Can I replace Polarised Electrolytic caps with a Non polarised one?

[Izzy]

Can I replace Polarised Electrolytic caps with a Non polarised one?

I am confused. I think I have heard we can use it. But i have also ehard in some cases its better to use Poalrised when they are mentioned Polarised in the schematics.
I was working on sansamp gt2 and I mistakenly bought a non polarised caps instead of polarised ones. That Polarised electrolytic is at output buffer of Sansamp GT2.

So can I replace it? I was about to sodler. 

[rockgardenlove]

I can't think of an instance in which you couldn't.

[Izzy]

So why do they put electrolytic polarised ones in the circuit?
Is there any particular reason?

[R.G.]

Can I replace Polarised Electrolytic caps with a Non polarised one?

Yes, you can use it.

But i have also ehard in some cases its better to use Poalrised when they are mentioned Polarised in the schematics.

Polarized are cheaper and a little smaller for the same value of capacitance, so they are "better" in some senses of the word.

So why do they put electrolytic polarised ones in the circuit? Is there any particular reason?

Polarized are smaller and cheaper for those applications where you can use them.

[Mark Hammer]

Before we proceed too much farther, shouldn't there be a distinction made between caps in the signal path and those on the supply lines?  Not that I expect anyone to spring for, fit, or even find a 100uf NP unit, but I would think that where a 1uf NP can work just fine in the signal path as a replacement for a 1uf electrolytic, perhaps it might be different as a decoupling cap on the Vref line?  Or am I being too conservative and naive?

[GibsonGM]

I get thoughts about this topic sometimes...not that it keeps me up at night, but still, LOL....
The polarization of an electrolytic cap is a function of the part, not the circuit it's to go into (such as a diode, which has a definite polarity).  You're not supposed to use them on AC, either, but sometimes I do and they are fine (altho in 5 yrs. from now, who knows? The dielectric may break down).   

My logic would say that a NP unit is "better" than a polarized one because of this inability to be used at AC.  Basically, I will use NP anywhere, and try to keep polarized isolated to DC (Vref etc), and always have good results.  As mentioned, cost and availabilty of higher uF are the only issues I worry about.  I am not sure how they affect noise, though....

Somewhere I also heard that there is some percentage of signal you should operate an electrolytic at (example, 100mv guitar signal and 16v electrolytic = bad, not enough voltage to keep the dielectric healthy).  Worth looking into, maybe?

[oldrocker]

Okay, what about the TS-808 schems using the two NP 1uF caps.  What would happen if you were to use polarized caps there?  I was lazy and I was going to use polarized caps since it's what I had laying around at the time.  I changed my mind at the last minute so I wouldn't get confused at the debug stage.  All schems for TS builds show the NP's used so assume it's important.  Has anyone tried it?  If so what was the results.

[R.G.]

Electrolytic caps are made by growing an oxide layer on aluminum foil. The oxide thickness is a function of the voltage applied, because the applied voltage actually pulls ions onto the aluminum through the previous oxide layer to oxidize more aluminum. Each weak spot gets plugged by more oxide growing because current flows through there, making more oxide.

If you take aluminum and put it into a proper oxidizing bath, then put an electric charge between aluminum and the bath, oxide grows like crazy for a while, but as the thickness grows and any holes and weak spots get plugged, the withstand voltage gets bigger and bigger until it equals the forming voltage on the bath, then it stops. For the purists, it does not really stop, it heads for zero growth asmyptotically. If you then lower the voltage on the bath, current effectively does stop because the oxide can hold more voltage than you have across it.

So if you want a 25V cap, you grow the oxide to maybe 35V. That's the "surge voltage". 25V is where the oxide will resist leakage to the guaranteed specified amount after some number of years of use under recommended conditions. 25V is then the rated voltage. I made the numbers up to illustrate.

If you use the oxide at a lower voltage, it eventually gets thinner, and un-forms itself down to the actual applied voltage. This is a slow process. The process is made faster by having zero volts on the cap, and even faster if you reverse the voltage. If you hold the voltage on the cap at the recommended max working voltage and don't otherwise abuse the cap by heating, it lasts for a very long time - decades at least. There is actually a competing re-forming process which happens if you use the correct electrolyte in the cap while it's working. If you leave it sitting on a shelf at 0V, the cap will eventually un-form itself until it has places which will punch through if you suddenly apply the rated voltage.

If you reverse the voltage, the cap actively un-forms. The oxide layer is actually a form of diode. Metal oxide rectifiers were a competing technology for tube rectifiers for a long time, usually in copper oxide. In aluminum the diode effect exists, and lets through lots of current in the reverse direction, and that large current eats at the existing oxide, which is why copper was used for metal rectifiers, not aluminum.

NP aluminum caps are made by forming BOTH plates. That is, each side resists a positive voltage being applied. The other side is in diode mode, conducting, but the one that is correctly polarized keeps current from forming, so you don't get destructive un-forming. This is also why two ordinary 1uF caps hooked up back to back make a 1uF NP equivalent, not a 1/2uF equivalent. On each half cycle, only one is capacitating, not both. The other is conducting.

Electros and NPs in the signal path/supply line
The same principles apply, the currents are just smaller. An NP cap and a polarized electrolytic cap can both be used in polarized applications. The NP cap is just always used one-sided,which is fine with it, as the oxide on the other side may deteriorate, but that doesn't matter as the working oxide is kept in shape by the applied DC voltage. NP caps do have have a slightly higher ESR, but it's not that bad. NP caps have the same number of foils/plates - 2 - it's just that both foils have an oxide layer grown on them, not just one. They'd probably cost the same (might as well oxide all your aluminum, not just half) if the volumes were bigger. In power bypass applications, you really would like to have lower ESR in rectifying applications, so bulk rectifying filters are best not NP. But for decoupling, sure, NPs work, they're just bigger and more expensive because of the lower sales volume and lack of competition to make smaller ones. For signal applications, you can use polarized and NPs interchangeably in places where there is a DC voltage to block that's bigger than the applied AC signal. If the applied AC signal has peaks that reverse the instantaneous voltage on the cap, you must use NP for long life.

100uF NPs are commonly available for use in speaker crossovers.

This is completely aside from the esoterica discussion of polarized caps having grit and grunge, dielectric absorption and other audio-uglifying effects. We're only discussing the cap's health here, not the inner definition of the notes and sharp focus of the sound stage.

You're not supposed to use them on AC, either, but sometimes I do and they are fine (altho in 5 yrs. from now, who knows? The dielectric may break down).

The current flow in the reverse direction is an exponential function of the applied reverse voltage. For small reverse voltages, not much current flows and the oxide breaks down at a slow rate. In many applications, the device is mostly sitting on the shelf with only occasional periods of use, so if the reverse current*time is small, the cap may last a longish time. If the reversed voltage is intermittent (AC signals), small (say 0.1 to 0.5V for aluminum), and occasional ( two hours a week) then the apparent life may well be many years. It's different if the voltage is DC, large, and continuous.

My logic would say that a NP unit is "better" than a polarized one because of this inability to be used at AC.

It's "better" if what you want is AC use. If you have the situation where a polarized cap can be used, the polarized cap will do the same job for cheaper and be slightly smaller. Once again, it comes down to what you think "better" means. If you have a DC rectification situation, especially at high frequency, the polarized cap is better because it will have lower ESR losses and live longer because it doesn't self-heat as much, as well as being cheaper and smaller.

I will use NP anywhere,

Fine for signals and non-rectification use.

and try to keep polarized isolated to DC

Which constitutes the recommended application.

I am not sure how they affect noise, though....

They don't. Caps do not generate thermal noise other than in their ESR.

Somewhere I also heard that there is some percentage of signal you should operate an electrolytic at (example, 100mv guitar signal and 16v electrolytic = bad, not enough voltage to keep the dielectric healthy).  Worth looking into, maybe?

Capacitor makers publish application guides. Good reading.

Ideally you should specify a cap with a working voltage between 75% and 90% of the rated DC voltage. This keeps the re-forming process I referred to above healthy. However, using an electro rated at 16V with 3V across it and 100mv of signal on top of that will work fine for a long time, maybe so long that you don't need to worry about it.

There is another consideration - capacitance tolerance.

Electros are not tight -tolerance devices because the capacitance changes with time. As the cap un-forms, the oxide gets thinner and capacitance goes up as the withstanding voltage goes down. As you form oxide on an aluminum surface, the capacitance goes down as the oxide gets thicker. Working voltage and capacitance are competing processes, which is why cap makers rate the "goodness" of caps as the C*V product. Another measure is CV product divided by volume.