My optical pulse monitor (with a phototransistor, a transimpedance amplifier, a high-pass filter, and a second stage to increase gain) was having problems with capacitively coupled 60Hz interference. Because the combined gain of the two stages is over 700MΩ at the gain setting I usually use, even very tiny currents at the input cause large output signals. On the design I had in the lab to demo the method, I was getting 60Hz interference that was almost a large as the pulse signal (though grounding myself reduced it by a factor of about 4).
I have been relying on a low-pass filter in the transimpedance amplifier (a capacitor in parallel with the feedback resistor that sets the gain) to reduce the 60Hz interference, and I realized last week that I could set the corner frequency much lower, from around 15Hz down to about 2.2 Hz. The signals I’m interested in are blood pulses, so the interesting signals are about 0.3–3Hz. I don’t need to keep the detailed shape of the pulse waveform, so a little attenuation at the highest and lowest parts of the passband is a good tradeoff for reducing the 60Hz interference over 6-fold.

The transimpedance amplifier I wired up with a rather low corner frequency for the low-pass filter. The filtering is built into the transimpedance stage, by putting a capacitor in parallel with the resistor, so that the gain of the amplifier is (R||Zc).
Here is a plot of the gain of the minimum gain of the two-stage amplifier (with pot set to be 10kΩ):

The narrow bandpass of the new design peaks around 0.95Hz (59.4 bpm), but has adequately high gin over the entire range of likely pulse rates. The old design gave more high-frequency detail to the pulse shape, but at the expense of too much 60Hz pickup.
Although the minimum gain for the amplifier is about 55MΩ, I usually need to set the gain higher (up to around 700MΩ) to get a clear pulse signal to record with the PteroDAQ data acquisition system.
Leave a Reply