You have to promise not to laugh or to suggest that perhaps the Irish whiskey is a contributing factor.
There were many, but here's a couple examples:
Jan developed an interface for using sound cards as audio analyzers (I wrote a series of articles about this). His circuit has two interesting twist over the crude circuits I was using- it autoranges the input signal to keep it near the 2V sweet spot of most sound cards and it has a tracking notch filter for a 1kHz input so that if the frequency is off by a few Hertz, the notch adjusts. It seemed to work well, but we kept getting -110dB 3rd harmonic regardless of the range or where the notch filter was centered or the fundamental frequency. "Huh," he said, "maybe I need to change out the opamps..." After playing around a bit more, it suddenly occurred to me- I had been almost exclusively using the AP to generate digital outputs, using the analog generator to measure tube amps. In this case, I was using its analog generator for an ultra-low distortion op-amp-based circuit. So we did a baseline loopback measurement which showed... yep, -110dB 3rd harmonic. His Autoranger was working perfectly. I sheepishly went to the AP website and checked the APx525 distortion specs, which were, "-110dB typical." Oops.
We were doing noise measurements of his Silent Switcher, which is a module that plugs into a USB port and outputs +15, -15, and +3.3/5/6.5 (depending on a resistor setting). We of course did a baseline noise measurement with the probe leads shorted, then measured the +15 output, which showed more noise than baseline. Fine. Then we ran the -15, and its noise spectrum tracked the +15 almost perfectly. "The +6.5V doesn't have a post regulator, so it should be noisier," he explained. And when we tested it, its noise spectrum overlaid the +15 and -15 curves. That's not right (because we understand what regulator noise spectra should look like), so I grabbed my bench voltmeter and checked the outputs. Yes, they were all zero! He furiously started checking connections, changing jumpers, then suddenly yelled, "SHIT! I installed the USB connector on the wrong side of the board!" A few minutes of desoldering and resoldering later, the voltages were right and the noise spectra made sense (by the way, this is a superb power supply).
Later when I have time, I'll talk about how I figured out the inadvertently bad gating selections made the loudspeaker frequency response measurement plausible but totally wrong. And how we figured out that my clip leads were hooked to the wrong BNC jacks of the AP. But at the moment, we have some drinking to do because it's 5 o'clock somewhere.