Let's talk about testing capacitors

[Helmholtz]

Wow - those must have been very expensive. It certainly is possible to tweak the chemistry and process to get very long life. Those things must have been from before etched foil construction, even.

I'm sure we didn't pay more than 2€ (high volume).
They have etched anode foils and were used in an SMPS (boost converter for power factor correction).

I've had a few die abruptly. In my previous job I designed power supplies for a while, so I tended to be who got yelled at when something melted down in a customer's office, for whatever reason. It's unusual, but it does happen. Design for reliability is a demanding task, and all you have to go on is the statistics.

Might have been a production defect.
Significantly increased leakage or ESR can cause thermal runaway.

For a while I was in quality management, being responsible for a group of EEs.
We did things like product failure analysis (customer claims), component and supplier approvals, custom component specifications, derating guidelines and product lifetime calculations (using fit numbers and other data).
We actually life-tested hundreds of our products at elevated ambient temperature over years and analysed failure causes.

[Mick Bailey]

I've seen a fair number of exploded caps in amps - sometimes catastrophic failure when an amp that's been unused for many years has been switched on, and others where I've suspected an unnoticed pre-failure condition. A typical example is a JCM900 where one cap had ruptured explosively, but the adjacent one was bulging and had leaked, causing chassis corrosion. It must have been on its way for a good while. The owner said until it went bang he hadn't noticed anything wrong.

After gutting many old (mainly 60s) can caps for re-stuffing, here's my generalised observations;

1. They're obviously much larger for a given capacitance.
2. Cans were made of a heavier gauge metal.
3. The foil and absorbent material are quite thick.
4. Sometimes the can is damaged internally, causing it to become brittle. This is possibly caused by reaction with the electrolyte.
5. The lip is often work-hardened during the rolling process.
6. Older end caps can be made of paxolin with a rubber seal, which can be hardened and brittle.
7. Many have no signs of any provision for venting or controlled rupture of the casing as with modern cap.

Some of these contribute to a much greater explosive force during rupture. The larger volume means more discharged material from the can. By far the worst amp I've ever worked on was an Ampeg where someone had fitted an old used can cap for a repair. The acrid gel and the sheer amount of shredded fibre stuck to everything was remarkable. As there are patents for asbestos fibre construction in old electrolytics caps, I had the owner get an asbestos test before I'd work on it. It was like someone had covered the inside with glue and emptied a feather pillow into it.

[Helmholtz]

Yes, ecaps can blow out or explode (if they don't have a safety vent) but hardly without showing excessive leakage or ESR before,
Both leakage and ESR can cause internal heating resulting in gas production and high internal pressure.
As leakage increases with temperature this is a runaway effect.

[g1]

One thing sometimes overlooked about comparing life of old caps compared to modern is the environment they are living in.
Lots more components fitted tightly together, often on both side of the board means you are sort of using the board as a heatsink. I was quite shocked to see a newish Mesa amp that had lots of electrolytics with shrunken wrappers (due to heat). New replacements felt just as hot with no faults present. It was just how hot the board was being run.

[R.G.]

One thing sometimes overlooked about comparing life of old caps compared to modern is the environment they are living in.
Lots more components fitted tightly together, often on both side of the board means you are sort of using the board as a heatsink. I was quite shocked to see a newish Mesa amp that had lots of electrolytics with shrunken wrappers (due to heat). New replacements felt just as hot with no faults present. It was just how hot the board was being run.

There is no exception to the need for good thermal management. It doesn't matter how many components how close if you have >calculated< the thermal load and >provided for< getting rid of the heat at a reasonable temperature. If the components show heat damage, someone didn't do their homework and just relied on "that's the way we have always done it, and it worked fine." With today's tools, like thermal cameras, at least you need to test what you did and fix the hot spots.

Of course really capable and malicious engineering can run the temps high enough to raise the failure rate to get the sear-out time closer to 1.1 times the warranty period.

I'm sure people are getting tired of me using this illustration, but a T-1 incandescent bulb can be rated for 5V and 15ma. That's 75mW. The bulb makes it so hard to get rid of that heat that the filament heats up to 3600 to 4600F before it can get into thermal equilibrium. The power generated matters less than how the heat gets away from the hot components. Temperature always rises until it can force enough heat to escape to cause heat escaping to equal heat generated.

[The Dude]

One of my pet peeves is designers putting electrolytic caps right next to heat sinks, which sadly seems commonplace these days.

[R.G.]

One of my pet peeves is designers putting electrolytic caps right next to heat sinks, which sadly seems commonplace these days.

Yeah. Engineering is actually bigger than designing. It takes a while for a newbie engineer to automatically think of things like
- How will I mount this in the box/on the PCB? Will it actually fit, or will it crowd out other things?
- How hot will this get? How hot will the part right next to this part get? Do I need fans, heat sinks, whatever?
- Have I checked the datasheet on the part for voltage/current/speed/etc. ratings?
- How much does this thing cost?
- Can I actually but these things? Do they really exist and how long will it take to get them here (a) for prototypes and (b) in quantity for manufacturing? What's the lead time?
- What's the failure rate like?
- How much abuse will the part take? What's its overload capacity?
and on and on and on. The more time you spend designing, the more these come naturally.
The degree, as such, doesn't matter. That's only useful for demonstrating that you have been exposed to the theoretical underpinning of the discipline. Education goes on long after the degree if you're any good.

[Mick Bailey]

Some Blackstar amps have electrolytics very close to the PI MOSFET tabs (no heatsinks) and the caps run very hot and consequently have a high failure rate. I fit higher quality replacements and move them to the underside of the board on the opposite side and they don't have any further issues (at least with the caps.....). Seems to me they designed the amp with no consideration of thermal aspects.

[R.G.]

It might be possible to hack amps set up like that. Inserting a vertical sheet metal plate between the hot things and the caps would help, as long as it doesn't block convection airflow