I did not lecture on Wednesday, but had a guest lecturer from the biology department, who gave a lecture on action potentials, based on a similar lecture she gives in a neuropsychology course. The lecture went a bit slow for the bioengineers, I think (with everything repeated 3 times), but there was some good information about the use of the Nernst equation to get resting and action potentials, the sodium and potassium channels, and the use of a voltage clamp to measure permeability of a squid axon to ions at different potentials. If I were to give the lecture, I would probably pick up the pace a bit, but use a longer wait time when asking the class a question—the biology professor tended to answer her own questions as soon as she got any sound from the class.
The EKG lab this afternoon had mixed results—everyone finished with a working EKG board, but the last group to leave took 7 hours, so I was in the lab from 2 p.m. to 9 p.m. I started the lab by demoing the board that I had build Tuesday morning, so that they could see that the project was possible. It was a good thing that I did so, since several of the EKG boards the students designed and built worked fine with the electrodes I was wearing, but not with their electrodes on their bodies. I’ve not figured out why this happened. Had their electrodes dried out? Was their skin prep or electrode placement poorer? Is my pulse larger than theirs, or my skin more conductive? Some of the groups did manage to get their boards working with their own electrodes, and everyone’s worked with some set of electrodes.
One of the slowest debugging problems was a component value error: assembling the board with a 4.7µF capacitor, rather than the 4.7nF capacitor they had designed in their schematic, resulting in a low-pass corner frequency of 0.05Hz instead of 50Hz, eliminating their signal. Another component value error by a different group was using an 8Ω gain resistor (giving a gain of 10,005) when a gain of 8 was desired—this resulted in saturating the amplifier. There were other component errors that were less catastrophic (a 47nF capacitor where a 4.7nF capacitor was intended, eliminating the 60Hz ripple, but also eliminating a big chunk of the signal), but all were eventually found and fixed.
Almost everyone had a lot of 60Hz ripple in the output (as did I, though they generally had more with their electrodes). I was confused on Tuesday about what I saw as a 40Hz ripple, but of course that was just aliasing of the 60Hz ripple with a 100Hz sampling frequency (with a Nyquist frequency of 50Hz, a component at 60Hz and a component at 40Hz are indistinguishable). I think that most of the 60Hz ripple is coming from electromagnetic pickup in the loop from where the wires stop being twisted together and spread out to the three electrodes, but some may be capacitive also. I may play around with better wiring to see if I can reduce the ripple. I could also try sampling at 60Hz, so that the ripple is aliased to DC. The only risk there is that it may not be exactly 60Hz, and aliasing to near DC could interfere with the meaningful part of the signal. I believe that professional EKGs sample at a higher rate, then use a digital notch filter to remove the 60Hz ripple.