Capacitor Voltage Rating

I'm a bit unclear on what capacitor should be used in terms of voltage rating.

On a previous post I read it was safe to run 32VDC on a capacitor rated @ 35VDC.

My understanding and from what others say - a 20% safety margin was the norm? My current project ; Pass's Balance Line Stage quotes "Substitute capacitors should have a 100+ volt rating". The power supply voltage it uses is 80V and after regulation it drops to around 60V. So why spend the extra $ when a 63V cap will do in the 60V circuit?

I know electrolytic capacitors can have a tolerance deviation of +/- 20% That is a huge variance and what does the tolerance imply if running capacitors at near 100% rated voltage level? I know capacitors explode like a bomb when exceeding their rated voltage. But I also have seen them explode within normal voltage range.
 
The voltage rating of a capacitor is based on several things. Most importantly, it should not be lower than the expected voltage across it. If the capacitor is located after a regulator, then the voltage rating doesn't need to be much above the regulated voltage (a few percent is normally OK). If the designer is paranoid, they will sometimes set voltage ratings on the output capacitors so that a regulator failure won't kill the capacitor. If there's a crowbar circuit, or if the circuit can function with the higher voltage (with some added noise or hum), this would work just fine.

On the other hand, the filter capacitor that's located after a rectifier needs a healthy voltage margin. This is because the mains voltage is poorly regulated. A variation of 20% (up or down) is quite acceptable to the power company, and to the electricians who wired your house. There are few commercial devices that aren't designed with this in mind.
 
First off, the +-20% margin on electrolytic capacitors isn't on their voltage rating, it's on their capacitance. It has no effect on their ability (or lack of it) to stand rated voltage. The voltage rating means what it says.
When Nelson says use 100V caps for an 80V rail, you've got to keep in mind what the standard voltages are. In general, the voltages in that range are 75V and 100V. (You can find other voltage ratings between those two, but they're rare.) Since he's already past 75V, obviously it's necessary to jump to the next higher voltage.
Where this 20% idea came from is a mystery to me. I've never seen it in a book, and all the equipment I'm familiar with (both stereo and mainframe computer) runs between 5-10% less than the rated voltage, and that's only to allow for line spikes and such. Actually, like most things, you can push it a bit if you like living dangerously. I remember one amp back in the 70's that ran a rail about 5-8% *over* the rated voltage of the power supply caps.
On the other hand, the company is longer around. Draw your own conclusions.
Anyway, for an 80V rail, I'd run a 100V cap, unless I came across a 90V or so, but that's not a common voltage. For a 60V rail, you could run a 63V as long as the rail is regulated. For unregulated voltage, that would be cutting it close, so I'd recommend a 75V cap in that case, since the common voltages generally jump from 63V to 75V.
A related point is that, given quantity price breaks, it may be cheaper to buy 10 (or 100, or 1000) of the higher voltage rating, and use them throughout the circuit, rather than buy 2 of this and 2 of that and 4 of the other one...
Yes, electrolytic caps can explode, but I wouldn't use the phrase 'like a bomb.' The only ones I've had literally explode are the smaller ones that don't have a vent (let's say 'like a firecracker'). Large computer grade caps have a rubber plug that pops out, releasing the pressure. I've had a few of those go, too, but they're pretty tame; just messy and smelly.
It's your money. If you feel that it's worth jumping two or three steps or more in voltage, go for it. Just make sure that you put enough voltage on it so that it forms, or you won't get the capacitance you're paying for.

Grey
 
PSU filter Cap's, and others connected directly to unregulated PSUs, should have appx. 20% ( or more) higher rating over nominal for several reasons.
Your power company has a nominal tolerance for the power they deliver,- here in Europe +/- 10% is rather common,- but it can often be more than +10% if you live close your mains supply transformer.
Also,- usually the nominal voltage for toroid transformer are spec'ed with ref to full load, where as E/I cores and similars are spec'ed at idle. Thus, - pushing the voltage load on cap's can give a few rather unwanted surprises...........
 
I forgot to mention about the life of the capacitor. Surely running the capacitor at 95% of it's rated voltage would not last nearly as long as running it at 80% of it's rated voltage. I've heard you may only get 30% of the life of the capacitor if you run it AT rated voltage. Comments?

For power supply filtering, I figure the 20% margin is right when using toroidal transformers in an unregulated power supply.

Biggest capacitor explosion i've seen was in my friend's stereo in his truck. He blew one of those big capacitors and the whole truck inside was filled with paper like shrapnel.
 
I had an idle moment, so just for fun I called my power company to get their input.
Nominal voltage here is regarded as 120VAC. The fellow said that if you were near a transformer that you might get a little over--no more than 122VAC--and that the lowest they allowed was 119VAC.
He was amused/horrified that anyone would think that +-20%, or even +-10% was acceptable or normal in any way.
Low voltage isn't really that much of a problem, so let's look at high voltage. 122V represents a little over 1% increase over the nominal voltage. Supposing that you were to choose, say, a 32V rail. A 1% increase would lead you to 32.32V...clearly still within margin for a 35V cap.
If in other contries the voltage varies more, then that should be taken into account, but here in the US 20% overvoltage would be 144VAC. When's the last time you heard of an outlet that measured 144V? If you're measuring wall voltage that high, you need to have a talk with your power company. There's no need to put up with that kind of nonsense.
People start talking about things like this, and it gets passed by word of mouth without anyone ever checking things. Eventually it becomes something "everybody knows" without any factual basis.
Running the cap at rated voltage should not reduce the life to 30%. By definition, running a cap at rated voltage should give you 100% of the rated life. At a guess, it sounds as though there was a substantial amount of ripple current there. Lots of ripple means someone did a poor job designing the power supply, not that the cap is at fault.

Grey
 
Grey

If your mains supply differs so little in the US then you are very lucky. Here in the UK, the statutory obligation is to maintain the mains supply between 230V +10% and -6%. In reality, I have measured line voltages, in different locations and over a period on time, of between 205V and 255V. Add to this the variation of up to 10% due to transformer regulation (no-load to full-load) and the design of power supplies, particularly regulated ones, (and the rating of capacitors) becomes far more problematical.

Geoff

[Edited by Geoff on 09-20-2001 at 08:15 PM]
 
Geoff,
As I said above, if you live elsewhere, take local conditions into account. That's part of why companies have Export models of their equipment which vary internally from the domestic versions.
The power I have at my current house is the worst I've ever had, but even then it only varies by about 2V, or roughly 1%. I'm assuming that it's load related, as I'm nearly at the end of the distribution line. In my previous house (same power company, and only about two miles away as the crow flies), the power was rock steady...always. I don't remember it ever varying as much as .5V. I don't remember where the transformer was in relation to the house.
Given that the weather here is mild now (mid-September, highs in the mid to upper 80s, lows in the upper 50s to mid 60s...all degrees Fahrenheit), I ought to measure my voltage to see what it's doing with minimal load on the line from air conditioning, etc. I'll try to remember to do so when I get home.
Naturally, this is a separate concern from noise on the line, which we've poked at elsewhere. The AC line voltage I have now is the worst I've had in that aspect, too, but excepting really big spikes, then it wouldn't enter into the voltage question. Even then, you'd be better off tackling the problem with MOVs, caps (small ones across the incoming AC line, not the bulk caps), or fuses, rather than going to something like 10kV caps on the off chance that it might save your bacon in the event of a lightning strike.

Grey
 
Is there anything wrong with using 250volt caps on a 70 volt rail? My local electronics shop has four surplus 26,000uf 250v caps for the the same price as four 10,000uf 125volt caps. I'd like to use the bigger caps but i'm not sure if the voltage ratings will cause problems.

Thanks in advance, Chad
 
certainly, there's nothing wrong with using an overspec'd cap. In fact, I was thinking that Grey's example of 32.5 volts being within the range of a 35V cap was probably cutting it a little close for my comfort. For anyone who's had a cap overvoltage and blow on them, you know what i mean ...they make an extremely loud bang (like an explosion, really) and spray all kinds of lovely fluid into a cloud of mist which settles on everything. I typically add a minimum of 25% to the supply voltage to get the cap rating.

If I were you Chad (wait a minute, I *am* Chad... hmmm... ;) ), I'd snap up those caps. 25,000uF caps with a 250V rating are rather hard to come by, even the 10,000/100V caps are probably a steal. But, I'd be careful to check their age first! Electrolytics degrade slowly, and the tolerance may be high because it covers the spec'd lifetime of the cap (MTBF usually).

Chad.
 
Hmmm....

Clearly,- in some of the posts, there must be some misunderstandings.

The tolerance of +/-20 % relates only to the capacitance value. There is as such, no tolerance stated for the voltage rating, - but usually in "pro" circles, we put a finger in the wind, and expects a cap to forgive us some overvoltage spikes, and to tolerate some 10% overvoltage on an intermittent basis. Cap's should of course NEVER be designed in to run on permanent over voltage. Giving room for a 15-20 % overvoltage is concidered good design practice.

ESR does not diminish by increasing voltage, it increases.
An example,- a RIFA comp. grade, - 22.000 uF/ 25V has an ESR of 5 momh, and the 100V version has 15 mohm.
The relative value of ESR also, quite logically, increases with lower capacitance. In this light, the practice of some manufacturers to parallell "many" smaller value caps, relies totally on the idea of parallelling the ESR of each individual cap, which in turn really depends of the resistance of your parallelling connection,- copper plates or "thin" wires, anyone???

Running a capacitor on relatively lower voltage than rated does not lower its capacitance or lifetime. Where does this come from ?? High temperature though, reduces the lifetime, and often conversely also the capacitance, by aging.

Somebody mentioned MOVs, and this is of course a very good idea, and commonly used in pro gear of all kinds. This is the remedy for high over voltage spikes, but not for marginal capacitor voltage rating. If you live in middle latitudes and/or have air born power lines to your house, OV spikes from discharge are rather common.

Even if some of you US residents have rock steady line voltage in your houses, I do feel quite sure that US power companies also have tolerance values of at least +/- 5%, as this is also a part of their mechanism to protect themselves from claims of damage resulting from varying line voltage, among other things also from load variations. Designing a power grid without allowing for load variations is impossible.

Just to sum up,- you have to allow both for line voltage variations, and the load variations you create by e.g. turning your amp up or down, reflecting to voltage variations from the amp's own transformer.
Allowing for up to 20% variations of the voltage on your caps is indeed a good idea.

BTW, - there is an excellent book available on the net -
the "Rectifier Applications Handbook", from Motorola. Try a search, - my net connection is somewhat clogged at the moment. But beware,---it's huge !! (some 350 pages....)



[Edited by AuroraB on 09-21-2001 at 04:32 AM]
 
Two ideas for those who are afflicted by power fluctuations:
--Build a regulator circuit into the line from the rectifier to the main bulk caps...but don't design it for the target voltage, set it for, say, 3 or 4 volts higher. Under normal AC line conditions it will be in full conduction and add minimal resistance to the power supply as a whole. In overvoltage conditions, it would clamp the voltage at whatever level it was designed for. Power dissipation in the circuit would be minimal. Obviously, it would always carry full current, but the voltage drop under normal AC line values need only be the normal drop across the pass transistor, so heat would be limited to a few watts at most. In an overvoltage condition, you would need somewhat more power dissipation capability, as the circuit would then be producing more like 10 or 20 watts. The circuit could be built for a few dollars worth of parts (excluding a heatsink for the pass transistor) and would surely be much cheaper than buying caps with higher voltage ratings.
--Purely by coincidence, I ran across a blurb last night on a new power box from Monster Cable. It doesn't work like the usual box with filtering, etc. Instead, it has a variac inside with a motor drive. It detects the incoming voltage and corrects it via the variac. Conceptually simple. Implementation might be a bit more difficult, but I can't see that it would be *that* bad. The flow chart would probably go something like this:
AC->variac->rectifier->comparison of rectified DC to reference voltage->correction via motor
What kind of motor to use to drive the variac, I'm not sure, but the monitoring circuit would be trivial. Needless to say, they gussied it up with LED readouts on the front and such, but you need not worry about that. The variac alone would cost perhaps $75 to $200 depending on current rating and whether you were able to source it surplus or new, so it wouldn't be a cheap project. Still, it'd be something to consider.
Model: AVS2000
List price: $1499 (surely it could be built for $200 or so)
Web: http://www.monstercable.com

Grey
 
Thanks Everyone!

Great discussion on capacitors.

Here's a link for the caps i've bought for my next project (warning it's a large jpg); Pass's Balance Line Stage pre-amp (4 channel).

http://www.geocities.com/super_bq/HiFi/Rubycon_200V_1200uF_Electrolytic.jpg

Notice the 105°C rating. Capacitors may be an overkill but I just could not pass up on the great price. Bought 14 of them @ $7cdn each (includes taxes). Also Rubycons (which make the Blackgate audio capacitors) are great for audio.