About tantalum capacitors

[caius]

Hi all,
I repair arcade boards and electronics just for fun and hobby since some years.Today I repaired a japanese PC, you can read log with pictures here:

http://www.jammarcade.net/nec-pc-9821ap2u8w-repair-log/


A doubt is floating in my head after done this repair : can a tantalum capacitor leak its dielectric like electrolytic ones do usually?Judging from what I've seen inside this computer I would say yes..

P.S.
Machine came out in 1993 so it's more than 20 years old.

[T3sl4co1l]

Impossible: dipped dry tantalums are just that, dry.  Perhaps they're some bizarre dipped electrolytic instead?

Tim

[caius]

Thanks for reaply.Obviously also to me this sounded very odd but it's the only aswer I could find.
Yes, I thought me too they can be some kind of electrolytic capacitors whose shape reminds of tantlaum ones.

[c4757p]

Cut one open and see what's inside?

[caius]

Ok, I will do.Anyway these capacitors seems undamaged as they never leaked.I measured their capacity and ESR out-of-circuit and values are normal.But, as you can see from pictures, corrosion on PCB is located just in correspondsnce of them, no other near components could have caused it.Then, I don't think in hardware design electrolytic capacitors have been ever used as decoupling of digital ICs, usually they are ceramic or tantalum.

[jimmc]

That looks like some kind of glue used to hold the caps in place, has been known to become conductive with age, I wonder if this was the problem here.
Properly used tantalums have no wear-out mechanism unlike aluminium electrolytics.

Jim

[caius]

That looks like some kind of glue used to hold the caps in place, has been known to become conductive with age, I wonder if this was the problem here.
Properly used tantalums have no wear-out mechanism unlike aluminium electrolytics.

Jim

Yes, there was some glue to hold the capacitors in place but this is not the responsible of corrosion, I cannot see traces of this flowing towards the TTLs, that's impossible.

[MAS3]

Hi.

The tantalum caps i've seen that have failed were very easy to identify.
A tantalum cap has some resemblance to the head of a match.
If it fails, it will fail spectacular, also resembling lighting a match.

Glue might have some acid in it, which will corrode copper surfaces over time.
jimmc's suggestion sounds about right to me.

[Yansi]

Hi
I think tantalum caps (especially the old ones) can leak aswell. Some old tantalum caps do have liquid electrolytic. I've just opened an old (1986) catalogue and confirmed, that some of them are described as "liquid dielectric tantalum". I've heard somewhere, that they contain some acid (sulphuric or phosphoric?).

I've looked into my junk boxes and found some specimens.  As the catalogue says, the body is pure silver, leads are tin plated copper. Sintered anode, liquid electrolyte. And sure, if you squeeze them with a pliers, something nasty leaks out. These must have been very expensive, back in the days...

[helius]

These are called wet tantalums. They contain a liquid electrolyte such as sulfuric acid.

http://www.mouser.com/Passive-Components/Capacitors/Tantalum-Capacitors/Tantalum-Capacitors-Wet/_/N-75hr5/

http://www.vishay.com/docs/40021/wtintro.pdf

The case is made of either solid silver or tantalum, and hermetic versions have a glass frit. They are very expensive.

Edit: I just looked at your pictures and those don't look like wet tantalums. Those are epoxy dipped, which to my knowledge always means a solid electrolyte. Also leaked sulfuric acid wouldn't be thick and yellow like that.

[Yansi]

No yellow stuff goes out of it, just relatively clear acidic (by the indication paper). And sure, these on the photo are wet tantalums, confirmed by their datasheet.  And yes, the plug is epoxy.

I'll try to translate the § from the old catalogue:

Tesla TE154:
Tantalum capacitors with liquid electrolyte, hermetically sealed in a silver package. Wire leads are tinplated copper. The negative pole lead is connected to the case, positive pole is fed through a seal made of epoxy rosin. The package is lacquer coated,...

[coppice]

Wet tantalums are used by the ton is defence systems. Dry aluminiums are much cheaper, and don't have the fire hazard issues of tantalum capacitors. Once people launched those, in the 80s, the civilian market for wet tantalums pretty much diappeared. You'll find quite a few in ancient civilian equipment designed to run 24 hours a day.

If a wet tantalum bursts its usually due to physical damage, rather than corrosion. The contents are pretty nasty, so be careful.

[caius]

Here attached a close picture of one of these 'misterious' capacitor.As said, it seems to me a nornal tantalum one.

[tautech]

Here attached a close picture of one of these 'misterious' capacitor.As said it seems to me a nornal tantalum one.

Nothing mysterious: a tantalum bead, looks like a match head and behaves like a match head. 

Keep the applied voltage WELL below the labeled value.

[caius]

Yes, I was joking, they are tantalum, no doubts atall.But what about the corrosion on the PCB?Could it be due these leaking?

[tautech]

Yes, I was joking, they are tantalum, no doubts atall.But what about the corrosion on the PCB?Could it be due these leaking?

I must be one of the lucky few to have not met "Wet Tantalums".
Long may that continue IMO.
I'm sure one day my luck will run out. 

[SeanB]

Not the bead tantalum, but the glue they used or a coating that they applied to the capacitor after assembly that has corroded the board.

As to the wet tantalum capacitors, the ones with the glass seal are going to last forever, but the ones in a package with an epoxy or rubber seal are a lot cheaper. I had a fun year changing out a few thousand of those that failed as the seal degraded with age, thermal cycling and pressure changes.  The spares in the store were also failing, and the first time I sent them back as failing capacitance and ESR limits, the inspector did a __VISUAL__ check and sent them back. The second time they went back sans leads. I did then request ( and got) a computer printout of everything in the stores computer, along with a quantity and location, of everything with the letters cap in it. Took me 3 months to read that pallet of phone books that came as a reply, and the incinerator was pretty busy with the used books as well. Stores loved me after that though, I think there are still orders trickling in still.......

[caius]

Not the bead tantalum, but the glue they used or a coating that they applied to the capacitor after assembly that has corroded the board.

It's really incredible that NEC, a Japanese giant, didn't think about the risk of applying a glue that would have corroded the PCB with age.I have another same computer, same production year and same tantalum capacitors glued on pcb but there is no trace of corrosion.Probably also environmental conditions have had their role unleashing some chemical reaction in the glue.

As to the wet tantalum capacitors, the ones with the glass seal are going to last forever, but the ones in a package with an epoxy or rubber seal are a lot cheaper. I had a fun year changing out a few thousand of those that failed as the seal degraded with age, thermal cycling and pressure changes.  The spares in the store were also failing, and the first time I sent them back as failing capacitance and ESR limits, the inspector did a __VISUAL__ check and sent them back. The second time they went back sans leads. I did then request ( and got) a computer printout of everything in the stores computer, along with a quantity and location, of everything with the letters cap in it. Took me 3 months to read that pallet of phone books that came as a reply, and the incinerator was pretty busy with the used books as well. Stores loved me after that though, I think there are still orders trickling in still.......

This is a nice story
 

[coppice]

It's really incredible that NEC, a Japanese giant, didn't think about the risk of applying a glue that would have corroded the PCB with age.I have another same computer, same production year and same tantalum capacitors glued on pcb but there is no trace of corrosion.Probably also environmental conditions have had their role unleashing some chemical reaction in the glue.

The glue served its purpose, and the level of corrosion within the working life of the machine was probably quite small. I agree that its amazing how well NEC chose their glue. 

[askjacob]

I have seen those leak. Not even from "over voltage" but just from long term ripple, at close to running at near their running voltage. They leave a browny-orangey goop like the very old contact adhesive looking goop they used to use before hot-snot or silicon became all the rage.

If you had to replace or repair the power supply as well, then that would explain the strain on them. If you haven't check that yet, then I highly recommend it - not just the voltage, but the ripple too - as that is what will cook them eventually.

In another story, I had an older PC (well, mum did) with a whole bunch of them (about 24) on it instead of electros for the rail supplies on board. The power supply finally crapped out after about 11 years of low performance email and browsing, and dumped ~150v DC into the motherboard. Every one went off like popcorn, and the few electros on board left confetti inside too.

She called me to say her PC is not working - and when I went to fix it I saw the carnage (smelt it first!) - she didn't hear a thing! I can only hope she was getting a cuppa at the time.

Also killed the mouse, keyboard and the NIC. I guess it was a time for her to upgrade...

[tautech]

The power supply finally crapped out after about 11 years of low performance email and browsing, and dumped ~150v DC into the motherboard. Every one went off like popcorn, and the few electros on board left confetti inside too.

Yep, evil little B'sards.

My rule of thumb is NOT to use them in excess of 66% of their voltage rating.

[voltz]

Interesting thread. In all my years of repairs, which are many, I have never seen or heard of a wet tantalum capacitor. 

Things to watch out for with dry Tantalum caps:
over voltage = bang. Reverse polarity = bang, high age = eventual short circuit.

In my designs, I try to avoid them as much as possible.