Making non-polarised caps from polarised caps, ok for high current AC?

[TMM]

So everyone probably knows the trick where you connect two polarised capacitors back to back to create a non-polarised capacitor of half the capacitance. This works great for small signal coupling but i'm trying to find info on reliability (e.g. oxide layer degradation) in a high current application. Say i have polarized caps with ripple current ratings of 800mA. If i put them back to back do I get a non-polarised caps with a ripple current rating of 800mA?

The actual application is a speaker crossover, so my options are:
-Use plastic film caps = huge/expensive
-Buy wun hung lo NP 'audio' electros of unknown origin/specs.
-Buy fancy pants audio brand NP electros for a ripoff price.
-Buy Panasonic/Nichicon/Chemicon/etc NP electros with laughable ripple current ratings because the intended application is low current signal coupling.

Meanwhile i have drawers of polarised caps with respectable specs lying around, so using them seems attractive.

I've tried to find info on audio forums but nothing concrete turned up just a bunch of noise from people who simply say 'don't do it' or raise concerns about distortion. Distortion doesn't bother me because i can measure it, I'm more worried about them becoming a dead short over time or something potentially destructive like that.

Thanks in advance.

[DanielS]

I have always kind of wondered about that but never bothered researching it.

The most interesting (but dangerous) way to find out would be to put a pair of 400V 220uF caps back-to-back in a lexan tube, plug them in and see what happens.

[rob77]

I have always kind of wondered about that but never bothered researching it.

+1  ...the very same here

[eurofox]

 

[DanielS]

After further thinking about it, my initial uneducated guess is that reverse leakage current will allow the capacitors to build a DC bias which will eventually cancel out the negative voltage - assuming the capacitors survive long enough to reach that point. The signal amplitude would likely need to be brought up progressively to avoid large instantaneous reverse-current spikes that might instantaneously damage or kill the caps.

[SeanB]

If you use the ones meant for CRT monitors, where you typically have a non polarised capacitor of around 2u2 to 10uF 50v non polarised which has a ripple current of around 3A peak through it in the scan coil system, you should have no problems.

In a speaker crossover just use the 2 in series you have, the current is going to be much below the 800mA rating most of the time in any case. I have seen very expensive speaker systems ( as in the ones where you pay with an equivalent mass of silver for them, and they have damping mass in them) where the crossover capacitors are the cheapest mystery small case variety. Any brand name cap should outlast them easily.

[NiHaoMike]

Use diodes to prevent reverse polarity.

[TMM]

Diodes would prevent total destruction of the capacitors if one went short circuit, but what about long term effects?
Even with diodes each capacitor is still going to be subject to a small reverse bias for a half wave and then a large forward bias for a half wave.
My concern is that the half wave of reverse bias will deplete the oxide layer slowly and i end up with a shorted capacitor which harms the speaker or amplifier. If a capacitor goes short circuit there is the potential to be sending low frequencies to a tweeter or presenting a very low impedance to the amplifier.

[David Hess]

Electrolytic capacitors can tolerate reverse voltages greater than the forward voltage drop of a diode so using anti-parallel diodes will work fine.

[T3sl4co1l]

At liberty to divulge the volts, amps, frequency and capacitance this thing is required to withstand?

Tim

[TMM]

A typical requirement would be 20uF, 50V, 2Arms @ 1KHz.

Electrolytic capacitors can tolerate reverse voltages greater than the forward voltage drop of a diode so using anti-parallel diodes will work fine.

I was under the impression that any sustained reverse voltage would degrade polarised capacitors, just that it is severe enough at <1V that they go bang. What i'm trying to find out is that if the average voltage is positive and it doesn't heat up to the point that it goes bang, do they still degrade long term or does the forward voltage repair the damage that was done in the negative part of the cycle.

[T3sl4co1l]

A typical requirement would be 20uF, 50V, 2Arms @ 1KHz.

Ok.  Well, 16Vrms (20uF @ 1kHz is 8 ohms, * 2A is..), 50V peak then I guess.

That's a little high in current for NP but it sounds like a typical use case.

I was under the impression that any sustained reverse voltage would degrade polarised capacitors, just that it is severe enough at <1V that they go bang. What i'm trying to find out is that if the average voltage is positive and it doesn't heat up to the point that it goes bang, do they still degrade long term or does the forward voltage repair the damage that was done in the negative part of the cycle.

I think the reverse figure is more like 5-10% of rating, or 3-5V, whichever is smaller.  Not sure if the mechanism is un-forming, which would then be re-formed.  At frequencies above the diffusion rate (pretty much anything), it shouldn't have much if any effect on structure.

But I don't really know if that's true, or something else happens in reverse..

Using diodes is a standard way of preventing reverse bias; or if you have DC available, you can use a large resistor to polarize the middle node, so both caps see proper forward bias.

Tim

[DanielS]

I was under the impression that any sustained reverse voltage would degrade polarised capacitors, just that it is severe enough at <1V that they go bang.

Well, in an AC application with two caps back-to-back, the reverse bias would only be transient: until the opposite capacitors blocks the whole voltage and on the opposite cycle, that other cap would only go down to ~0V while the opposite cap picks up all the input voltage. So, the two capacitors only see reverse bias until they both achieve high enough DC bias through the other capacitor's reverse leakage and they do not reverse anymore. In principle, the capacitors should not see little more reverse-bias energy than what is necessary to charge/bias the opposite cap and once both are biased, they no longer see reverse bias.

How much damage can that occasional transient reverse bias do? That is a good question. Enough to warrant protection diodes?

[Circlotron]

Much easier to use AC power factor correction caps like the ones used in fluorescent lighting fixtures.
Rugged, cheap, lots of values, very widely available from electrical wholesalers.
B.Y.U.
http://www.ambercaps.com/lighting/seriescp.htm

[SeanB]

20uF 250VAC motor run capacitor in a plastic case, with either wire leads or push on connectors then will be cheap and last essentially forever. They handle 15A in use with motor loads. Plus they will self disconnect if you do manage to overload them.

[Artlav]

I was under the impression that any sustained reverse voltage would degrade polarised capacitors, just that it is severe enough at <1V that they go bang.

Hm, i think i read somewhere that it was an electrochemical effect that destroys them - the insulation layer begins to dissolve.
That would imply some minimal voltage for it to start happening, similar to how water does not electrolyse below at least 1.23V.

Anyone know if that's right?

[David Hess]

For aluminum electrolytic capacitors:

Application of overvoltage on the order of 50 V beyond the capacitor’s surge voltage rating causes high leakage current and a constant-voltage operating mode quite like the reverse conduction of a zener diode.  Application of reverse voltage much beyond 1.5V causes high leakage current quite like the forward conduction of a diode.  Neither of these operating modes can be maintained for long because hydrogen gas is produced by the capacitor, and the pressure build up will cause failure.

http://www.cde.com/catalogs/AEappGUIDE.pdf

For tantalum capacitors:

Solid tantalum capacitors are capable of withstanding peak voltages in the reverse direction equal to 10% of the DC rating at +25C and 5% of the DC rating at +85C.

For wet tantalum capacitors, the peak reverse voltage does not exceed 1.5 V and the peak current times the duration of the reverse transient does not exceed 0.05 ampere seconds. In addition, the repetition frequency of the reverse voltage surge is less than 10 Hz.

http://www.vishay.com/docs/40110/faq.pdf

Reverse operation is a lot like an electrolytic rectifier:

http://en.wikipedia.org/wiki/Rectifier#Electrolytic