Page 168 - Troubleshooting Analog Circuits

13. Letters to Bob

My series in EDN on troubleshooting generated myriad letters from readers. Because
so many of the letters contained worthwhile troubleshooting tips and amusing per-
sonal anecdotes, we decided to collect some of the best letters into a chapter. along
with my replies and interjections. The tips just keep on coming. . .

Dear Bob:

Here are some tips and gotchas:
1. A significant source of noise in my lab is the ever-present video-display terminal. It
couples especially well to audio-frequency transformers.
2. I head off a lot of trouble by providing RF bypass on audio and DC circuits. Their
audio and low-frequency-only inputs can pick up AM radio. Having music come out
of a speaker that is supposed to be a monitor on a telephone circuit is very bad form.
3. My computer brethren frequently fail to consider what happens during reset. I saw a
thermal printer catch on fire once when its internal pP was reset. The reset 3-stated
the printhead’s drivers, which allowed all of them to tum on continuously. Later a
software bug turned them on continuously again. I finally made the printhead com-
puter-proof by capacitively coupling the drivers so that the pP had to produce a con-
tinuous sequence of pulses to keep the heads turned on.
4. My inexperienced brethren frequently forget to calculate total power supply require-
ments.
5. Vishay (Malvem, PA) produces some very accurate, very stable (0.6 ppm/”C) resis-
tors, which I keep around to check ohmmeters.
6. Some companies think they are helping designers by taking instruments to the cali-
bration lab without letting the designers know. They do not understand that small
day-to-day drifts are less annoying than an unexpected step change produced by
recalibration.
7. My computer brethren frequently lose scope-probe ground clips-the clips get in
their way and sometimes short a power supply. I gave up and bought a pile of the
clips that I keep in my bench.
8. Sad but true, sometimes adding a scope probe to a malfunctioning circuit makes the
circuit work. The probe adds enough capacitance to kill a glitch or stop a race. On
floating CMOS, the DC impedance of the scope can be low enough to pull the signal
down to a valid level.
9. At one place I worked, I was called to the factory to make my “no-good” circuit
work. The complaint was that the DC offset of an op amp was drifting. When I got
there, I found the technician had a good DVM connected to the op amp through a
piece of coaxial cable to keep out noise. Of course, the cable’s capacitance was
making the op amp oscillate. You can’t measure DC parameters when the op amp is
oscillating. Sometimes I find a scope connected this way because the tech wanted
more gain or could not find a 1OX probe.
I55