What's the weirdest fault you ever fixed?

[Circlotron]

I've fixed zillions of electronic things in the last 40 years, but this one takes the cake. Today at work I was testing a bunch of boards of my own design. They have a Motorola 'HC908JK3 micro and one of the timer ports drives a mosfet that pulls in a solenoid. If you turned it on and just let it sit for 3 or 4 minutes the mosfet would eventually gradually turn on over a 10 second period, pulling the solenoid on. No big deal I thought - the gate drive is open circuit and the gate is floating up from leakage. Other than that the thing worked quite normally. Eventually I found that the port pin was going tri-state. It would float up to 3 volts or so and when you put the scope probe on it the pin would rapidly fall to zero. It was like the software was misbehaving or something. Put in a new mcu and no difference. I fiddled with it for about 2 hours and finally probed the reset pin and found it to be about 2.5 volts, not 5 volts. The reset cap is a 10 uF electrolytic and it was in backwards and the cap's leakage current would gradually increase over several minutes pulling the reset pin lower and lower, shutting off different sections of the chip at different reset pin voltages. As I said, the rest of the thing worked ok, just not this particular port pin. I never saw the like of it before.

What's the trickiest thing you have ever fixed?

[star882]

There's no Schmitt trigger on the reset?

I couldn't distinctly remember any strange problems I have experienced (I do know they happened, though), but a friend of mine was building a network router out of an old P3. He underclocked and undervolted the CPU, replaced linear regulators with switchers, and replaced the hard drive with a CF card. That happened to decrease the power usage to the point where the ATX power supply he was using would intermittently shut off. He spent quite a while trying to find an intermittent bad connection or short circuit that didn't exist. I suggested to him to try connecting car light bulbs to the power supply and then it worked just fine. Of course, he then ended up developing his own power supply for it and the end result was a nice router that would outperform any home use router (and rival some low end enterprise routers) and only use about 10w in the process.

[infinia]

Was doing initial development Bit Error Rate tests on our new digital demodulator chip. The demo PCB was built to include various STB TV tuner cans. In the Lab everything tested well for awile untill we noticed there were random periodic bursts of errors. We tried a few different test setups without joy. Then I noticed it wasn't a problem untill we were around. I then hit the bench causing massive errors. My backround included designing digital PLLs so I suspected correctly that the STB tuner had microphonic issues in the loop filter (DC-audio band). The fix was to change a 0.12 ceramic cap to a polester film type.

[star882]

Originally posted by infinia
Was doing initial development Bit Error Rate tests on our new digital demodulator chip. The demo PCB was built to include various STB TV tuner cans. In the Lab everything tested well for awile untill we noticed there were random periodic bursts of errors. We tried a few different test setups without joy. Then I noticed it wasn't a problem untill we were around. I then hit the bench causing massive errors. My backround included designing digital PLLs so I suspected correctly that the STB tuner had microphonic issues in the loop filter (DC-audio band). The fix was to change a 0.12 ceramic cap to a polester film type.

I have a digital TV receiver with that problem. Does it mean that I can fix it by finding the capacitor and replacing it with a polyester film? (I'm thinking of putting it in my car so passengers can watch TV. But obviously, if I can make it act up by tapping on it, it won't work very well in a car.)

[infinia]

Yep, real slight tapping esp. near the cap will pixalate or freeze the picture then open up the tuner/demodulator can and find the loop filter cap. On older designs usually between the PLL chip and VCO. It might be hard to find and read the value tho without a schematic.

[Enzo]

Oh I've been soldering for over 50 years, and have had some adventures. I recall one not so high tech problem.

Servicing an old Crown DC300 power amp, one very distorted channel. This is a real old one with the panel on the front with individual fuse holders for the channel power rails.

QUickly determined a missing rail on the bad channel. Removed the fuse, looked OK< returned it, no change. Pulled fuse and measured with an ohm meter. Fuse definitely OK. Still no voltage.

After fooling with it longer than I should have had to, I swapped fuses between holders - I was too lazy to walk over and grab a different fuse. Now the amp worked.

ANy other fuse worked in that holder, and I did try new ones from the drawer. ANd that funny fuse worked in the other three holders. But for some reason that one fuse would not work in that one holder.

COuldn;t see anything different about the fuse, nothing bent or odd about the fuse holder and cap.

I demonstrated the situation to our then new junior tech. His response was, "That can't be, that is impossible." Well you are looking at it, I said. "No, it can;t be."

The lesson was never take anything for granted. SOmething as automatic as a good fuse in a good holder is not something you can count on.

And at first, having verified the fuse was OK, told myself I didn;t need to measure across the holder, which would have shown an open.

Something I always teach my trainees, NEVER think up reasons not to check something.

[panomaniac]

Great stories!
The fuse reminds me of a story from my youth. Not electronic, electric.

At the art gallery were I worked they asked me to replace some of the burnt out halogen bulbs all around, high up.
First one. Climbed up the ladder, popped in a new lamp and - nothng. Climbed down, got another - nothing. Went to get my meter. All lamps checked OK! ??? Climbed up the ladder and checked voltage. 220V (Europe).

Now I was confused. All the lamps tested good with the ohm meter. The fixure had the proper 220V. Why didn't it work? You've probably guessed by now. (No one in the gallery could).

Bad, slightly corroded contacts in the light fixture. They would not pass enough current to light the 300W lamp, but no trouble passing enough current to "light" the 1M ohm meter.

I don't remember how I figured it out, but it took me some time and a lot of head scratching.
Learned a very good lesson that day! Current and voltage are not the same thing.

[Conrad Hoffman]

Worst thing I ever fixed was a GR Impedance Comparator. It "worked" except that there was a phase shift that prevented correct calibration. I spent 40 hours probing and measuring things with no success. Even changed a few parts that I couldn't prove were bad- a sinful method in my book. No joy. Finally I found a hairline crack in a solder joint where a signal was fed to the grid of a tube. Being high impedance, the ac signal was getting through ok, but the capacitance across the crack introduced just enough phase shift to cause a problem. One touch of the soldering iron and it worked perfectly.

Normally old GR stuff is very reliable, but another fun repair was an old GR signal generator with excessive distortion. It just should have been cleaner. I finally determined that the coupling cap to the output amp had a tiny amount of leakage. On a cap bridge it looked fine in every regard, but it had a few hundred kohms of DC resistance. This biased the (tube) output stage away from where it should have been, increasing the distortion. Replacing the black plastic paper cap (an infamous Black Beauty perhaps) with a new film cap made it perfect again.

[Geek]

Service call to a client's house for a dead TV.

Spent about four hours on it. Dead power supply, dead everything and I couldn't make it work.

The homeowners were dumbfounded as I was, because it had worked the previous night.

They were going out and the babysitter had arrived. They explained to her there would be no TV that night.

She said "Oh!" and flipped the switch on the wall and the TV came on perfectly.

The homeowners and I, totally embarassed, agreed to never speak of this again and I didn't charge for the service call

[Steve Dunlap]

The strangest error in a piece of equipment I have encountered was a shorted resistor. It was suppose to be 47 ohm but measured as a dead short. It still measured that way out of circuit. Replacing it fixed the problem. I kept the resistor as a souvenir.

Perhaps the dumbest failure I have seen was a Japanese receiver that was dead. Would not power up at all. Everything checked good. It turned out to be a burned out lamp in the power switch that caused the protection circuit to see a fault in current draw and refuse to power the unit up. After changing the lamp, the unit worked fine.