Normal Electrolytic Capcitors as non-polarized (bipolar) ones in AC circuits?

[Electrobrains]

Has anyone knowledge or experience concerning electrolytic capacitors in AC circuits?

Of course you have the specially made non-polarized (NP) versions, but they are expensive and don't go high with capacity.

In Wikipedia, it's stated that the reverse voltage should not exceed 1 to 1.5V. Still I have seen published circuit diagrams with the configuration of two electrolytic capacitors connected with the minus (or plus) poles together and used in AC circuits (see diagram).

As I understand, what usually takes place when an electrolytic capacitor is getting problems, is that the electrolyte starts boiling (evaporating gases) and the thing explodes in a more or less spectacular way!

Could it be that it's possible to use electrolytic capacitors in AC circuits, as long as they don't heat up "too much" (like in normal DC circuits with much ripple)?
Could it be that the changing of polarity would protect them from being chemically destroyed and allow them to be used at much higher reversed polarity voltage?

Has anyone read about this or made some own research?
(please not just the "you-shouldn't-do-that-answer")

Thank you for answers!

p.s. Once I really got shot by an electrolytic cap. One of my dear children had helped me to solder some components on a PCB. When I tested the board, one small electrolytic capacitor had been reversed polarized. Obviously the thing didn't have a venting system, but came flying like a bullet on my hand. Luckily I was only bruised, but I can tell you it really hurt.

Here a guy showing what you (officially) shouldn't do to these things: exploding capasitors

 

[Harald Kapp]

Even if you ask

please not just the "you-shouldn't-do-that-answer"

I have to say that you should not use normal (polarized) Elkos with AC.
The series circuit you attached is a possible solution to the dilemma. It works because at any time one of the capacitors is correctly polarized.

But why the question mark? The capacity of the series circuit is calculated as any series connection of capacitors.
You can improve the stability of the circuit by adding a resistor in parallel to each capacitor (e.g. 100 kOhm) if your circuit can tolerate the DC component that results from the two ressitors in series. The resistors will help to balance the charge on the capacitors.

 

[Electrobrains]

But why the question mark? The capacity of the series circuit is calculated as any series connection of capacitors.

Hello Harald
The question mark did not refer to the capacitance or how to calculate it.
It refers to the symbols (i.e. if two polarized capacitors can be seen as one non-polarized under certain conditions).

It seems not to be very much said about this and some people obviously already have practiced this approach in some areas. Also the Wiki statement "1 to 1.5V" shows that there is a gray zone "on that side of zero" - for DC. For AC the conditions are probably better, as also hinted at on the Wiki page.

There are no specs (as far as I have seen) given from Capacitor Manufacturers on this topic. That's why I ask here, if anybody has knowledge or experience.

Anyway, thank you for the hint about the stabilizing resistors! But does this really help anything in a (symetrical) AC circuit?

 

[john monks]

I work on circuits that use electrolytic capacitors, back to back, in AC circuits.
What I discovered is that a DC voltage develops on the capacitors that prevents the AC peaks from reversing biasing the capacitors by very much. 1 volt sound right after running for a while. It seems that the capacitors work as rectifiers to some extent. I have not seen this arrangement fail so far.

This arrangement is frequently used in speaker systems.

I do this with electrolytic capacitors and tantalum capacitors.

 

[Electrobrains]

I work on circuits that use electrolytic capacitors, back to back, in AC circuits.
What I discovered is that a DC voltage develops on the capacitors that prevents the AC peaks from reversing biasing the capacitors by very much. 1 volt sound right after running for a while. It seems that the capacitors work as rectifiers to some extent. I have not seen this arrangement fail so far.

This arrangement is frequently used in speaker systems.

I do this with electrolytic capacitors and tantalum capacitors.

Very interesting!

I suppose you are into audio then, with rather soft/low frequency signals?
My application would be a square wave, power transmission type.
Maybe those diodes could help anyway, to keep the reverse polarity within limits at start-up and at the "surgy" edges.

 

[john monks]

I work on high voltage pulse generators and laser pulse generators.
I recently finished a project requiring to have a DC level that will go anywhere from -500 volts to 500 volts with about 200uF of filtering. I used two 470uF capacitors back to back without diodes.
From what I have seen in the lab a little bit of voltage reversal does not seem to harm a capacitor at all. And I'm afraid to place a diode across capacitors because an excess amount of current discharging the capacitor string might wipe out a diode.
So I feel safe with what I did.

 

[john monks]

When I connected two 470uF capacitors in series I got 235uF. I only needed 200uF to meet specifications. But I'm happy to have some overkill.

Your are right. If you discharge or that is bring the ends of the capacitor strings to the same potential, 0 volts, the diodes are not affected.
Using back to back electrolytic capacitors in a power supply is an unusual application. I did in my application because I needed to go from -500 volts to +500 volts.

About the ripple current in the datasheet I would have to see to be sure.
Now the ripple current in a series string is going to be the same in both capacitors. Nothing is going to be halved in terms of ripple.

As far as the diodes are concerned there is a surge current that is specified in the datasheet that is typically many times the average rated current.

 

[Electrobrains]

Hm.. I don't see how you get half of the capacitance in the AC case, if there is this self-limiting reverse voltage feature... where is the theoretical error?
I think I really have to connect some of those caps and measure it.

 

[john monks]

Let's look at it this way.
If you have a 10 volt power supply and connect a 1 farad capacitor to it you will develop a charge of 10 volts across the capacitor and you will will have 10 coulombs of charge or electrons in the capacitor.
Now if you connect the capacitor to a 5 volt power supply you will have 5 volts and 5 coulombs of charge in the capacitor.
Now if you charge the capacitor over a time of one second you will have a current of 10 amperes and 5 amperes during that second because current in amperes is equal to the change of charge in coulombs divided by time in seconds.

Your experimentation will show you that your charge is directly proportional to your voltage on the capacitor.

Capacitance in Farads is defined as coulombs divided by volts.

Now if you have a 10 volt power supply and have two of the same value in series you will have 5 volts in each capacitor.
So just using two capacitor instead of one you are only moving half as many coulombs of electrons.
Therefore you will only have half the the capacitance or half the total Farads.

So adding capacitors in series, that is in a string, your capacitance will be decreased.
Adding capacitors in parallel you will get more capacitance. In fact the capacitance simply add up.

This is true in AC circuits and DC circuits.

When an electrolytic capacitor is reversed biased, that is charged backwards, tends to conduct and thereby prevent it from charging very much when connected is series with another capacitor. This characteristic is way I was able to connect two capacitors back to back, that is in opposing polarities, and be able to charge the string to plus or minus 500 volts. But in doing so I got less capacity.

 

[Electrobrains]

You are right concerning the calculations of capacitance!

What I yet cannot fully understand, is that their asymmetrical behavior does not influence the final capacitance. I have already measured on a couple of electrolytics and will post the results when I have more time.

Thanks for the help!

 

[Raven Luni]

I take it this is only necessary at high power levels? I've been using single electrolytics with small audio and other AC circuits for ages - and most diagrams you'll find do this.

 

[Harald Kapp]

If you look closely at an Elko used in an audio circuit you will find that it is DC biased such that the AC on top of the DC bias never reverses the total voltage across the Elko.
It's fine to use an Elko in this way.

 

[john monks]

Electrobrains, I don't understand your confusion. IF you mean that the capacitor conducts in one direction but not the other I can explain.
When you have two capacitor connected in series back to back and apply a voltage to the series string one capacitor charges up while the other tends to look like a short. So for this moment of time the capacitance will be greater because this is much like only one capacitor. After you are done charging both capacitors start looking like capacitors again and the capacitance of both affects the total capacitance. When you discharge the series string, that is bring the ends of the string back to zero volts relative to each other both capacitors now are charged to some voltage and the series string of capacitors assume a value of less than either capacitors. If both capacitors are 1 Farad your string will be 1/2 Farad.

It may be helpful to get a capacitor meter and try various series, parallel arrangements.