Cap polarity, non-polarity, DC blocking, amp circuitry.

[Thecomedian]

I don't think I'll be able to find any polarized .002u caps.

What is the point of the polarity of the DC decoupling caps, or is it that if you do use a polarized cap, it must go in the proper way, and non-polarized caps don't care which way they go?

[R.G.]

I don't think I'll be able to find any polarized .002u caps.

And I'm pretty sure you won't.

What is the point of the polarity of the DC decoupling caps,

I'm not sure what you mean by that. Can you give an example? Did you see a schematic with a 0.002 cap marked for polarity?

or is it that if you do use a polarized cap, it must go in the proper way, and non-polarized caps don't care which way they go?

This is certainly true.

[Paul Marossy]

One thing I have always wondered about is when you have to use say a 10uF polarized cap in a signal path, how the heck do you determine which way it has to be put in the circuit? Does it really matter in this case?

[R.G.]

One thing I have always wondered about is when you have to use say a 10uF polarized cap in a signal path, how the heck do you determine which way it has to be put in the circuit? Does it really matter in this case?

Although modern caps are much better in this regard than older ones, it is never good practice to install a polarized electrolytic cap in a circuit with the same DC voltage on both sides of it; that is, zero volts across the cap.

Electro caps need the polarizing voltage to keep self repairing. Plus, with a signal of any size at all, one polarity of signal reverse biases the cap, and even if the signal is small ( say, half to one volt or more) it speeds up the erosion of the oxide layer.

So the best answer is a step sideways: don't do that. If you have a polarized cap in the signal path, and it's not obvious which end is more positive than the other, replace it with a non-polar.

[gritz]

One thing I have always wondered about is when you have to use say a 10uF polarized cap in a signal path, how the heck do you determine which way it has to be put in the circuit? Does it really matter in this case?

Although modern caps are much better in this regard than older ones, it is never good practice to install a polarized electrolytic cap in a circuit with the same DC voltage on both sides of it; that is, zero volts across the cap.

Electro caps need the polarizing voltage to keep self repairing. Plus, with a signal of any size at all, one polarity of signal reverse biases the cap, and even if the signal is small ( say, half to one volt or more) it speeds up the erosion of the oxide layer.

So the best answer is a step sideways: don't do that. If you have a polarized cap in the signal path, and it's not obvious which end is more positive than the other, replace it with a non-polar.

In addition to this, electrolytics that aren't subject to a polarising voltage can exhibit pretty bad hysteresis (memory) effects, similar to the dielectric absorbtion of certain grade ceramics (and measurably rather worse in some cases I've seen). These thing worry me - and I'm pretty slack about the stuff that the audiophiles worry about.

At least in our 9V single supply pedals it's usually pretty easy to ensure a polarising voltage (and our high impedances mean that small input coupling caps are fine and we don't need flat response down to 0.0001Hz anyway), but with split supplies those big input / output coupling caps in a lot of "pro" audio line level gear are just floating about in no-man's-land. Ugly.

[Paul Marossy]

Although modern caps are much better in this regard than older ones, it is never good practice to install a polarized electrolytic cap in a circuit with the same DC voltage on both sides of it; that is, zero volts across the cap.

Electro caps need the polarizing voltage to keep self repairing. Plus, with a signal of any size at all, one polarity of signal reverse biases the cap, and even if the signal is small ( say, half to one volt or more) it speeds up the erosion of the oxide layer.

So the best answer is a step sideways: don't do that. If you have a polarized cap in the signal path, and it's not obvious which end is more positive than the other, replace it with a non-polar.

Interesting... especially true for tantalums which are rabidly polar. Yeah, I like the idea of non-polar electros better for various reasons. I often see polarized caps in the signal paths of various things, and I've never noticed there being an obvious convention about which end goes where, which is why I ask.

[Thecomedian]

One thing I have always wondered about is when you have to use say a 10uF polarized cap in a signal path, how the heck do you determine which way it has to be put in the circuit? Does it really matter in this case?

Positive side of the cap goes towards positive voltage, negative side goes towards negative voltage or ground

[Paul Marossy]

One thing I have always wondered about is when you have to use say a 10uF polarized cap in a signal path, how the heck do you determine which way it has to be put in the circuit? Does it really matter in this case?

Positive side of the cap goes towards positive voltage, negative side goes towards negative voltage or ground

Duh. But how about in a AC signal path that has both a positive and negative swing? That's what confuses me.

[Thecomedian]

If one side or the other has higher positive voltage, I'd put the negative terminal on the opposite side.

When positive is pointed towards + side of cap, holes/carriers move to charge the cap up, while electrons move to charge the negative side. When voltage swings negative, electrons move away from the negative side, and holes/carriers move away from the positive side.

Just like diodes and transistors, the Positive Side is actually negatively charged, and negative side is actually positively charged.

[Paul Marossy]

If one side or the other has higher positive voltage, I'd put the negative terminal on the opposite side.

That part makes sense.

[gritz]

One thing I have always wondered about is when you have to use say a 10uF polarized cap in a signal path, how the heck do you determine which way it has to be put in the circuit? Does it really matter in this case?

Positive side of the cap goes towards positive voltage, negative side goes towards negative voltage or ground

Duh. But how about in a AC signal path that has both a positive and negative swing? That's what confuses me.

Well... you can just stick two electrolytics in series, with similar "poles" facing each other. You can then attach the junction to a polarising voltage via a high value resistor (making sure not to inject any noise). Then the cautious audiophile might attach a film cap in parallel with the polar cap pair. Or you can try to design away the need for a large cap (not always possible when e.g. you can't predict the input impedance of whatever-it-is that might be plugged into the thinger that you're building). Or you can just wing it, like everyone else does. It's only a couple of volts at a trivial current, after all.  

[wavley]

I'm going to complicate things just for fun.

Remember when foil caps had the outside foil end marked?

[gritz]

I'm going to complicate things just for fun.

Remember when foil caps had the outside foil end marked?

That's polystyrene "Swiss rolls", rather than electros in a can, isn't it? Useful for shielding tho.

[R.G.]

But how about in a AC signal path that has both a positive and negative swing? That's what confuses me.

There can be no confusion. The voltage on a polarized electrolytic cap must always be more positive on the + marked side. Full stop. If there is a signal where the *signal* causes the + side to become more negative than the (-) side, then it speeds up the aging/failure of the cap. Even if the reversal is instantaneous, not constant.

With the voltage between + and - sides at zero, the cap ages at a certain rate, and will eventually be no good. This is the shelf life, and it's generally 5-10 years.

With a voltage on the + side more positive than the - side, the cap will last longer than its shelf life.

With a voltage on the + side that is more negative than the - side, the cap is aged faster than the shelf life. More than a fraction of a volt, and especially with lots of current available, the life in this mode may be seconds or less. With tiny reverse voltages, such as signals alternating, the damage is slower.

This is one of those things where designers get away with it for a while, perhaps long enough for the warranty (if there is one) to expire. The caps will take small reverse voltages for short times before failing.

But if there is a polarized electro cap in a design with 0V dc on each side and signal through it, it is a poorly thought out design. Electro caps survive best/longest if they are constantly held at some large fraction of their rated voltage all the time.

[Paul Marossy]

The voltage on a polarized electrolytic cap must always be more positive on the + marked side. Full stop. If there is a signal where the *signal* causes the + side to become more negative than the (-) side, then it speeds up the aging/failure of the cap. Even if the reversal is instantaneous, not constant.

OK, got it. Thanks for taking the time to explain that, it's something that I never quite understood until now.

[wavley]

I'm going to complicate things just for fun.

Remember when foil caps had the outside foil end marked?

That's polystyrene "Swiss rolls", rather than electros in a can, isn't it? Useful for shielding tho.

Pretty much all of the dielectrics of film caps used to be that way (I still see it on a lot of new caps too) and it is useful for shielding.

But it is, even though it's not a polarity issue, an issue of orientation in the circuit.

[Lurco]

? http://www.diystompboxes.com/smfforum/index.php?topic=102654.msg911131#msg911131

[chromesphere]

? http://www.diystompboxes.com/smfforum/index.php?topic=102654.msg911131#msg911131

I was waiting for one of the experts to offer an answer, but here goes.  What voltage appears at the base of q1?  I'm assuming if E is connected to ground it will be 0.5-0.6v?  If so your guitar signal is probably going to be around / higher then that.  Which makes the base less than your guitar signal.  That's my interpretation of it, although that doesn't mean much because I'm probably wrong.

[R.G.]

By the same reasoning, the input cap on those does look to be backwards. The base-emitter voltage of the input will be about 0.5V positive.

The thing probably works for quite a while, as the reverse voltage is so low, but it's not really good for the capacitor, and you could expect early wear-out, depending on how hard the manufacturer worked on making it reverse-tolerant - or how lucky they were.

For best life in a low-voltage, low-signal setup like that, I would use a non-polar cap.

[Paul Marossy]

? http://www.diystompboxes.com/smfforum/index.php?topic=102654.msg911131#msg911131

I was waiting for one of the experts to offer an answer, but here goes.  What voltage appears at the base of q1?  I'm assuming if E is connected to ground it will be 0.5-0.6v?  If so your guitar signal is probably going to be around / higher then that.  Which makes the base less than your guitar signal.  That's my interpretation of it, although that doesn't mean much because I'm probably wrong.

See, that's a good example of what I was getting at. It's not always so clear about what the polarity of the cap should be.

The voltage appearing at the base would depend on how hot your pickups are. These newer high output pickups can have a peak voltage of at least around 500mV (or even 1V as I understand it). But low output vintage pickups are probably one third to half that (of 500mV).

So why don't manufacturers use bipolar caps instead in these scenarios? Let me guess, it's probably because they cost a few cents more?