Self healing component failures

Electrolytic caps immediately come to mind as the most likely suspect. I have sure seen this sort of thing before. This can happen with either aluminum or tantalum caps.

There are other things like bad solder joints and internal wire bonds that can be intermittent, but they usually are not predictable.

Jon

RoHS solder joints. Hands down.

ent

Are these failures occuring in development circuits you've actively been working with, or in returned product? Because product that's been out in the field can have problems like dendritic growth shorting adjacent pins (that can open up if enough current goes through it), parallel circuit paths caused by crud that becomes conductive under humidity, etc.

Are the failures microphonic ( does Fonzing* the thing make them start or stop; is vibration a factor? )

-- Les Cargill

Not really enough information to make an informed comment. Virtually any component can have odd failure modes including intermittent ones, but I don't think I can come up with any examples that are consistently allied to any particular component type, and that are not temperature related. By far the most common time related failures these days, are down to bad electrolytics, but if you dig down, there is almost always a temperature angle to the failure. For instance, caps or resistors in the startup supply of a switcher, will often fail due to heat issues, but you don't notice because as long as the item remains powered, the cap or resistor does not cause a problem. Now remove the power completely, and try and restart it a few hours later. Chances are, it won't now come back on. But leave it a few days, and it might well restart the next time ...

Another possibility is bad joints, particularly with anything manufactured in the last 5 years, which will make use of lead-free solder. Bad joints in this hateful stuff, can give rise to the most bizarre intermittent symptoms, which can on occasion have apparent self-healing behaviour.

You need to tell us a bit more about the actual failures you are seeing, and what the actual equipment is.

Arfa

capacitors, especially the electrolytics.

Jamie

Almost always you find that there is a variable of which you weren't aware. Strangely sometimes in equipment with absolutely rock solid power supply regulation line voltage can affect it. Shouldn' but does. Operating conditons when it comes to thermal aren't always apparent. Air currents could be different. A piece of dust could blow off or fall out of a fan or heatsink. Then there are those %&##^&# software problems now.

However in rare cases it is actually a component failure. What transistor was in old Pioneer recievers that was always an intermittent problem ? 2SA828 or something like that ? I forgot now but it was so common people replaced them even if they had nothing to do with the current problem. And it was intermittent.

More common is an open terminal on a bipolar. Normally any of the terminals going open will result in no conduction, but in my experience when I could determine which it was usually the emitter or base. Open collectors have been extremely rare in what I have touched. What's more usually with an open base or emitter it is hard to tell in circuit which. Taking it out of circuit by desoldering has just changed everything due to the thermal shock. Sometimes even wiggling the leads does it, in that case, it's usually a defective case.

Tubes can be even more finicky. One of the biggest problems in tubes was gas, and believe me they had adequate vacuum pumps, what the problem is, is that they must be heated to get the gas that is disolved in the metal elements of the tube out completely before it's sealed up. The dark spot is a getter and any gases that boil out during use are supposed to be absorbed my that metal, I don't remember what it is offhand. But those gases floating around the envelope aren't always in the same spot appaerently or something. I know it doesn't sound right according to the physical laws of gases, but if the quantity is very very small who knows. A few molecules between elements with a high potential between them is not the same as a vacuum.

Transistors though, when they short and then unshort, they are the biggest problem. Across the whole junction, SOMETHING blew that short out. What's more there is no way in hell it didn't take part of the junction with it. The problem here is it would take alot of work with an electron microscope to even see something like that. Also, when that happens of course the ratings of the device are compromised and the thing is going to fail again. However if it keeps on cycling there is obviously little or no damage each time.

Perhaps it is sometimes just a slight innacuracy in how the semiconductor was formed. Perhaps one of the junctions decides to act like a tunnel diode from time to time. It's also slightly possible that magnetic fields could affect it. Poor analogy and a longshot, but then we are not talking about something that happens every day.

Speaking of diodes, what about high speed rectifiers that slow down. It happens, and it happens enough that is has made it to case histories (symcures) in TVs at least twice. In these the diode tests good, and would probably work fine on a 60 Hz sinewave, but at high speed it won't. Check with the ohmmeter and it has .5V drop, in the circuit it has over ten times that. Is it a hybrid diode that has two, one for the initial peak and the other to take up the slack after to protect the more delicate high speed diode ? Or is it a dual junction of some sort ? What else couild explain it ?

That is about all the far fetched shit I got at the moment. At this point all I can say is if you have a case of the gremlins and the thing has anything in it like data and clock lines, suspect a software problem first unless you KNOW it's something else. Like obvious power problems and such.

Like in the case of a Zener that would short when it warmed up. this was another symcure on a TV. It seems that once the Zener shorted it was no longer dissipating any power and thus cooled down and was no longer shorted. I don't ko what you know about TVs specifically, but when the horizontal output transistor was shorted you had to replace a Zener in the tuner. Ain't it great ? No sanity required.

Cazapitors:

That also applies to electrolytics and metalized film caps. If the hole is small enough, it will self-heal. Google for "self healing capacitors".

I guess resistors might also qualify. They drift when they get hot, but recover their original value when they cool down.

Same with most semiconductors. The only one's that won't return to their original condition when cooled are those that suffer from thermal runaway. Some geometries also suffer from latchup, where they form and SCR. Those will recover, but only after you remove power.

If it's none of these, then I suggest you have your device exorcized. Once the demons are gone, the device should operate more normally.

"Paul Hovnanian PITA "

Most intermittent faults behave like you describe if viewed over an INSUFFICIENT time scale.

Think about it a bit harder.

.... Phil

independent

Capacitors for X and Y rated mains purposes , mainly polypropylene, don't know about temp/time chracteristics

basically instant self healing, if they didn't they'd be a short across the line and thus explode or take out the fuse.

The two that come to mind are aluminium electrolytic capacitors (possibly) and surge tolerant resistors, which used to be carbon composition. I think surge tolerant R's tend to rise in R after an overload. No idea about how these behave with temperature.

Thinking further... batteries?

A temperature independent failure mode implies, to me, it's not a semiconductor and the cause is mechanical, like vibration. That could result in a cable's insulator being worn through as it passes a sharp object... and some squidgy plastic insulators seem to sort of self anneal after a while if squashed together.

Hi!

The simple fact of the matter is that any electronic component can become temperature sensitive. You may want to use a heat gun or "freeze spray" to try and isolate the fault more quickly.

If you don't have freeze spray, I'm not above suggesting that you could use your household freezer...though this comes with significant caveats, not the least of which is what condensation might do if it causes inappropriate electrical connections. And it should probably never be done with anything line powered. It also wouldn't be terribly precise, though it may "reset the clock" and let the device work normally for a while.

William

This jogged my memory a little. Back in the day, under bias conditions you could sometimes invert the surface of a bipolar transistor enough to latch it on. Unplugging it and baking it out would cure the problem for the moment. Letting it sit would also let it heal.

ent

Mechanical switches and relays are some frustrating ones. Relays meant for switching power circuits can fail to pass a low-level signal even when apparently closed. Many people don't know to look for dry- rated contacts (e.g., gold-plated) when specifying such parts.

Also, when building or repairing your own gear, you need to be careful when reusing parts that have previously been used to switch high currents or voltages. Carbonization and general dirtiness in an old power switch can make a nice humidity sensor when you reuse that power switch to drive a logic gate (use a pullup/pulldown resistor, duh!) or even the LED in an optocoupler (not so obvious). Meanwhile, a relay with gold-plated contacts may no longer have gold-plated contacts after it's been used to switch loads near its full rated power.

-- john

"William R. Walsh" wrote in news:bNXmr.14916$ snipped-for-privacy@newsfe01.iad:

helpful hint; when using freeze spray,it's better to spray a Q-tip and apply that to the component than to spray the component directly. you can CAUSE "problems" to appear with freeze spray.

Isn't that the point?

This is a further symptom of some weird problems my truck's electronic ignition module is having. It seems that when it begins to misbehave badly, if I park it for a week or so, the problems go away for a few days. Then they start to reappear. Temperature doesn't seem to have anything to do with it, as the failures are just as likely to occur on a cold start as after having driven on a warm day.

I don't think its due to lead free solder. The truck is 33 years old. I re-heated all of the solder joints I could get to (most of the module is potted). I added an external cap (on the theory that an internal cap inside the possting may have failed open). No difference. Cleaned all the connections and added a dedicated ground lead from the battery negative to the module (on the theory that body rust somethere might have taken out the original ground path). No change.

The symptoms are that every once in a while (about once a day) the ignition will just cut out. Every time a coast to the side of the road and start poking around with a voltmeter, all is well (12 volts is present where its supposed to be). I've replaced everything except for the module. Those go for about $350. If one can be found. The really funny thing is that: in order to test the coil output, I hooked up an old timing light (a direct connect type that triggers off the coil secondary). The presence of that strobe in the ignition circuit fixes whatever is wrong within a few seconds. And the fix lasts for a day or two. I don't have to leave the strobe connected. Once it starts, its OK without it. Otherwise, it just never starts unless I leave it parked for a week or two, which seems to be when the 'self healing' kicks in.

Since the problem has been narrowed down to that module, its pretty certain I'll just need a new one (something other than the $350 OEM part). I might try building my own (there's not much to one of these). But the failure mode has me puzzled and I'd like to understand what might be wrong here. Even if it has no bearing on the eventual fix.

Since I have enough vehicles at my disposal, I don't actually have to fix this any time soon. So I'm trying to solve the puzzle rather then just throw parts out.

I've been watching this go on for about 6 months now. Since its a redundant piece of equipment (one of my trucks), I'm more interested in understanding the failure mode than actually fixing it.