Ultimate Screw-ups
Comment
WKetel
Capacitors do have quite a few specifications, and sometimes it is vital to ...
David Ashton
I reckon connectors for power are one comonent that is difficult to overspec. I ...
Pop goes the capacitor!
David Ashton
9/13/2010 5:52 PM EDT
In the early 1990s I ran my own business leasing terminal systems to travel agents in Zimbabwe. I got the terminals second-hand from British Airways (whose mainframe Air Zimbabwe used for reservations). BA renewed their terminals every 5 years, so I was able to get their old stock pretty cheaply. But there was a reason BA replaced their terminals every 5 years: They were beginning to require regular maintenance.
My second-hand terminals were dumb beasts, big and ugly and heavy, but they did the job. They were based on the old Z-80 microprocessor and had CRT monitors and big iron transformers and linear power supplies. But that made them easy to maintain, and I had full service data for them.
The power supplies probably gave me the greatest number of problems. They were linear with 3-terminal regulator ICs – the good old LM323K for the +5V for the logic, an LM317K for the +24V monitor supply, both on a big finned heatsink, and 7812/7912s for the +/- 12V. A big heavy toroidal transformer supplied them. (I later made a great lab power supply out of bits from an old terminal!)
But although they were simple, they were only just sufficient for the job. They got very hot and failed occasionally. But the weakest point was the connectors. Both the +5V and the +24V supplies reached their respective boards through only one pin on a totally inadequate connector. A bad connector would go high-resistance and get hot and go even higher resistance, sometimes eventually melting the connector body. Fortunately I had access to new connectors, and once installed they usually gave me no more trouble for a year or two.
The most interesting problem I had, though not the most difficult to diagnose, had to do with the vertical output circuit. These were magnetic deflection CRT displays with a yoke around the neck of the tube, and there was a large non-polarized 6.8 µF electrolytic capacitor in the vertical output circuit. They were about 2 cm in diameter and 5 cm long (for our American friends that’s about 3/4 x 2 inches).
When the first one went wrong on me (it failed open circuit giving a horizontal line across the middle of the screen), I scoured the local electronics suppliers and eventually found some 10 µF non-polarised capacitors with a suitable voltage rating. They were, however, only about 1 cm in diameter and 2 cm long (3/4 x 3/8 inches). I marveled at how component design had progressed since my terminals were built. I installed one of the capacitors in the faulty terminal, ran it up and it was back to normal. I left it for a few minutes and all seemed well. I put the terminal back into my spares pool. Some time later it was used to replace a faulty terminal at one of my favourite travel agencies.
A few days later I got a call from the agent.
“Dave, you know that terminal you replaced last week??”
I got an unaccountable sinking feeling in my stomach.
“Yes…???”
“Well, it’s exploded!”
I got out of the agent that it had made a loud pop and sent up copious smoke signals, causing no small amount of alarm to the poor girl who was operating it. The “pop” noise gave me a clue as to what had happened. My new, marvelously small capacitor had vented itself to the outside world. And when I opened the terminal, that is exactly what had happened, and the inside of my terminal was covered in bits of aluminium foil and fluffy dielectric.
I apologized profusely to the agent, told him what had happened and gave him a month’s free rental for the pain and suffering caused thereby. I also learned a valuable lesson about the AC series current rating in capacitors, and how component design had NOT progressed as far as I thought it had. I managed to find some 3.3 µF polyester capacitors of which two in parallel worked as well as the original non-polarised electrolytic. And you can be sure that I gave that terminal a one-week soak test before I let it anywhere near one of my agents again!
David Ashton is an Australian engineer. To this day he says he has a name for taking ownership of problems, which he says he doesn't regard as particularly praiseworthy, it's just how he is.
|
|
|
|
|
Mark Wehrmeister
9/13/2010 11:58 PM EDT
Exploding capacitors, that brings back memories! I had an electronics teacher in high school that wanted to get in good with the students (or scare the heck out of us) so on the first day of class he reversed the polarity on a capacitor connected to a breadboard and set it on a table in the middle of the room. A few minutes later, Pop! Needless to say all of us figured out much faster ways to make capacitors explode and we created our own "fireworks" shows for months after that.
Sign in to Reply
David Ashton
9/14/2010 8:03 PM EDT
I have had a few over the years, but the one described was, let's say, the most public one! I find the results are most spectacular when the cap has a bit of time to build up a good pressure inside, as this one did. These days they tend to put weak points on the top of the can for safety, which diminishes the effect somewhat....spoilsports.....
Sign in to Reply
Frank Eory
9/16/2010 5:19 PM EDT
I remember that back in school, some students struggled with the complex number representation of AC circuit quantities, and the idea that the current in a capacitor was the "imaginary part" of the loop current. There is nothing like an exploding capacitor to convince a student that there is nothing imaginary about that current!
Sign in to Reply
John.Donovan
9/17/2010 3:34 PM EDT
In college I worked in the capacitor lab for a microwave company. They made their own caps and my job was to test them for Q, breakdown voltage, etc. We'd put the caps in a lucite box and crank up the voltage until the puppies exploded. The small ones were so ear shattering that we never had the nerve to try it on a large electrolytic!
Sign in to Reply
David Ashton
9/17/2010 6:19 PM EDT
That would have been almost as good as working for Mythbusters!
Sign in to Reply
Donscoggin
9/17/2010 4:40 PM EDT
Back in the late 8os' at an Itty Bitty Machine companyon a Mainframe type product , we had large 450k ufarad DC output caps exploding confetti. After months of investigating by a product engineer manager who refused to accept that brown stuff happens; he found thru X rays etc that ants were crawling into into the caps during manufacture in south carolina to have a drink of ethylene glycol and when positioned beween plates correctly caused a short. If the caps had an extra vent plug inside (caused by overzealous manual insertion of the first vent plug )then the secondary would block the initial vent causing the eventual explosion. Talk about root cause analysis!!
Sign in to Reply
David Ashton
9/17/2010 6:18 PM EDT
Ants...I once had ants getting into a 25-way RS232 plug on a modem and causing intermittent shorts. I think it was something to do with the formic acid they make. I ws going to write it up for these columns but I can't remember enough of the details. You should write the above one up in more detail, it's a good story.
Sign in to Reply
Nate Ocean
9/17/2010 5:11 PM EDT
But what's to be said of today's poorly designed power supplies? I routinely receive from friends dead LCD TVs or Monitors barely a year or two old. Never has there been a digital problem or an actually LCD panel issue. Chips and panels are made in billion-dollar factories and they make solid stuff.
Inevitably the problem will be either in the back-light inverter or in the main switcher power supply. Typically you need little more than a pair of eyes to look for the capacitors with bulging tops. These are always on the low voltage side (+5V, +12V, +18V sections). There they are routinely abused because of an inadequately designed switcher and often these are standard grade caps, instead of the necessary low-ESR grade components.
Occasionally just for grins (or because I'm an engineer) I'll measure the defective parts. Typically the measured capacitance is about but a fraction of the original spec'd value. In a few cases the electrolyte had dried up and the part was essentially an open circuit.
Back-light inverters are also poorly designed. If the capacitors don't go out (as described above) then it's likely the inverter switching transistors. My theory is that the back-light design has such poor margins, that as the bulbs age they require increasing voltage to run. This over-taxes the inverter circuitry which soon fails.
It's ironic that on the shelves of Goodwill you can find a *dead* LCD monitors or TVs that are only a few years old--but shot. Sitting right next to them are quality CRT-based monitors--still working--but now obsolete.
Monitors that I've repaired include Dell, HP, and Gateway (my current monitor) and Polaroid, Walmart, and a Samgsung TV (my current TV). The guts are from various OEMs, Benq is inside the HP as I recall. In all cases the sub-assemblies are from China, as are the defective capacitors.
The Internet is littered with web sites about bad capacitors and how to make repairs.
Sign in to Reply
David Ashton
9/17/2010 6:23 PM EDT
A later line of terminals that were made by the same company that made mine above had exactly that problem. It was a sod to find the first couple of times but after that we'd just replace all the low side caps and the PSU was good for another couple of years. ASTEC PSUs too, from what I remember, they ought to have known better. It's great if they bulge, dead giveaway, but they don't always.
Sign in to Reply
Erickk
9/17/2010 6:55 PM EDT
About 25 years ago, I enocuntered an application note in the specification sheet for a multi-output switching power supply from a Japanese manufacturer. In factured english, the note said that the life of the product could be extended many times over by periodic replacement of all the filter capacitors. Despite this note, the power supplies were actually extremenly reliable, perhaps becuase we operated them at only about 75% of rated wattage. But it does point out that electrolytic capacitors have always been a known weak link in electronics.
Sign in to Reply
W1PK
9/20/2010 9:28 AM EDT
That happened to my first ham transmitter. I bought a second-hand AT-1 that had been sitting on a shelf for some time. When I powered it up for the first time, there was a sizzle from inside, then all four B+ capacitors blew their seals and sprayed electrolyte all over the inside. I washed it out with a garden hose and left it in the sun to dry for a couple of days, before doing anything else. A guy at the local radio club gave a talk last year on how to bring up old tube gear with electrolytics inside. Use a variac, and bring up the voltage in small steps over a period of a day or so, to re-form the caps. Or just replace them.
I've gotten away from designing power supplies with electrolytics, tantalum or otherwise, in the last couple of years, though, since high-frequency switching supply ICs have become easy to get. I'm using almost all ceramics, and not Z5U types, either.
Sign in to Reply
WA7TUV
9/27/2010 4:15 PM EDT
I used to be a TV transmitter maintenance tech back in the days before remote control. I was once working on a 285 volt 1 amp power supply with tube regulators and had the supply upside down on the workbench. I leaned over to the right to grab my meter when the main electrolytic capacitor vented electrolyte steam that left colored speckles on our 12-foot ceiling! I had been looking down at the bottom of the capacitor just two seconds before. Timing is everything and I found a new respect for the bottom of those capacitors.
Sign in to Reply
camey
9/20/2010 3:47 PM EDT
In my early days of building computer systems a friend brought around an S-100 motherboard for me to test. After we had connected the +/-5 and +/-12 and powered it up I noticed the electrolytics down the side were arranged in random orientation. I commented that it was unusual given the otherwise symmetric nature of the backplane. My friend looked at me and asked "They go in a particular way?"
Before any of us could hit of the off switch one exploded and we were left rather dazed standing in a cloud of mylar and dust. I've seen many other larger caps explode since but nothing quite as impressive as that little 100uF firecracker aka directional electrolytic.
Sign in to Reply
Duane Benson
9/23/2010 12:24 PM EDT
Electrolytic capacitors have always worried me when used close to their published specs. A good engineer will always give plenty of headroom, but when the cost constraints come in, that headroom is often one of the first targets. Some components in some designs seem to be able to deal with minimal headroom better than others, but I don't think that electrolytics are in that set.
These days with RoHS I have even more concern. As an assembly house, we see a lot of these and while most RoHS components are in fact truly RoHS compatible, the metal can caps seem to be closest to the hairy edge. One big electrolytic with lots of space around it may reflow quite well. However, put a couple close together and their combined thermal mass often prevents the inside connections from soldering thoroughly. Slow the oven down a bit to fully solder it all and the cans are likely to bulge.
If they are that close to maximum thermal tolerance, are they typically slightly damaged even if they don't show any external signs? Are they now pre-weakened and ready to pop if anything else is on the marginal edge?
Sign in to Reply
ylshih
10/1/2010 1:29 AM EDT
In reference to your side comment about inadequate connector pins for the power lines, the number of marginally spec'd parts in even modern designs is discouraging. I've just had to repair a dishwasher control board (personal fix-it) where a relay controlling the heating element runs through a totally inadequate PCB trace and will burn out over a few years. The failure mode is well known and documented on appliance repair websites. The discouraging news is that this fault seems to have been in this product for at least 10 years and there seems to be no evidence that the manufacturer has ever issued an updated control board with a thicker/wider trace for this power run, a simple no-cost solution. They seem content to charge $200 for replacement control boards - I replaced the burned out run with heavy gauge copper braid and reinstalled the board.
Sign in to Reply
David Ashton
10/3/2010 7:29 PM EDT
I reckon connectors for power are one comonent that is difficult to overspec. I had a similar experience to yours, with a washing machine. 2 months after the warranty ran out it stopped spinning. I found a relay on the control board with dark black lines radiating from under it on the PC board. Trouble was, the whole PC board was encased in a clear resin type stuff. So with much cursing I forked out $180 for a new board. Then a week after I'd got it going, I happened to see a recall notice for that model. I did get a refund for my $180 but they wouldn't give me anything for my labour! In this case I think it was inadequate soldering on the relay pins - a common problem with automatic soldering processes.
Sign in to Reply
WKetel
11/24/2010 8:40 PM EST
Capacitors do have quite a few specifications, and sometimes it is vital to consider all of them, or at least more than are usually considered. AC current and power dissipation were overlooked in this tale of woe. Leakage versus temperature is a good "gotcha" parameter, a bit more subtle than the "life versus temperature" rating. One very interesting rating is ESR,(effective series Resistance), which does cause problems with capacitors used on power buses, but can also cause problems in high-frequency capacitive impedance matching circuits.
It winds up being very important to understand all of the parameters that are important when specifying capacitors, and also to make sure that purchasing department decisions do not substitute 100pF disc capacitors for the 100pF Silver Mica caps originally specified. The results from such a substitution, while not explosive, are impressive in the lack of correct circuit functionality.
Sign in to Reply