Heart beat monitor with AD8232 and slow CPU

[SanchoSK]

Hi, everyone.
I am re-building my training rotoped and I want to connect it to Home Assistant.
Part of the rotoped is also heart-beat sensing - on both handles the device uses conductive pads to measure heart rate.
The circuit inside is a big blob, so I don't have any way how to connect to it.
Anyhow, I bought AD8232 as this is used for ECG. But when connected to the pads instead of the regular probes, the signal has quite a lot of noise.
I investigated in using my cheat oscilloscope and found out, that during the heart beat, there is a peak of around 2.8V and above and all the noise is below the voltage.
This would be, in general, also possible to process with micro-controller, however, the one I want to use also drives the display and has wifi connection to handle, so it cannot monitor and process the signal all too often.

My idea is - I'd like to connect the output to some comparator that will then trigger output pulse of predefined value - let's say 200-250ms and then go low again. This would be plenty for the MCU to detect the change from low to high, count the pulse, calculate the BPM and return the control back to the rest of the processes (refreshing display, reporting to wifi, ...).

I remember from ~30 years ago that NE555 was able to work in such mode - both as comparator and then as pulse generator that will auto reset.
But I have no idea what to search for on the internet and all the schematics I found were only for oscillators based on 555.

Another problem is that if I look at the datasheet of the 555, it requires minimum of 4.5V, while my display and the AD8232 work with 3.3V. Is there some low voltage version of 555 that would work the same?

Any advice is appreciated.
I am attaching the sample from my cheap oscilloscope.

 

[Tony Stewart]

In order to use the CMRR benefits of the AD8232 you need to use the RLD output. Maybe a connect that to a foil or braid strip on the seat and the coupling capacitance may be enough to suppress the 50 Hz. The next option is to use a high inductance CM choke with a 50 Hz filter.

 

[SanchoSK]

Thanks, but I'd like to do it the same way as the original rotoped was done - as I tend not to sit with my naked buttocks on the seat
In the meantime, I came up with circuit like the one in the attachment.
I used a pulse generator with biased input of 2.5V low with 5% duty cycle and 2.9V high. This simulates quite closely the inputs I am getting from the AD8232.
It seems the output does exactly what I need - generates a pulse with ~220ms duration, that should be plenty to detect with the microcontroller.
I am just not sure if the simulation of the circuit will reflect the reality - I guess there's only one way to find out

 

[Tony Stewart]

Your body picks up 50 Hz E-field right through your clothing. You do not need to be naked to detect it when you have a large surface area. To prove this, get a a sheet of aluminum foil and probe that with your 10 Meg probe but ground disconnected. Then using a T shirt or cloth, put your hand over the foil to detect 50 Hz. By injecting the inverted Vcm it will can cancel the noise you see.

Your signal/noise ratio is insufficient to avoid errors.

 

[Tony Stewart]

You can also add a Notch filter

 

[SanchoSK]

This looks cool. I'll test that, thanks a lot!

 

[danadak]

Since the 50 Hz is quite stable in frequency, one seeks high Q (depending
on rejection goal) in filter to get the best rejection possible. That forces one
to consider sensitivity issues of components, including OpAmp used. Or
consider tuning it.

Some ref material, attached.

Adjustable - https://www.ti.com/lit/an/snoa680/snoa680.pdf

Of course DSP based on clock/sampling frequency, coefficient numerical accuracy, can
achieve real high performance low sensitivity results.


Regards, Dana.

 

[danadak]

There are single chip easy to configure solutions that handle mixed signal
manipulation :

This is drag and drop, each onchip component (in PSOC language a component is an onchip resource)
and a lib of f() calls to manipulate it. Folks have used this to create single chip oscilloscopes, and lots
of ECG EKG activity. Even impedance analyzers, very versatile.

Whats onchip, multiple copies in most cases :

 

[SanchoSK]

This is a bit overwhelming for amateurs of my rank
I will check this, but still reading the material above and trying to simulate in circuit simulator...

 

[Tony Stewart]

Dana's solution performs far better, with bandstop and bandpass performance.,

The twin T demands precision values to cancel amplitude and phase shift exactly, 0.1% preferred or one with a trimpot. It was just a quick design, not verified.

You can scale all R's x10 and caps C/10 or more