Super crazy problem - do I have an invincible capacitor?

[spargo]

So, I think I just figured out a problem one of my pedals was having after a new build.  The pedal worked for a few hours, then started cutting out.  I have five tantalum caps in the circuit.  For some reason, I could have sworn I read somewhere around here that tantalum caps were non-polar.  So as a result through all of my breadboarding and building the final circuit I just slapped them in there whichever direction (not noticing the + on one side of them, since they are indeed polarized).

Now here's where my luck came in.  In all my breadboarding and building the final circuit, by chance I happened to place all of the tantalum caps in the right direction!  With the exception of one on the final circuit...

Now this one that is in the wrong way still has an interesting thing about it.  THE CIRCUIT STILL WORKS some of the time with the cap in backwards.  I thought it was supposed to explode?  I had this thing running for a few hours the other day.  What's the deal?!!

[JKowalski]

Tantalum caps will only explode put in backwards (or even sometimes the right way!) if they can conduct large currents.

Localized heating in the capacitor changes the tantalum dielectric to a conductive material, which turns the capacitor into a low value resistor in that area - which draws more current, which converts more of the dielectric and so on until the cap is practically a short and conducts as much current through it as it can and explodes. (I think this is correct, reading from memory)

Now, for example if you have a high value resistor in series with it, then it can't conduct enough current in the worst case scenario to actually explode (limited by the resistor). If it is across the power supply though, it will end up shorting your power supply through the cap. Never use tantalum for bypass caps like this.

The best thing you should do is correct the problem. It may wildy change values or not work properly sooner or later. Just do it right!

[spargo]

What do you mean if it is across the power supply?  I have no tantalums in the power filtering circuit if that's what you mean.

And in general in a pedal circuit, if the audio signal is flowing through a tantalum cap, how do you know which direction to put it in?  It's obvious if one leg of the cap goes to ground, but what if it's fully in the circuit?

[jasperoosthoek]

That happened to me too. I mistakenly thought a line meant a big minus sign but there was a little plus on the end of that line. It worked for several days and then broke down (shorted). I found out that they are very tough indeed.

[R.G.]

As demolition experts know, sometimes even blasting caps and dynamite won't go off. Doesn't mean they are inert, though. 

[JKowalski]

What do you mean if it is across the power supply?  I have no tantalums in the power filtering circuit if that's what you mean.

And in general in a pedal circuit, if the audio signal is flowing through a tantalum cap, how do you know which direction to put it in?  It's obvious if one leg of the cap goes to ground, but what if it's fully in the circuit?

Yeah, if you use it for power filtering. If it is connected one end to V+ and the other end to ground with nothing in series with it.

For audio you will typically have the signal riding on a bias voltage. One end of the cap will often be the audio signal riding on ground (for example an input) and one end will often be the audio signal on say 4.5V (for op amps) or a voltage from biasing a transistor stage. This will be the plus end.

If you have the audio riding on the same voltage on each side (say Vref to Vref) then your best bet is to use a nonpolar capacitor.

[John Lyons]

In other words, the side of the cap with the highest voltage is the + side.

[JKowalski]

In other words, the side of the cap with the highest voltage is the + side.

Well said 

[John Lyons]

Meant to say that the positive of the cap should be
placed at the point in the circuit with the highest voltage.

[spargo]

Well, I bit the few dollar shipping bullet (and the wait) and thought it would be wise to stay away from the tantalums and get some metal films, as long as they fit on the PCB well.   It's too confusing for me to figure out their polarization correctly.  Thanks for the help guys!

[jasperoosthoek]

Haha, I replaced two working bipolar caps with tantalum caps today because the big Small Clone board didn't fit in the small box I used. The 24mm Alpha pots plus bipolar caps on top were just too high and the back coved didn't close.

The plus side on the Ta cap has a longer lead wire. No reason to check what's printed on them. 

[amptramp]

Tantalum caps will only explode put in backwards (or even sometimes the right way!) if they can conduct large currents.

Localized heating in the capacitor changes the tantalum dielectric to a conductive material, which turns the capacitor into a low value resistor in that area - which draws more current, which converts more of the dielectric and so on until the cap is practically a short and conducts as much current through it as it can and explodes. (I think this is correct, reading from memory)

Now, for example if you have a high value resistor in series with it, then it can't conduct enough current in the worst case scenario to actually explode (limited by the resistor). If it is across the power supply though, it will end up shorting your power supply through the cap. Never use tantalum for bypass caps like this.

The best thing you should do is correct the problem. It may wildy change values or not work properly sooner or later. Just do it right!

I think you are quoting fromwhat I said a long time ago on another thread.  The main feature of tantalum is that it can be anodized to a pentoxide which offers a pinhole-free film.  This translates to very low DC leakage current when used as a capacitor dielectric.  But if sufficient curent flows through a defect, it can change to the dioxide, which is conductive.  This starts a firestorm around the capacitor that results in fire or explosion.  The best use for tantalum caps is for low-frequency timing such as an LFO where you are guaranteed to have series resistance.  The low DC leakage offers excellent performance in this limited application.

[R.G.]

The plus side on the Ta cap has a longer lead wire. No reason to check what's printed on them. 

... unless someone, sometime, somewhere had a pair of cutter and ill intentions. 

[PRR]

Tantalum will take "small" reverse voltage. It used to be used where the final DC was uncertain: supposed to be "zero", but might come out a tenth-volt either way.

It may "work" up to a few volts reverse, depending on other things (like how bad it can be before "not-working").

> WORKS some of the time

If this is an output DC-block cap, and the next box/amp has an input DC-block cap, then the tantalum can go totally limp and leaky yet the rig still "works". Into another box/amp without an input DC-block cap, it is likely to throw that next input out-of-bounds. So it works "sometimes".

They explode only when there is enough energy available. Used as a coupling cap with an op-amp or a 5K collector resistor, the available energy is too small to even warm the cap. Across a 9V battery, the battery will go flat before the cap gets hot enough to burst. A good wall wart can do it if the tantalum is directly across the wart's full output.

An Aluminum (ordinary "electrolytic") cap can also explode. "Re-cap" your 100 Watt tube amp. Put the first main power cap in *backward*. Don't try this at home: it stinks and spews snot-paper all over the room. OTOH, put a bias cap in backward, the tubes may shatter, but the small bias supply usually won't make more than a "pfft.."

There ARE non-polar tantalums. They are expensive and rare.