0
\$\begingroup\$

I am using a heart rate sensor whose schematic is provided here. The circuit uses an operational amplifier (towards the right of the figure) which among other things, does a filtering operation. I would like to get a hint on what kind of filter configuration this is. Specifically, I am interested in the following:

1- Is this an active low pass filter, high pass filter or both?

2- What is the frequency range of this circuit? In other words, what are the cutoff frequencies of these filters.

If someone can provide any relevant hints, that would be great. I would also appreciate if somebody can be point me to relevant resources on this.

The APDS-9008 is a Broadcom package that is a photo sensitive LED with a unity voltage follower and the MCP6001 is a low powered, 1 MHz op amp. Lastly, the 0603 is a Schottky Diode.

enter image description here

\$\endgroup\$
3
  • \$\begingroup\$ The signal conditioner ought to match the bandwidth spectrum of the signal. YOu can use Falstad's simultor to sketch the circuit with An OpAmp and trace the result with a log swept sine input or a step response. Or your favorite tool. THis is something you must learn. \$\endgroup\$ Sep 18, 2018 at 2:55
  • \$\begingroup\$ It's clearly a bandpass of some sort and the interesting frequencies are clearly in the ballpark of kilohertz, but it's beyond my skills to analyse it more throughly than that. \$\endgroup\$ Sep 18, 2018 at 3:18
  • \$\begingroup\$ Also are you sure the sensor has a voltage follower, the datasheet is crap but it talks about output currents rather than output voltages. \$\endgroup\$ Sep 18, 2018 at 3:19

1 Answer 1

1
\$\begingroup\$

The signal conditioner ought to match the bandwidth spectrum of the signal. You can use Falstad's simultor to sketch the circuit with an OpAmp and trace the result with a log swept sine input. Due to limitations in this AC analyzer for low freq=1Hz, I scaled all frequencies up 3 decades by reducing all C's down the same.

As expected it matches the spectrum of a heartbeat at 120BPM or 2 Hz. In theory , if my assumption is true, this is called a "matched filter" the optimum for signal conditioning to max SNR.

enter image description here

With lots of practice, this took me 10 minutes, and another 10 to edit the screen shot and make it pretty.

Here's a built-in filter

\$\endgroup\$
4
  • \$\begingroup\$ A simple bandpass in this case. A matched filter has a more complicated kernel, a bit more than what a simple opamp+RC can handle. \$\endgroup\$ Sep 18, 2018 at 6:35
  • \$\begingroup\$ @aconcernedcitizen Nope. A Matched Filter is exactly as I defined. The theory for complex brick wall filters is also true but does not exclude the fundamental. requirement to match the spectrum and the phase \$\endgroup\$ Sep 18, 2018 at 12:41
  • \$\begingroup\$ It's a compromise of the signal spectra because BPM changes. Of course, I have seen the Fourier spectra of heartbeats and recall its shape. is an approximation. \$\endgroup\$ Sep 18, 2018 at 12:53
  • \$\begingroup\$ Are you saying that the wikipedia page is wrong? Or that this is, indeed, a matched filter? \$\endgroup\$ Sep 18, 2018 at 15:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.