How close to capacitor voltage rating?

[Axtman]

How close can you get to a capacitor voltage rating to be safe?

For example I have a filter capacitor that is rated 25 volts and the circuit measures 20 volts. Should I use a 35 volt capacitor or 50 or 100?

[mozz]

Where in the circuit is this? Is there any chance of the voltage going higher or surging?

[Enzo]

The voltages on caps are "working voltage". Meaning if it is a 450v cap, it is OK in circuits running 450v. They also have a surge rating, so it is common in tube amps for the voltage to spike high until the tubes warm up. SO if the 450v supply goes to say 480v for a few seconds and the cap has a 500v surge rating on top of its working voltage of 450, we are all OK.

If you are ordering parts, might as well step up one voltage notch. If you already have 25v caps, use them.

I don't see it so much any more, but used to be the cap said right on it "450WV".

[R.G.]

I would not hesitate to use a capacitor at 85-90% of its rated voltage IF it was pretty certain that it didn't regularly spike up to 110-130% of the rating. It's OK if it overshoots once every power-on. Cap manufacturers specify most of their caps with a surge voltage rating specifically for this reason. What will kill them is continuously being run above the rated voltage from a low impedance source, or being run as a power supply filter cap with heavy loading and corresponding high ripple current. The voltage overage causes high leakage current, and power is voltage times current, so the cap cooks itself from the leakage, or possibly develops a local hot spot inside and shorts. The high ripple current use, even below the rated voltage, cooks the cap by sheer heating caused by the ripple current.

So - modest overvoltages over the rated voltage if they're truly intermittent and the cap can cool down and "recover" from the overage before the next one comes along, are OK. So is running the cap very near the rated voltage. But don't over volt them all the time, and don't run them with big overvoltages frequently.

[Axtman]

Thanks for the responses. In practice I typically match or exceed the voltage rating if I am replacing a capacitor. Usually one voltage rating up. I never reduce the voltage rating but in this case the existing old capacitors were 35 volt rating but the voltage on the circuit measures 20 volts.

I am puzzled by Fender's cathode bypass capacitors. They are typically 22 uf 25 volt. They will never see more than a few volts so why did they make them 25 volts?

[R.G.]

Thanks for the responses. In practice I typically match or exceed the voltage rating if I am replacing a capacitor. Usually one voltage rating up. I never reduce the voltage rating but in this case the existing old capacitors were 35 volt rating but the voltage on the circuit measures 20 volts.

Your practice is a good, conservative one from the standpoint of conservative design for long product life. What is not included in my previous comments is what you do in design. And that depends on what your are designing FOR. You might be designing for reliability and long life. The only good way to do that is to pick components and circuit techniques that derate the components.

Component life is probabilistic. That is, components do not go right out to their 5000hr, 10,000hr, 20,000hr life as specified on the data sheet. Some fail early, the so-called infant mortality. Once those failures are done, the remaining parts fail at a reasonably constant and lower rate until the wear-out processes make the failure rate start rising again. The failure rate accelerates a lot at wear-out. Manufacturers employ statisticians to evaluate what kind of lifetime they can specify so they can claim an X hour lifetime. What that really means is that 9x.xx% of the parts we sell you will last longer than the specified lifetime, and we'll pay you back under warranty for the few that don't.

The thing that complicates this is that the percentage of infant mortality and the base failure rate are dependent on conditions. Run them cool and quiet, they fail less fast. Run them hard and hot, they fail faster. That's the real reason for burn-in programs. Run them hard and hot when they're new, and the infant mortality is done early without shipping the unit to a customer and having it fail in his office/shop/home. It is far cheaper to scrap the early failing units in the factory than to ship them back and try to repair the customers' trust later.

The point of that long-winded rundown of component reliability theory is that you can make things - on average, which only counts when you do a lot of identical things - last longer if you derate the components so their wear-out rate is lower after infant mortality. So choosing a 35V cap where a 25V cap will do comes down to whether you're optimizing a design for long life or for making a zillion of them as cheaply as possible. The unseen issue there is that in production, parts are chosen to (a) work long enough to get past warranty and (b) be as cheap as possible. There is generally a modest cost difference between a 25V cap and a 35V cap of the same capacitance. As do-it-yourselfers, the cost of the 35V part is an extra dime or less. If we had a manufacturing run of 10,000 to do, we'd have to trade off the expected extra warranty costs of the 25V part against the extra up front parts cost of the 35V part.

I am puzzled by Fender's cathode bypass capacitors. They are typically 22 uf 25 volt. They will never see more than a few volts so why did they make them 25 volts?

I'm pretty sure that this decision was originally made by the guy designing the circuits back before we had modern failure rate analysis, but on much the same basis. Plus, before the transistor era, manufacturers of caps didn't make many 10V caps. 25V caps may have been the low cost/availability sweet spot. It's a guess.

[olddawg]

The old rule of thumb used to be to overrate everything by 20% over the max you would see happening in the circuit. 20v in circuit for a cap, use a 25v cap, etc. Same for resistor size..., Zener wattage, etc.

[the fatch]

Running an electrolytic capacitor below its rated voltage increases its working life, as does running it below its rated temperature. Randall Aiken explains in his Technical Q & A at: www.aikenamps.com/index.php/technical-q-a

[Leo_Gnardo]

I am puzzled by Fender's cathode bypass capacitors. They are typically 22 uf 25 volt. They will never see more than a few volts so why did they make them 25 volts?

It can be a matter of what's available. Furthermore, what's available cheap. From the era when the classic Fenders were made, I have seen other manufacturers use as cathode bypasses caps rated for 10 volts, 6, 4, even 3 volts. Try to find anything like that now. Low voltage caps, sure they're available - starting with capacitance values in the hundreds of uF, and in recent years supercaps rated in Farads. Fender quickly settled on 25 uF 25V, mostly dualies (2 caps in one package), must have bought millions of 'em over the years, and you can be assured they paid rock bottom prices. Sure, you can order up anything you like when you're buying 100,000 pieces or more for manufacturing, but they settled on those caps probably for reasons that had more to do with economics and availability.

[Chuck H]

It can be a matter of what's available. Furthermore, what's available cheap... Fender quickly settled on 25 uF 25V, mostly dualies (2 caps in one package), must have bought millions of 'em over the years, and you can be assured they paid rock bottom prices. Sure, you can order up anything you like when you're buying 100,000 pieces or more for manufacturing, but they settled on those caps probably for reasons that had more to do with economics and availability.

This^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

In other words, Fender probably spent less on a batch of 10,000 standard 25uf/25V caps than they would have ordering 10,000 custom, and otherwise unavailable 25uf/5V caps even if those caps had lower materials cost to make. Supply and demand.

[nevetslab]

I am puzzled by Fender's cathode bypass capacitors. They are typically 22 uf 25 volt. They will never see more than a few volts so why did they make them 25 volts?

As stated by R.G., long term reliability in using a part that's typically 10 times the working voltage in the case of the cathode bypass caps. Another aspect to this, when selecting caps for better performance (more linear..it's Xc slope remains closer to the ideal out to a higher frequency...often goes hand in hand with voltage ratings....up to a point. With the much wider range of capacitor designs these days, a typical 22uF/25V cap is a tiny-tiny part compared to what we found with 1980 date codes. I'll typically use 22uF/100V part, even over a 22uF/63V part, having seen the impedance & phase curves on a particular brand/model. Will you hear the difference, it being a cathode bypass cap? I doubt it, but, it's more of a quality thing, knowing the replacement part installed will last and is a superior part to what had been there.

[Enzo]

And I think it pays to put things in context. When recapping a 50 year old Fender amp with 25/25 cathode caps, is it really a ting to worry about longevity of caps to be gained from going to 50v caps? I mean the poor overworked 25v caps lasted 50 years as it is.

I probably use 100v parts where 50v or 25v parts are good myself. As caps are so small these days, the size is no longer an issue, so I might onl;y stock the 100v rather than both 100v and 50v and 25v, just as I only stock 1N4007 diodes instead of also 1N4001,2,3,4,5,6 types.

[Merlinb]

Aluminium electrolytics are usually quite happy running right up to their voltage rating. But tantalums are not; for DC power applications (i.e. where the voltage is applied permanently) they should be derated to maybe half their working voltage.

[Dave H]

Aluminium electrolytics are usually quite happy running right up to their voltage rating. But tantalums are not; for DC power applications (i.e. where the voltage is applied permanently) they should be derated to maybe half their working voltage.

Very true, I hate tantalums. I found out the hard way that they go short circuit at the slightest hint of overvoltage.

[Enzo]

And they faint at reverse voltage just as quickly.