I spent 7 ½ hours in the circuits lab today, helping students finish their optical pulse monitors. One group finished on time, and a couple of others finished with only an extra hour, another group gave up (taking the project home to see if they could finish there), and one student went to a meeting, but came back before I had left, so I stayed until she had finished also. Everyone got the first stage working, and were able to see pulse signals with a 2–10mV amplitude (on top of 100–200mV DC and about 20mV 60Hz noise). All but the group that took the project home got the second stage working and able to record clear pulse signals on the PteroDAQ software (though some had the gain a bit too high and got clipping).
I don’t think that this lab needed to take as long as it did for the students, and I’ve been thinking about ways to get them to do it much quicker next year.
- The first thing to do is to separate the electronics from the scaling of the light signal. Their math is too slow to put it on the critical path to getting the lab done. Instead, I’ll make the analysis of the size of the signal they ought to see be done after a more tinkering approach to design has been done.
- The next thing to do is to require that a schematic for the first stage be turned in on Monday, so they work on it over the weekend. They’ll have to come up with a guess at the initial resistor size, but I can give them a rough estimate (within a factor of 10, say) of the current to expect from the phototransistor.
- I can also teach active RC filtering a little sooner, so that I can have them design the low-pass filter to reduce 60Hz interference in their first design.
The 60Hz noise is a much bigger problem in the circuits lab than at home—probably because of all the fluorescent lights and benchtop equipment. It was reduced enough by a single RC low-pass filter that it did not saturate the second stage, and it was well removed by synchronous sampling (so I’m glad the PteroDAQ software allows specifying sampling frequency and not just period—it is easier for students to enter 30Hz than 33.333msec (which was not a period the old software supported, anyway—they would have had to go to 50msec).
One problem that some students encountered today that I had not anticipated is that some fingernail polish is opaque to the 627nm LED light we were using. I don’t normally wear fingernail polish, so it had never occurred to me that designing the fingertip sensors to shine through the fingernail might be a bad choice in some circumstances. I suppose that this is yet another reason why we need a diversity of engineers—unrealized assumptions of one subculture may be obvious problems in another.
I don’t know what I’ll do about the fingernail-polish problem next year (this year the student just scraped one fingernail clean and got the circuit to work). I suppose I could simply announce that fingernail polish may be opaque, so students may wish to keep one finger clear on lab day. I could let students discover the problem on their own (or with a little guidance from me), as happened this year. Or I could try to design a different way to mount the LED and phototransistor.
One problem with the setup I used this year is that to get a good signal you need to apply enough pressure to your finger to be between the systolic and diastolic pressures. That can be difficult to do, though with a bit of practice I got pretty good at it today, as I helped students debug their circuits. It would be good to have a simple adjustable spring clip that would hold the two optoelectronics components and apply the appropriate pressure (around 80mm Hg, 10kPa, or 1.5psi). The reason my first attempt at an ear clip was such a failure was that it squeezed too hard and cut off all circulation—if I had made a clip that squeezed with only 10kPa, it might have worked fine.
Over the summer I might play around with different designs for mounting the optoelectronics, to avoid the fingernail-polish problem and to get a more controllable squeeze for whatever part of the body the light is shining through.
Tomorrow I’ll have to cover strain gauges, Wheatstone bridges, and instrumentation amplifiers. I checked today that I can still derive the gain equations for both 2-op-amp and 3-op-amp instrumentation amps. I’ll assign students to do their block diagrams, schematics, and layout for the pressure sensor amplifier to turn in on Monday, so that they’ll be able to start lab promptly on Tuesday. With any luck, Thursday can be a catch-up day for any labs that needed to be redone with a little more data.