Safety margin for voltage rating on electro caps? Nil?

[brett]

Hi
I had 2 16V aluminium electro caps that I used in a small power supply.  The supply was +/- 16 to 17V, so I figured that my caps would be just ok.
Unfortunately, no.  After half an hour or so, the cans started to release steam from the top vents.

I was assuming there's some sort of a safety margin, but maybe there isn't.  Any similar or contrary experiences out there?

PS the project is a simple 6W amp, to be powered from a 24V wall wart.  Kinda like six Rubies.  It'll be called the Mighty Midget. 

[rockgardenlove]

I've popped off some 16v Xicon's with a 13v wall wart.  Granted, it was a cheap one that fluctuates up to 15v sometimes, but I was still surprised. 

[JimRayden]

For AC rectification, the general rule for cap rating is sqrt2 of the AC voltage (plus a bit more). I guess it'd be safe to use that here too. So a circuit of 16 volts would be well off with 25V caps.

---------
Jimbo

[zachomega]

I know from tube amps, that you don't run the caps at their maximum rating.  For example, if the voltage is 350 volts, you wouldn't pick a 350 volt cap, you'd probably go for 500 or maybe even more. 

Another thing to keep in mind is that heat lowers the maximum working voltage of the cap. 

-Zach Omega

[Paul Perry (Frostwave)]

+1 to the replies above. And one of the reasons stuff -like caps - is so small today, is that the 'margins' aren't there any more, I suspect. It used to be that manufacturing tolerances meant you had to design with a bit to spare, now it is "just" able to do the job.
Yeah, I've blown up a 25V electro in a filter, usign an 18V AC plugpack. Now I use a 35V electro there.... another thing, it's unlucky to put a TO220 regulator flat up against an electro - that doesn't do anyone any favors.

[brett]

I was using a 24V AC plugpack.  Multiplying by 1.4 and subtracting 1.4V for the bridge rectifier gives +/- 16.something volts.  Under low load I measured it at +/- 17.1 V, and I figured that the voltage would fall under load and the caps would be happy, but the reverse happened.  The load (about 400mA) seemed to play a part in the downfall of the caps, as they burned when the baby amp was cranked.

thanks

[R.G.]

Electros should ideally be run at about 80% of their rated maximum voltage. You really never want to go over the rating. Electros have a surge rating of something like +110% or 115% of the maximum rating, but that is truly a surge rating. They will withstand that overvoltage for seconds to a few minutes. Maybe. Sometimes the surge is specified as a number of cycles of the AC line.

[Mark Hammer]

Although there are a variety iof identifiable circumstances where lower voltage-rating electrolytics CAN be used, caps are so cheap and so small these days (compared to 25 years ago when you had to pay for the "privilege" of being forced to make a bigger board to accommodate those clunkers) that here is very little excuse or not simply making it a point of protocol to buy/order 25v caps and be done with it.  Heck, I've got 50v rated units in my parts bins these days that are smaller and cheaper than caps I bought when I was starting out that were 1/4 the capacitance value and rated 10v.

Keep your eyes peeled for old 8-bit and 16-bit Soundblaster cards in garage sales.  Those suckers often have a whack of tiny 4u7, 10u, and 47u caps rated at 50v.  If you have a method for removing them safely and easily you'll be in great shape since the leads are usually long enough to re-install in other boards.

[Tubebass]

I built a PS 20 years or so ago that operates 20 volt computer-grade electros at 21 volts! The supply has been powered almost continuously for all these years. Guess the older caps did have some safety margin!

[R.G.]

1. Computer grade caps are called "computer-grade" for a reason. Computers used to be things only corporations or governments could afford. Computer grade caps were made with higher surge voltages and reliability in mind.
2. Overvoltages that are both steady and small do not kill the cap immediately.
3. The way an electro cap is made is that bare aluminum is exposed to an electrical current inside a bath of electrolyte. The leakage current grows insulating oxides over the surface. Powering a cap only a volt or so over its rating and leaving it forms the oxides to a higher voltage. You have to never let the leakage current get too high. If that's true, then high quality caps will form up to stand the mild overvoltage.

Like all specifications, there is some margin. A "25V" cap specification indicates that the manufacturer will give your money back (in theory at least) if it fails at 25V. So the makers have to ensure that their failure rate is well under 1% for example at that specification. Because all manufacturing processes show a statistical variation, you have to put the mean (average) of the distribution well above the guaranteed value. If your manufacturing is tight and you can make caps that have a small sigma (standard deviation) then you may be able to make caps where 99.99% are over 25V and under 26V. If you're sloppy and sigma is larger, you may have to make a nominal 30V cap to ensure that 99.9% are over 25V and not get warranty returns.

Modern manufacturing has made distributions much, much tighter. That means that they don't have to put as much margin in, and so the newer the cap, it's possible that the tighter the voltage margin will be.

[Paul Perry (Frostwave)]

  The load (about 400mA) seemed to play a part in the downfall of the caps, as they burned when the baby amp was cranked.

Maybe a coincidence - but maybe not! because, when the load increases, ther is more ripple across the caps & thus a lot more heat generated. Heat kills......