Doh' moment!!

[unknownparticle]

Have just been making a minor repair to a nicely made, home built multi-voltage power supply that I picked up from a UK vintage radio club auction some time ago.  The unit was made to substitute mains power at various preset voltages to supply older transistor radios. So it has 3v, 4.5v, 6v, 9v, 12v and 18v settings, plus a fully variable position.  It's based around a TO3 variable linear regulator, with switch selected presets to set the output voltages.  Quite a neat little unit, nicely built and very useful.
So, having used it for a number of years I decided to have a peek inside to make sure everything looked ok, good thing I did because one of the 2 4700 uf reservoir caps had started to puke itself!
So, checked the value and voltage and obtained a couple of nice Panasonic replacements.  Had to mod the mounting arrangement because the original, older caps were much larger.  Did all that so it looked ok to my OCD vision, turned on and all was working fine. Here is the doh' moment though, I decided to check the voltage across the caps and discovered to my horror it was 10 volts higher than the rating of my nice new caps
Even though the new caps were exactly the same rating as the old caps, at 25 volts!  So the original constructor had either thought, oh it'll be ok, or never noticed the error!  No wonder that original cap had started to puke!
So, a reminder for me, don't take anything for granted!
All my careful mounting will now have to come apart and new caps bought, bu66er!!!
Although, I do wonder how long those new caps would last running at 35 volts   No, stop it!

[George Edmonds]

HI
Mesure the transformer output AC voltage

Multiply it by 1.414 to get the DC voltage

Add 50% to the DC voltage result for a long capacitor life

G Edmonds

[unknownparticle]

My mistake was assuming the original constructor had done that. Anyway, the replacements will be 63 volt.  Could have been worse I suppose, the 25 volt ones could have ejected their internals all over the bench!   

[electr_peter]

With unique or older deigns it is best to triple check everything before finalizing repairs because you never know all aspects of circuitry. Otherwise one small repair may lead to bigger repairs.

Big caps is always a risk due because of their size - desoldering can be a challenge (80-100W soldering iron with big tip sometimes required), mounting is tricky due to weight and size, etc. And always treat them as fully charged, otherwise solder wire will be unexpectedly vaporised and your ears will ring for a while
I would add high value discharge resistors just in case (so that you can check circuit quickly after power off and not wait for some minutes).

[Wallace Gasiewicz]

Divide the AC volts by 1.414.
Or multiply by .707

[BrokenYugo]

I've learned to up the voltage rating considerably when dealing with stuff pre 1980 or so, between running the caps up around the limit and sometimes you find that those much smaller caps of equal specs run a bit warm.

[vk6zgo]

I've learned to up the voltage rating considerably when dealing with stuff pre 1980 or so, between running the caps up around the limit and sometimes you find that those much smaller caps of equal specs run a bit warm.

I got caught out on a black & white TV monitor, where a bipolar electrolytic was used to filter a DC supply derived from an overwind on the HO transformer, which provided, what in valve days, was called  "boost HT" to the Vert OP stage. (too long ago to remember why they used a BP cap, but they must have had their reasons)

Looking up RS Components catalogue, I was pleased to find that the "march of progress" had yielded much smaller BP caps of those ratings, so I bought a few, & fitted one.

Sure enough, the monitor came up OK, & all was well,-----for about ten seconds, then the pix height dropped  to a small strip across the screen, & the new, high tech cap started to exude steam.

It turns out that the nice, small, BP caps have terrible ESR.

The special one was unobtainium, so the monitor ended up with a "Christmas tree" of Polyester "greencaps".
Gorilla-like? -----Yes,but it worked, & worked well!

[srb1954]

I've learned to up the voltage rating considerably when dealing with stuff pre 1980 or so, between running the caps up around the limit and sometimes you find that those much smaller caps of equal specs run a bit warm.

This is more to do with the ripple current heating the capacitor than running the capacitor close to its voltage rating. The heat contribution from the leakage current will be negligible unless the capacitor is run considerably higher than its rated voltage or is faulty. 

For a given amount of heat generated by the ripple current and ESR a smaller capacitor will appear to run hotter because it has a smaller surface area to radiate and convect heat away. Conversely, a higher voltage capacitor will appear to run cooler merely because the larger can's surface makes it more efficient at getting rid of heat.

[George Edmonds]

Divide the AC volts by 1.414.
Or multiply by .707

WRONG , no wonder electrolytic capacitors fail.

The peak to peak voltage of an AC voltage is 2.828 times the RMS voltage, so when rectified you get 1.414 times the RMS voltage.

G Edmonds

[IanJ]

Re the transformer, is there multiple input windings that could mean it's on the wrong tap thus it's high on its output?

Ian.

[Circlotron]

My bedroom digital clock had a backup battery that I replaced with five paralleled 4700uF 10V caps. Has enough capacity to cover about 25 seconds of loss of AC mains. Anyway, these 10V caps have been going 24/7 for at least 15 years with about 12.5V applied. Temperature is low and ripple current is essentially zero. Don't know what the measured capacitance is now though.

[unknownparticle]

Re the transformer, is there multiple input windings that could mean it's on the wrong tap thus it's high on its output?

Ian.

Good point, based on the cap voltage issue!  It's all good though, it's one of those RS transformers from the 80's with a dual voltage primary, 120 and 240 volts. The voltage selection is by a wire link and I'm so used to seeing them that I noticed it was correct as I was looking around the insides.

I have to fess up that I've been running it on the new, 25 volt caps, just to see what happened, naughty, naughty, very naughty
They don't even get warm!  Even with a full load, at 1 amp, nothing, not a peep!  I wonder if there is a voltage tolerance on these Panasonic caps?  Would have to be over 30% though, which seems extreme.

[inse]

Isn't it that the caps grow a thicker oxide layer if the voltage is raised slowly, so that no excessive leakage current can overheat the cap?
And afterwards have gained withstand voltage on cost of capacity?

[unknownparticle]

Wouldn't that cause internal expansion though?  So unless there was space for the internals to expand the cap would burst?

[inse]

We are talking about micrometers or even  nanometers.
In fact it’s all explained in the wikipedia article about aluminum electrolytic capacitors…