I started class today with a do-now problem:
Only a couple of students had any clue how to do this problem, though they’ve had all the pieces needed to do it for quite a while. I got them to do it as a class (making a big thing out of the block diagram of op-amp, impedance, impedance for a negative-feedback amplifier. The did eventually figure out what the circuit did, with a whole lot of prompting, but I worry about how little ability they seem to have to apply stuff they know well in even slightly unfamiliar contexts. Even if they never do electronics again, the same skills will be called on to make up new lab protocols in molecular biology, looking at what each protocol procedure does and how the needs of one protocol step constrain what steps can be done before or afterwards.
I think that there is way too much rote learning in the bioengineering program (particularly in the chemistry and biology courses, but even in the math and physics classes), and that students are not learning how to solve problems, but just applying solutions learned by rote—a fairly useless skill.
We then did quiz review for the quiz they took last Monday, looking at the stuff that lots of people got wrong. For example, one question asked what the gain equation (Vout as a function of Vin) was for a negative-feedback op-amp circuit, then asked what the constraints on Vin were for the equation to be valid. I had provided power-supply voltages for the op amp, so Vout was constrained to be between the power rails, and it was a simple matter of rearranging the inequalities using the formula for Vout from the first part of the problem, but no one thought to do that. I showed them the technique, and talked (again) about propagating constraints through a block diagram.
After that I explained why a number of students had been seeing a spike on their gate voltages in the class-D amplifier, on the edges of pulses. The problem is the reverse transfer capacitance: coupling the large voltage swing on the drains of the FETs back to the gates. We did not address how to reduce the spike, other than to suggest playing with the size of the pull-up resistor on the gate.
I showed the Old Spice ad that demonstrates skin-electrode EMG (which I mentioned in an earlier blog post), then started talking about EKGs. We started with the need to have current-limiting resistors to keep currents below 50µA, even if the highest voltages in the system get connected to the leads. We then talked about the differential signal between left arm and right arm, and I talked about electrode placement below the collar bones to avoid interference from other muscles (see Better electrode placement for EKG blinky). After a lot of questioning about how to keep the LA and RA signals between the power rails of the instrumentation amp, I finally got a student to suggest the standard solution: adding a reference electrode connected to Vref. We did not have time to talk about active cancellation of the common-mode signal.
We also did not have time to talk about the DC offset in the electrodes, and how the ±1mV signal we are interested in can be buried in a ±300mV DC offset due to different half-cell voltages at different electrodes. Monday will have to cover the full block diagram for an EKG, since Wednesday will have a guest lecturer on action potentials and how the voltages for EKGs arise.