Page 173 - Troubleshooting Analog Circuits

I60 13. Letters to Bob

Figure 13.2. Letting lTL inputs float HIGH is a risky business. They won’t do any harm on your bread-
board-only when you go into production. It’s wise to tie them up towards +Vs with 1 kR
or so.

wearing ground straps when handling MOS ICs. My experience, and that of others,
has been that MTBFs plummet if you don’t handle MOS ICs with kid gloves. The
problem is that the device rarely fails on the spot, but having possibly been over-
stressed, will do so at a time and place of its own choosing.
Turning to the article in the September 28, 1989, edition of EDN (Chapter 4), I feel
bound to present a caution regarding tantalum capacitors. First, they are even less
tolerant of reverse polarity than electrolytic capacitors are. Reverse polarity can arise
when, say, coupling op amps with tantalum capacitors. Second, I have replaced more
tantalum capacitors than I can remember because they short-circuited for no good
reason.
The worst place for a bad tantalum capacitor is on a computer motherboard, which
is just where I found the last one I replaced. The job started out as a short that was
shutting down the power supply. Ascertaining which rail the problem was on was
easy enough. I decided that because of the number of components on the bus, I would
have to try something less radical than temporarily isolating a section of the rail. I
tackled the problem by feeding a 10-kHz sine wave at about 1 V rms through a 1 kR
resistor and monitoring the traces with a current probe. I found the offending compo-
nent in less than a minute. I may have been lucky, but, on the other hand, my test
signal wouldn’t cause the ICs to draw any significant current and was of a low-
enough frequency not to cause enormous current in the good decoupling capacitors.
In the same article, you said that floating lTL inputs is reasonably OK. In some
situations, particularly noisy ones, I would have to disagree. I have seen nasty prob-
lems where this practice causes random glitches. The circuit in Figure 13.2 is an
absolute nightmare. Yet this practice is common on IBM PC clones from Asian man-
ufacturers in, of all things, the glue logic of 80386 motherboards, where high-speed
clocks are the norm.

Malcolm Watts
Wellington Polytechnic
Wellington, New Zealand

Dear Mr. Watts:

Thank you for your comments. You question the practice of pulling current out of
a transistor’s base circuit. If you were to actually pull current out of the base and