caps in reverse

[electrolust]

Forgive terminology and even conceptual errors.  After all this is the beginner section!

I am wondering about polarized caps in reverse.

One way you normally use a decoupling cap is to block DC bias voltage, in the direction of your AC signal output.  eg, to remove the DC offset from the output of a single ended (virtual ground) op amp.

But also it's not uncommon to block a DC bias voltage in the "forward" direction of the cap, while generating an AC signal in the "reverse" direction of the cap.  For example, http://www.epanorama.net/circuits/dynamic_to_electretinput.html .  The output direction of the mic is in the reverse direction of the cap.

As best I can understand, this works fine as long as the AC signal peak voltage is not higher than the DC bias voltage.  If that were to happen, the more positive node would be on the negative side of the cap and this would (could) fry the cap.  Is that a correct understanding?

Considering the forward-only case again, I also understand that an AC decoupling cap has another effect in that it creates an RC circuit with your load.  So the value of the cap has to be selected properly depending on your frequency requirements and the load resistance.  eg 10uF cap with 22k resistor as a low pass filter for audio frequencies.

Does the cap work the same way in reverse?  I mean in my example where you have an AC signal that is going in the reverse polarity direction of the cap (yet still under DC bias voltage present in the forward direction), will the cap in the reverse direction still form an RC low pass filter?

[Zero999]

When an AC coupling capacitor is correctly used, the voltage across it never goes below 0V. It will see the DC bias across it, plus a small AC ripple and if the capacitor is large enough, the voltage across it won't change much.

The capacitor on the output of an oscillator, amplifier etc. forms a potential divider with the load resistance. Ideally the capacitor needs to be large enough to couple the lowest frequency of interest without any noticeable attenuation. The capacitor's impedance needs to be negligible, compared to the load so the AC voltage across it will be small.

[rjeberhardt]

In the circuit you link to, the coupling (not decoupling) capacitor is biased with dc by the base-emitter voltage of the transistor as a dynamic microphone has a fairly low dc resistance.  Signal voltages can flow through a capacitor in both directions but for a polarised capacitor the signal should be smaller than the dc bias.

That circuit is wrongly drawn as a 220 nF capacitor would not be polarised.  It would be plastic film or ceramic.

Russell.

[Deathwish]

That circuit is wrongly drawn as a 220 nF capacitor would not be polarised.  It would be plastic film or ceramic.

Russell.

http://www.protostack.com/passive-components/capacitors-electrolytic/0.22uf-220nf-50v-electrolytic-capacitor

[Cliff Matthews]

A 220 nF capacitor would not typically be polarized. That site had only one smaller. http://www.protostack.com/passive-components/capacitors-electrolytic

[electrolust]

Signal voltages can flow through a capacitor in both directions but for a polarised capacitor the signal should be smaller than the dc bias.

I thought so.  Thank you for that confirmation.

My main question is, since a polarised cap does not "work" in reverse (doesn't block DC voltage), does it still form an RC filter the way it would if you use the cap in the "correct" direction.

[TimFox]

With reverse bias across the capacitor, it conducts like a forward-biased diode (before it blows up), and therefore does not act properly as a filter.

[electrolust]

By reverse bias, I take it that means the negative side of the capacitor is more positive?  From your answer I understand that there would be no filter function, up to the point where the capacitor fries and at that point all frequencies are blocked.

That is not quite what I am asking.  For my question, the net positive voltage is still on the positive side of the capacitor.  The p-p volts from the mic is an order of magnitude smaller than the DC bias, so in my question the signal does not end up reverse biasing the capacitor.  Even for the generic case that isn't necessarily such a small signal, I only am asking about the case where the AC signal peak isn't greater than the DC bias.  As this AC signal propagates "backwards" through the capacitor I wonder if the capacitor still forms a voltage divider / filter.

[JacobPilsen]

https://www.eevblog.com/forum/beginners/when-a-capacitor-is-put-in-backwards/msg487876/#msg487876

[rjeberhardt]

Provided there is dc bias in the correct direction, it doesn't matter which direction the signal flows it will still act correctly as a capacitor and can be used in a filter.  Remember though that the capacitance tolerance on electrolytic capacitors is quite high so the resultant filter's frequency response might not be very accurate.

Russell.

[Performa01]

1) AC doesn't have a 'direction'.
2) Aluminium Elecrolytic Caps for instance can take up to 2V reverse voltage and will still function as a capacitor properly.
3) Coupling caps are always designed to have a capacitance large enough so not developing any significant AC voltage drop.
4) Filters ... see below:

For filters, there is indeed a significant AC voltage across the capacitor. This is the reason why we don't use high-loss capacitors such as electrolytic caps in filter circuits. We usually want stability and reliability in filter circuits, hence using high quality foil capacitors up to a couple of microfarads. If we really need to go higher, there are non-polarized electrolytic caps particularly suited for crossover networks in audio applications.

[rjeberhardt]

1) AC doesn't have a 'direction'.

. . . but signal and power flow do.

Russell.