Questions about High pass and low pass filters

[CWMC]

I have a few questions regarding High pass and low pass filters when calculating a pulse rate.

Why is it that in this link, https://mikroelectron.com/Product/Finger-heart-rate-sensor

The cutoff frequency is set to 0.5Hz? Why did the capacitor and resistor be chosen for the HPF?

Also why does the gain and cutoff frequency have to be 48 and 3.4 Hz for the LPF?
What is the use of the potentiometer and why is it 10k?

Any help would be greatly appreciated.
Thanks

 

[KlausST]

Hi,

The DC oparating point of the OPAMP is V_Ref, which is 2.0V.
But the sensor maybe has any other operating point.
Therefore you need to seperate both operating points. This is done by blocking DC.
A HPF is the usual way to do this.

Now why 0.5Hz.
Heart rate is appriximately 40 beats per minute (sportsman doing nothing) up to 200 beats per minute (heavy working).
40 beats per miunte is 40 bpm / (60 s/min) = 0.666 bps = 0.666Hz.
You want fc below the lowest frequency of interest. 0.5Hz is well chosen.

Gain of 48:
The input signal needs to be amplified to best fit to an ADC (or whatever the signal processes).
The signal amplitude of the input is unknown, the ADC input range is also unknown. At least to us.
But the developers will know.

3.4Hz:
200bpm / (60s/min) = 3.33bps = 3.33Hz.
Also well chosen. They just focus on the heartbeat rate.


10k pot:
You could use 1k ... 100k.
1k gives less noise, while 100k gives less power consumption.
10k is a good compromise.

Klaus

 

[CWMC]

Thank you for your fast response.

So say having a HPF cutoff frequency of 0.72 Hz that would be alright if the lowest heart rate wanting to measure was 50 bpm?

Having a gain of 100 when using a MCP6004 Low power op amp means what exactly when connecting to the INT pin of a microcontroller?

 

[KlausST]

Hi,

0.72 Hz cutoff frequency:
This is a first order low pass only. So don´t expect a steep attenuation curve.
The gain of an 0.72Hz signal (50bpm) is 70.7% (related to the input signal voltage)
The gain of an 0.67Hz signal (40bpm) still is 67.9% (not much difference)

***
If you want to feed a signal to the INT pin, then it should be a "digital" logic signal with valid levels (V_IH, V_IL), not an analog one.
I recommend to use a comparator with some hysteresis to generate this digital signal from the anlog signal.

Klaus

 

[CWMC]

So what would I feed the amplified signal from the pulse sensor into before going to the INT pin of the microcontroller?

 

[Akanimo]

Hi CWMC,

So what would I feed the amplified signal from the pulse sensor into before going to the INT pin of the microcontroller?

Pulses are already digital signals. But just like Klaus had mentioned in Post #4, the signal levels from the sensor have to be compatible with those of the MCU. If they are compatible, then you can feed it directly, else you'll have to have something in-between the sensor and the MCU that makes the signal compatible.

However, it would be preferable you choose a sensor with output signal levels with respect to the input signal levels of the MCU.

You may find it easier to let us know what sensor you're using.

 

[CWMC]

I am using TCRT1000 and PIC16F887. I am feeding the output signal straight to the INT pin of the microcontroller. Once connected, i am getting different values for the heart rate. Sometimes it read 35 bpm the next its 3000. Not entirely sure what i am doing wrong.

 

[KlausST]

Hi,

Follow my recommendation of post#4.

What else do you need?

Klaus

 

[Akanimo]

I am using TCRT1000 and PIC16F887. I am feeding the output signal straight to the INT pin of the microcontroller. Once connected, i am getting different values for the heart rate. Sometimes it read 35 bpm the next its 3000. Not entirely sure what i am doing wrong.

I took a look at the document you linked in Post #1.
The first thing to realize is that the waveform provided in the document is one obtained with the HRM-2155E sensor. The waveform obtained with the TCRT1000 sensor may not be the same as that waveform. I think you need to take a look at the waveform that the TCRT1000 gives to gain an insight on what how best to process the signal to obtain the required pulses.

Another point is that the TCRT1000 sensor is susceptible to a very small movement of the finger as the document states. You need to devise a means to take care of that.

If the waveform are the same, then the document is pretty straightforward. You need
Sensor-->1st-stage passive HPF-->1st-stage active LPF-->2nd-stage passive HPF-->2nd-stage active LPF-->buffer-->comparator(just like Klaus had recommended)-->MCU.

However, I strongly recommend that you post a picture of the waveform you obtain from the TCRT1000 sensor so we look at it together.