January 2005, Issue 174

Light-to-Frequency Conversion (Part 2)
Pulse and Oxygen Content

BEATS PER MINUTE

Based on the sampling period of 31.25 ms, a heart beating at 60 bpm, or 1 bps, would have 32 samples between maximum peaks. So, the beats per minute equals 32 divided by the number of samples (CNT) times 60 bpm: 1,920/CNT. With a design scope between 50 and 250 bpm, that would be approximately 39 samples (at 50 bpm) and seven samples (at 250 bpm).

Converting the number of samples to beats per minute is adequate for the low end where the number of samples is larger. At the upper end where the number of samples is lower, the resolution is unacceptable. This can be improved by averaging a number of samples. By averaging eight samples, the calculated beats per minute error (because of resolution granularity) can be cut to less than 2% in the 250 bpm range. Empirically, the 50 to 250 bpm range is a bit skewed, where 40 bpm seems like a better low end. When I was resting, my rate was approximately 50 bpm. At the other end of the spectrum, 250 bpm seemed too high a value because a resting infant measures approximately 110 bpm. Even under stress rates of 250 bpm seem extreme.