Conversely, you will commonly see 63 WV caps on nominal 50-55 V rails, not a great deal more usually. I think I've seen 50 V on 56 WV caps as well. So rail voltage might approach cap working voltage at the upper limit for mains voltage tolerance but that's about it. Like that, these generally last for many years with good-quality parts.
I know of a range of '70s Rotel receivers that are somewhat notorious for blowing their mains filter caps - rails were close to the limit to begin with, and the 220 V to 230 V transition in mains voltage did the rest. Even then, this only tended to happen once the units were already qualifying as vintage.
I know of a range of '70s Rotel receivers that are somewhat notorious for blowing their mains filter caps - rails were close to the limit to begin with, and the 220 V to 230 V transition in mains voltage did the rest. Even then, this only tended to happen once the units were already qualifying as vintage.
With big banks of energy discharge caps, this gets exciting very quickly.
At the lab I work at one of the experiments has several rooms full of large oil-filled pulse discharge caps. I want to say they're 330 uF 5 kV, but it's been a while since I've seen them. Switching is done using size E ignitrons. Long before my time, they have had caps short in that bank. The results are extremely violent and some of the senior engineers have seen bus bars get vaporized and chunks of welding cable get permanently embedded in concrete walls. With small caps it's a loud pop, with banks of large caps it is more like a bomb going off.
Back to the subject at hand, power amplifier manufacturers do seem to have a habit of pushing the voltage rating on power supply caps. They also seem to get away with it. MC2 MC650s ran 88V on 100V rated caps. They were 85 degree caps too, not 105. We've started to see a few MC650s that need recapping, but even the newest ones are 20 years old and most of them that were used in installations are probably pushing 100k hours. Crest ran about 60V on 67V rated caps in the CA9, and I haven't heard too many complaints about them having power supply problems.
How do they get away with it? The answer is that most of the time, these caps really don't see huge ripple currents. What percentage of the time are most power amps outputting more than 10% of their rated power? In a home listening environment, how much time do amps spend outputting more than 5 watts? Most of us are running far more power than we'll ever use.
Even in a live sound application, most of the time these amps only get pushed for very short bursts. In a touring amp, they probably won't run more than a few hours at a time. In an installation, they'll probably run 24/7, but idle 90% of that time. Cap manufacturers are often anticipating that their products will be pushed much harder.
If you're building a class A amplifier, all of this goes out the window because the power supplies sees a lot of stress all the time. In a normal, optimally biased Class AB amplifier, the power supply sees very little stress.
That said, this is DiyAudio. Shaving maybe $4 off what might be a $200 BOM is relatively insignificant. In manufacturing it makes sense, because that $4 spread across 10,000 units is a noticeable savings.
Last edited:
After rectification right. 400v test. Unless he had bipolar caps
They were bipolar.
I remember also in my apprenticeship someone connected a 100uF 350V electrolytic the wrong way round, and the can exploded and a reel of red hot burning aluminium foil travelled about ten feet across the lab leaving a trail of smoke.
I think the failure is generally catastrophic rather than gradual, breakdown of dielectric. It is going to be a empirical write-off situation - a sacrifice.
Why not just order a cap with an adequate voltage rating and remove all worries about it? It's likely that your caps will tolerate the slight overvoltage you mention. However, if there are increases in line voltage for some reason, the caps could get pushed too far. A good experiment you might try is to measure leakage current (after cap is charged up) first at the rated voltage, and then at your new higher voltage. If this current isn't higher than the reading obtained at the in-spec voltage, likely you'll be okay. It is very poor design practice, but if this is a onesi just for your own use, then I see no problem. Keep in mind that one copy of the cap may tolerate the overvoltage, while another may not. As the caps age out, they may fail prematurely at the overvoltage.
Testing is done with a current limited supply - as simple as something like a 1k resistor in series with DC voltage that can equal or just exceed (up to +10%) your cap rating. Don't keep the cap above the rated level for longer than 10 minutes - any part from a reputable manufacturer is compliant to that level of stress. If the cap has an internal micro-arc event then the 1k resistor won't allow the voltage to increase, and the pin-hole causing the arc will form up, as is what happens during normal manufacturer forming.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.