actually has 3 AFE9940's in it. They all exhibit the same 20 Hz noise.

The system is battery driven, so there is no connection to AC power.

I haven't been able to figure out what the source of the noise might be.

Can anyone suggest where the 20 Hz might be coming from? Is there any possible source within the AFE9940.

Thanks!

over 10 years ago

0 Amy Schnoor over 10 years ago

TI__Mastermind 18285 points

Ed,

We have received your post and hope to respond soon. A few questions...

1. You are referring to the AFE4490 correct? Not the AFE9940 as mentioned in the post.

2. Could you possibly give us a screenshot of the 20Hz noise you are seeing?

0 Ed Hepler over 10 years ago in reply to Amy Schnoor

Expert 1940 points

Yes, sorry... AFE4490 !

Here is a plot of the data as captured and plotted via Excel...

Thanks!

Noise_on_Pleth.docx

0 Ed Hepler over 10 years ago in reply to Ed Hepler

Expert 1940 points

Hi Amy,

Any additional thoughts on this?

I found the article: www.mdtmag.com/.../designing-heart-rate-monitor-wearable-devices
authored by TI employees... It appears that Figure 5 in this article also seems to have noise similar to what I am seeing (even in the "Finger at rest" portion of the plot).

0 Amy Schnoor over 10 years ago in reply to Ed Hepler

TI__Mastermind 18285 points

Ed,

Sorry for the delay. We have seen this before and it was caused due to the fact that there is a conductive path to GND when touching the surrounding circuitry of the sensor. A few workarounds:

1. Use the filter you have implemented
2. Use some potting material or tape around the sensor to prevent any contact around the sensor. This way the subject is only touching the sensor and nothing else on the board.

0 Ed Hepler over 10 years ago in reply to Amy Schnoor

Expert 1940 points

Thanks Amy...

I tried surrounding the sensor and LED circuitry with "cling wrap" clear plastic, but got the same results.

The filter would be a reasonable solution at the sampling rate that we are using for testing, but the ultimate application requires a much higher frequency sampling rate, so the number of taps needed to filter out the 20 Hz signal is prohibitive.

Any other thoughts?

Thanks,
Ed

0 Amy Schnoor over 10 years ago in reply to Ed Hepler

TI__Mastermind 18285 points

Ed,

Are you sure the plastic you used is not conductive at all? We have found some plastics to be slightly conductive. Can you try black electrical tape? We have used that and it was worked for us in the past as a temporary solution.

Data converters

Data converters forum

Noise on pleth signal