Most weird-assed Elect. problems I'll probably not see again ??

[CJay]

Here is a computer related one.

The Amiga SCA (computer virus).  This one caught me off guard when I was young.  Most likely the first Amiga Virus, 1987, if not, the first one to gain notoriety...

Good grief, there's a blast from the past. I remember one of the UK Amiga magazines 'gave away' a copy of Lamer to everyone who bought their magazine and used the coverdisk, it might have been Amiga Format magazine I think?

[SeanB]

Magnetic fields and an effective shorted turn.... Lot of current flowing in the conduit walls.

All right.  I think I see.  It's not really the conduits themselves, it the multiple holes in the cabinet?  You could effectively "fold" the holes together and end up with two interlocked rings, one with the AC flow and the other with the induced current, have I got that right?

Yes, you get conduit entries that are insulated so you need a separate grounding wire in the conduit ( or use plastic) to prevent this, though you also need to slit the cabinet between the conduits to leave a gap as well.  Only done where you cannot fit a bigger conduit to get them all in one.

[BrianHG]

Here is a computer related one.

The Amiga SCA (computer virus).  This one caught me off guard when I was young.  Most likely the first Amiga Virus, 1987, if not, the first one to gain notoriety...

Good grief, there's a blast from the past. I remember one of the UK Amiga magazines 'gave away' a copy of Lamer to everyone who bought their magazine and used the coverdisk, it might have been Amiga Format magazine I think?

It spooked me because I never heard of or seen a computer virus at the time.  Yes, I also heard one of the Amiga format magazines who offered a free disk with the magazine had been infected with the virus.  For that to happen, someone in the chain at the magazines HQ must have been playing pirated games, otherwise they would have never had the virus in the first place.  Also, that random, 1 in 20 or 1 in 100 boots before the message appears may have never been seen by them yet.

Even worse, that old simple D-Ram controller in the Amiga which had no mem-test at power up meant, if you powered down your Amiga, it could retain the virus for a few seconds in some cases at next power-up, even without an active d-ram refresh.

[radar_macgyver]

I once had a motherboard blow one of its CPU voltage regulators.  Not just burn out; it literally exploded, half of it blown clear off, with a clear audible bang, after almost a year of reliable operation (and a week or two of flakyness).

I briefly worked at a company that designed server motherboards. We were advised during board bringup to keep our eyes away from any OSCON caps in case they were inserted the wrong way (these were prototype boards that did not go through any QC).

My entry: when diagnosing a Klystron transmitter which would start up OK, but in a few seconds its output pulse would get jittery and phase noise would get really bad. We were almost at the tube replacement stage when we noticed that the triggering signal for the transmitter (not RF, just differential TTL in a shielded twisted pair cable) was dangling a bit too close to the HV modulator (~10 kV, 500A pulses). Re-routing the cable and adding a ferrite clamp solved it. What still puzzles me is why it took several seconds for the failure to show up, and not immediately.

Another one: on our dual-band radar (S- and X-band combined onto a single feed antenna) I noticed the S-band (2.7 GHz) radar would pick up a weak signal (only a few dB above the noise floor) whenever the X-band (9.4 GHz) transmitter fired. I would have expected something the other way around if the X-band picked up harmonics of the S-band, but why would the X-band transmitter produce any energy at 2.7 GHz? It turned out to be EMI from the X-band's HV modulator that was physically close to the radar antenna. Adding some EMI gasket around the X-band enclosure door cured the problem. What was most strange was this problem only showed up years after the X-band radar was installed.

[German_EE]

One-way cable

Go to any audiophool outlet and you'll see one-way speaker cables for sale where all sorts of nasty things will happen if you connect them the other way around, but this was a 2m length of RF cable with a PL-259 on each end, connect the cable one way around and SWR was OK, connect it the other way around and it was about 2:1.

The cause was the plugs, one had been badly soldered and some of the solder had crept onto the center pin increasing the diameter. Over time the SO-239 socket that it was connected to had expanded to cope with this. So.....

Socket A (wide socket) Plug A (wide center pin with solder)___________________________Plug B (normal center pin) Socket B (normal socket)

Works OK

Socket A (wide socket) Plug B (normal center pin)___________________________Plug A (wide center pin with solder) Socket B (normal socket)

Intermittent contact on the left connection and a stressed socket on the right.

I ended up replacing the cable and the damaged socket, things were then fine.

[GlennSprigg]

Ok... here's another Factory.... (huge automated processing plant).
I had a Chief Engineer above me, (he was non electrical), who could not understand (or be told)
why the main 1,500 amp factory feed was grossly overheating. (Split into 2 parallel sets of cables).

It came down to SIMPLE Ohms Law... :-)
Two 800-amp cables, (each phase), piggy-backed together Lugs, to one side of the huge BusBars.
Unbolted, I calculated about .6 Ohm on one run, and about .3 Ohm on the other runs.  Ok....
Sounds like nothing, but ONE had TWICE the current flow as the other !!!!
(1000 amps in one, and 500 amps in the other). Due to a difference of just .3 Ohm...
They either 'get it' or they dont...

[GlennSprigg]

After several visits to a site and workshop investigations on a PC with repeat claims of random characters appearing in a rather comely lady's word processor documents we figured out that moving her copy holder closer to her made the problem disappear as her ample chest wasn't hitting the keyboard when she turned the page.

Sounds just like a common legend. But I'm under the impression that area is very sensitive so it would be hard to miss the accidental contact? Closely related to that was a funny story a friend of mine told me about the time in computer repair class when she accidentally knocked over a box of screws...

I don't know about 'ample breasts', but I was called out to Yatala Labour Prison in Adelaide late one night, to their main control room.
The MainFrame computer had locked up. (Swamped with thousands of calls from a P.C. it seems)...
Turns out someone had a BOOK that was resting on the 'Print-Screen' button on one of the P.C.'s.
The 'MainFrame' had a 'Hissy-Fit' with it's buffers, and locked out the whole system !!!!!! Sigh..

[SeanB]

Ok... here's another Factory.... (huge automated processing plant).
I had a Chief Engineer above me, (he was non electrical), who could not understand (or be told)
why the main 1,500 amp factory feed was grossly overheating. (Split into 2 parallel sets of cables).

It came down to SIMPLE Ohms Law... :-)
Two 800-amp cables, (each phase), piggy-backed together Lugs, to one side of the huge BusBars.
Unbolted, I calculated about .6 Ohm on one run, and about .3 Ohm on the other runs.  Ok....
Sounds like nothing, but ONE had TWICE the current flow as the other !!!!
(1000 amps in one, and 500 amps in the other). Due to a difference of just .3 Ohm...
They either 'get it' or they dont...

Which connection was not crimped properly then? Did the lug glow or not.

[Eka]

After several visits to a site and workshop investigations on a PC with repeat claims of random characters appearing in a rather comely lady's word processor documents we figured out that moving her copy holder closer to her made the problem disappear as her ample chest wasn't hitting the keyboard when she turned the page.

At one of my former work places we had an amply endowed lady that sometimes accidentally activated the 10 minute count down timer for light shutoff when she reached for the paper stored on the shelf above it. They moved the paper, and maintenance added drawer handles on either side of the button that activated the timer so it wasn't possible to brush against it and activate it.

The reason there was a timer is it took about 10 minutes to walk to the normal exit door from where the light shutoff was. Also the system was sequenced so it didn't cut off such a huge load all at once. Once activated, it took 2 minutes to turn off all the banks of lights, and you could't stop it in mid process. Light turn on also took 2 minutes. They originally had it all turned on and off all at once, but if it and two of their large electric kilns all switched off at the same time it would cause an over voltage event at the plant's substation, and the plant's substation would kick out. With all the machinery, etc in the plant shutting off all at once it would also trip the area substation. Power would then go out for about 1/4th of the industrial area it was in. The electric kilns were also rewired so they had banks of heating elements that were never all turned on or off at the same time.

[Nominal Animal]

Before the turn of the century, I had a short IT support gig at an university.  After getting one classroom rewired, the uni network staff noted that both the panel-to-switch and socket-to-computer Ethernet cables were of the crossover type, and wanted to fix that, by swapping both to straight/normal ones.  It was fine otherwise, except that the panel-to-socket connections were mislabeled, and I ended up breaking three or four Ethernet cards on PowerMac 7200's.  It is a triple WTF: first, the mislabeling from panel-to-wall socket cabling; that was just pure evil. Second, that using a crossover cable instead of a straight one, or vice versa, could break the network card on PowerMacs 7200s. Third, I didn't realize what was happening until I broke the third one (or might even have been the fourth).

[duak]

Sean & bsudbrink: about the 3 phase wires in separate conduits.  It seems to me that this situation is a ground loop - the bane of low level signal engineering.   I'm guessing from the size of the conductors that the phase current is a few hundred amps?  I wonder if the NEC covers situations like this.  Surely others have encountered induced currents in metaliic conduits or raceways.  I'd bet there's be a measureable voltage on the conduits.

BTW, I understand that hospitals require all equipment attached to a patient be powered from one outlet.  Don't know if it's true though.  Makes sense because if, say, an ECG monitor probe set is a voltage or two above the motorized reclining bed there could be fault currents passing thru the patient. I think I've seen special medical receptacles that are ungrounded.

Cheers,

[janoc]

After several visits to a site and workshop investigations on a PC with repeat claims of random characters appearing in a rather comely lady's word processor documents we figured out that moving her copy holder closer to her made the problem disappear as her ample chest wasn't hitting the keyboard when she turned the page.

Sounds just like a common legend. But I'm under the impression that area is very sensitive so it would be hard to miss the accidental contact? Closely related to that was a funny story a friend of mine told me about the time in computer repair class when she accidentally knocked over a box of screws...

I don't know about 'ample breasts', but I was called out to Yatala Labour Prison in Adelaide late one night, to their main control room.
The MainFrame computer had locked up. (Swamped with thousands of calls from a P.C. it seems)...
Turns out someone had a BOOK that was resting on the 'Print-Screen' button on one of the P.C.'s.
The 'MainFrame' had a 'Hissy-Fit' with it's buffers, and locked out the whole system !!!!!! Sigh..

I remember a similar problem from the early 90s from dad's workplace (where the young me was moonlighting as an unpaid computer tech/support here and there). They had an SCO Unix server there, with an UPS for an orderly shutdown in case of a power failure. Both the UPS and the server were in tower cases, sitting on the floor under the desk in the office.

At some point the machine started to turn instantly off, losing power (and some data in the process). That was weird, the UPS was there meant to prevent exactly that and it didn't work? After a lot of testing of both the computer and the UPS everything seemed to be OK. Real headscratcher ...

This went on for a while, with several crashes a week. Until one day someone saw one of the office workers sit down at that desk to do some paperwork - and hit the big button on the front panel of the UPS with their knee in the process. POOF, server down. The protruding button on the UPS was the main shut-off switch for the load (really oddball design, IMO, don't recall the brand of the UPS anymore though). So the UPS remained up, beeping furiously about an apparent loss of power and the server was turned off instantly ... The result was lots of red faces and  a makeshift safety cover out of a piece of acrylic taped over that button. No problems since.

[dmills]

See the story behind the term "Molly Guard" for an older example of the same basic problem.

Regards, Dan.

[Gregg]

Sean & bsudbrink: about the 3 phase wires in separate conduits.  It seems to me that this situation is a ground loop - the bane of low level signal engineering.   I'm guessing from the size of the conductors that the phase current is a few hundred amps?  I wonder if the NEC covers situations like this.  Surely others have encountered induced currents in metaliic conduits or raceways.  I'd bet there's be a measureable voltage on the conduits.

BTW, I understand that hospitals require all equipment attached to a patient be powered from one outlet.  Don't know if it's true though.  Makes sense because if, say, an ECG monitor probe set is a voltage or two above the motorized reclining bed there could be fault currents passing thru the patient. I think I've seen special medical receptacles that are ungrounded.

Cheers,

Since I took the picture of the same phase conductors running through separate conduits, I probably should have explained the situation a little better.
The picture taken in 2005 is of a 480 volt transfer switch, the conductors are 350mcm copper THHN and the measured load was between 740 amps and 800 amps depending on how much HVAC was on line.  This was a small telco/data facility in Silicon Valley; which is still in business, so I won’t mention any names.
NEC article 300.20 strictly prohibits wiring parallel runs of conductors in this manner to eliminate the induced heating of the conduits, raceways etc.  310.4 refers to paralleling large conductors in higher power installations.

[jmelson]

Three phase power but each phase routed through a different metallic conduit.  Things were getting warm but the client didn't want to spend the money or down time to fix this problem.   

Oh NO!  Every industrial electrician knows you have to run ALL phases (plus neutral, if it exists) in each conduit.  Otherwise, you get induced currents in the conduits, which can be very substantial.

So, if you need to run 4 conductors for each phase, then you run 4 conduits, and one conductor of each phase ABC and neutral run in each conduit.

Jon

[NiHaoMike]

Oh NO!  Every industrial electrician knows you have to run ALL phases (plus neutral, if it exists) in each conduit.  Otherwise, you get induced currents in the conduits, which can be very substantial.

So, if you need to run 4 conductors for each phase, then you run 4 conduits, and one conductor of each phase ABC and neutral run in each conduit.

Jon

I presume that dedicating a conduit per phase is OK with plastic conduit? Or is plastic conduit not allowed/heavily restricted for such high power levels?

[basinstreetdesign]

Back in 1997 I was working for a small company run by two shysters.  I was planning to quit but one of the things they did to try to convince me to stay on was to give me a "company car" to use.  All I had to do was put gas in it.  The more I drove it the more I found what a turkey this boat was.  The idle on the carburetor was adjusted way too high so that just shifting into drive and NOT giving it any gas would make it pick up speed so that it would get itself into 2nd gear.  The brakes took a beating.

There were electrical problems, too.  The radio would change stations at random times and whenever I signaled left, the dome light would flash.      I'm sure that there was some kind of corrosion issue or lifted common ground connection or some such.  After a while I found that the battery was so weak that I could not start the car twice in close succession.    When I griped to the owners, they just laughed.  I was gone in about a month but they refused to stop paying me (at 50% of normal) to "keep me available". 

[BradC]

BTW, I understand that hospitals require all equipment attached to a patient be powered from one outlet.

Don't know about hospitals, but I played a gig at a pub once where the PA was plugged in on one side of the stage and my amp on the other. After copping a hundred volts (or so) between the guitar strings and my Mic, I now check sockets for a proper earth during set up and *everything* gets plugged into the same socket.

It turns out your lips are really quite sensitive to that sort of voltage, and the rest of the band thought the slowly decaying echo of "fark...fark...fark..." over the PA was quite comical at the time.

[Berni]

Ah yes disconnected grounds.

I had a case where went to the desk of a programmer to help debug the product that I built the hardware for. Tho I had noticed that touching the PCB would sometimes make it do weird things and cause glitches on the display. So I bring a scope over to see what's going on, I start looking for a good place to put my scopes ground clip while holding it in my hand and upon picking up the PCB I got a shock and threw the product on the table. I looked surprised for a second to what happened, the highest voltage in there was a 30V LCD backlight supply switcher, so how did I just get shocked. Then i found that I get shocked by touching the shield of the microUSB cable that was used to power the product.

Turned out there was 110V between his PCs ground and the earthed ground clip of the scope. Turned out that the power strip all of his stuff was plugged into was stuck into a red UPS backed outlet that had no earth.

Upon explaining it to the programmer who sat at that desk, he did not seam at all concerned by this and even argued that this is normal because it's a special UPS outlet. They did fix it some time later.

[TheWelly888]

BTW, I understand that hospitals require all equipment attached to a patient be powered from one outlet.  Don't know if it's true though.  Makes sense because if, say, an ECG monitor probe set is a voltage or two above the motorized reclining bed there could be fault currents passing thru the patient. I think I've seen special medical receptacles that are ungrounded.

Not quite from one outlet but from one phase; in the hospital I work in, each ICU bed has multiple sockets fed from one phase. I have never seen ungrounded medical receptacles ever!

[bsudbrink]

I presume that dedicating a conduit per phase is OK with plastic conduit?

If I understand correctly now (and I think I do), just plastic conduit won't help if you have different phases entering a conductive box through different holes.  You will have induced current (and presumably heating) in the conductive box.  The solution is to make the separate holes into one hole by connecting them with thin slits.

[jmelson]

So, if you need to run 4 conductors for each phase, then you run 4 conduits, and one conductor of each phase ABC and neutral run in each conduit.

I presume that dedicating a conduit per phase is OK with plastic conduit? Or is plastic conduit not allowed/heavily restricted for such high power levels?

Well, I can't imagine running 4160 V in plastic conduit.  And, running thousands of Amps can lead to at least humming of nearby metals, and possible failure over time due to vibration.  Best to balance and thereby null out the fields by not grouping the phase wires.

Jon

[duak]

I mentioned earlier that I had seen an ungrounded hospital receptacle.  I rummaged thru my collection and sure enough, I had one.  In fact, it was an Isolated Ground receptacle where the grounding pin is not connected to the frame but has a screw terminal to allow for a special insulated ground wire to a grounding rod separate from the equipment grounding wire or conduit.  I sort of remember looking it up and thinking it might come in handy to build an isolating transformer to safely work with offline power supplies.  Apparently, these receptacles have orange triangles near the ground pins.  In my case, the whole receptacle is orange.

I looked up sheath current which is the term for currents induced in conduits or other shields.  I had initially thought it was something to do with three phase circuits, and it sort of is but it really occurs any time the shield is ferromagnetic and there is an unbalanced current flowing thru the conductors within it.  In some cases, the conduit may get hot enough to damage or at least reduce the ampacity of the conductors within it.

Learn something new everyday, or in my case, re-learn.

[calexanian]

As far as the lights only working when another light is turned on it jogged a memory.

Here in the states we have split phase for our 240/120 systems and it was allowed to shrink the size of the neutral when 2 120 loads would essentially be in series across the 240 where nothing but the imbalance current would be carried by the neutral. If you lost the neutral or it went high impedance thats when the funhouse lights started. turn the washing machine on and some lights got brighter, and your washing machine motor burns out due to low voltage. things like that. if you totally loose the neutral then the lights would only work in that scenario when both switches are turned on.

[Brumby]

Where I am in NSW, you are allowed to lock your outside meter box/fuse box - but you can only use a padlock supplied by the electricity company.  They give you a key for your padlock while they hold onto a master key.

This stops nuisance attacks by people flipping switches on you while still allowing access to those who require it.