Oops !

Thanks to all for the suggestions. A few comments:-

Probing along a track sounds like a good idea, but the reality is that the surface tracks are 6 thou wide, covered with solder mask, and very easy to destroy. There are no surface power tracks, apart from the short connections to decoupling caps. The only practical access points are at vias and component pins. In my case, the current wasn't flowing along a surface track anyway !

Probing with a 4 wire ohmeter is doubly difficult, because you need two connections at each point. In practise this means soldering a contact in place. I got some very strange readings until I remembered about thermally generated emf's and allowed everything to cool after soldering.

So I did some simple voltage and resistance gradient 'maps', which didn't really show anything. Big puzzle - back to square zero.

Finally I realised that the reason that there was no voltage/current gradient across the board was that the measurement reference point (main Vcc decoupling cap) and short were in pretty much the same place. A few minutes later a near invisible solder splash turned up on a regulator chip which was ~ 5mm away from my reference point !

Normally I do some tests as I incrementally assemble a new board, but as this was a variant of a working design, I got somewhat careless. I'm not sure that it would have helped anyway, because the solder splash must have occurred quite a long time after the regulator was installed and tested ...

Looking back on it, I wasted hours on something that should have taken minutes, but I hope to have learnt enough to shorten the next one ...

Thanks again ! Dave

I was once debugging a board that had such a short. I wondered why I could see a light shining through the board even though there were no LEDs on it.

Unless you get a thermal imager and use that to look for hot spots.

--
Dirk

The Consensus:-
The political party for the new millenium
http://www.theconsensus.org

With the four wire ohm ohm meter, I put the "current" connections on the inputs to the board, and left them there, and probed with the "voltage" connections.

A good four-wire ohm meter should use AC (or at least reversing DC) excitation to get rid of any thermal emf. This is the real advantage that proper ohm meters offer over driving a few hundred mA through the board with a current limited power supply - with AC excitation you can reliably measure much lower voltage drops (microvolts rather than millivolts).

-------- Bill Sloman, Nijmegen

of

some

fully

circuits

current

cap.

but a

I agree that it's likely not a shorted cap - we must've placed at least 1/2 a million decoupling caps of various sizes, types, and brands in the last year, and have yet to have a significant problem with any of them, at least electrically. We have, however, seen shorted boards, which were supposedly electrically tested, but the quality of the boards was not the greatest.

Most likely the problem is a solder bridge somewhere. They can sometimes be hard to see, even with a well trained eye looking directly at the short (ex.

0.4 mm pitch part with short on the bottom side of the leads, or a soldering iron drug across a 0603 cap without any flux can leave a thin film of solder shorting it out. I once saw a customer's board with such a short, that was hard to actually tell if it was a short, even under a microscope!).

I made a nice KABOOOOOOOOOOOOOM 8" away from my face once when changing a breaker in a 200A 120/208V 3 Ph panel. I was getting ready to screw in a new breaker, and while trying to line up the screw to the hole that fastens the breakers tab to the bussbar, the bakelite housing between the breakers fell away (significant, but unknown not so visible damage to that area of the panel), alowing the breaker to drop, which caused the breakers tab to short out between 2 phases. I don't remember the bang being extremely loud, but other people said it was defining. I couldn't see for about 10 to 20 seconds, since the arc was so bright. My hair and eyebrows were a singed, and I had a bit of a sun burn from it. After brushing away all the singed hair and getting my eyes checked at the hospital (potential welders flash - my contacts may have prevented it), everything was fine. Looking at the damage afterwards, the bussbar had serious arc nicks in it, the breaker tab was partially missing, the screw no ware to be found, black and gray smoke stains all over the place, on the bussbar, bakelite, breakers, shielding plastic, the screwdriver tip and insulated shaft, etc.

What you say about AC measurements is true, but do you know of any specific bench multi-meter instruments that use ac excitation?

--
 Thanks,
    - Win

Got a call to bring some fuses to a remote CO once. When I got there a

2nd level manager RAN out to my truck, and got the fuses, and ran back in. Being curious what could possible get this lazy SOB to walk, much less run, I went in and inquired what was happening. Seems they had a contractor tightening the bolts on the battery buss bars, and dropped his socket wrench. It was not wrapped in electrical tape, as per standard, and fell across the 48V main busses. It blew ALL the fuses in the rectifiers, and they didn't have that many fuses in the office, so it was now off-line. Fortunately for the contractor, he was in the cable tray ABOVE the busses when he did this. They were sweeping up ball bearings for weeks...

Charlie

I read in sci.electronics.design that Winfield Hill wrote (in ) about 'Oops ! - Thanks !', on Wed, 28 Sep 2005:

Not a multimeter but I have a very old (toob!)4-wire milliohmmeter that uses a 50 Hz constant-current source. 1.2 mohm to 120 ohm FSD. Still as accurate as the analogue meter allows.

--
Regards, John Woodgate, OOO - Own Opinions Only.
If everything has been designed, a god designed evolution by natural selection.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

Polar Instruments

make some great gear for just this sort of thing.

Regards,

Gerry