Repair/Restoration of Vintage Computer devices with Tantalum Capacitors

[CkRtech]

I pitch this thread as a philosophy question regarding old computer components. I have in recent years started restoring and building old computers as a hobby - mostly working with what I have had in my possession for a couple of decades. I am looking at a new restoration project within the next few months with components that stretch into a period (<= 1994) of using of candy-dipped tantalum caps on expansion cards and motherboards. This is where my vintage computing hobby gazes at a potential pit stop with my electronics hobby.

These caps have potentially not seen power for 25+ years. I have seen one user experience an IBM 5150 have one tantalum on its motherboard "go popcorn" - leaving a noticeable mark around the area where it used to reside.

I would prefer to avoid situations like those, but my question to you guys is - how proactive should one really be? In the case of acquiring an IBM 5150 that hasn't been powered up in goodness knows how many years, I am convinced that those caps should just be replaced proactively. But what about something newer (1993)?

I have attached an example of a early-mid nineties video card containing some of the caps. It represents a component of a computer I should reacquire from a relative soon. In addition, I believe the motherboard also might have a few. We're looking at 486 era, here. I last had it powered up in 2010 - and nothing happened at that point. Still - it certainly won't be the last bit of hardware I acquire like this, and for sure won't be the oldest.

[helius]

Tantalum capacitors are "self healing", meaning that they regrow the barrier layer if there is a crack in it. The problem is that this healing process releases oxygen and heat, which in large amounts can allow combustion. It only happens if there are cracks in the dielectric, and those are not normally present simply because the assembly is old: age does not contribute to tantalum failures.
The short of it is that if you wouldn't replace all the tantalums on a brand new card, there is no reason to do so on an old one.

[woodchips]

Good grief! Age doesn't contribute to tatalum failures! So what does? Come 20+ years and popping tants is almost a musical score.

Must say my direct experience does not reflect that statement, but what else could it be?

[helius]

It is entirely possible that the operating voltage drifts high on boards where parts have aged. For example, if there is a trimmer pot on the power supply that controls the output level, and it has gotten dry and dusty. This can kill tantalums when the electric field strength is high enough to begin the "self healing" process in locations it has not yet happened.

[eugenenine]

looks like VLB.

I have an Amiga 2000 on my bench I'm trying to revive.

[KE5FX]

What kills tantalum caps is dI/dt, so look at the schematic and/or PCB.  Any tantalums that are fed with a lot of series inductance (and ideally a few ohms of resistance) can probably be left in place. 

There's a good case for replacing any others, especially the ones that are parked directly across a high-current supply rail.  I'd replace those with either modern OSCON electrolytics, or tantalums whose voltage is derated by 2:1 or 3:1.

[stj]

i work on stuff a hell of a lot older with lots of tants - generally from 1977 to 1983.

they dont fail from age, what shorts a tant is psu problems, spikes and reverse spikes especially.
and that brings me to another point, a tant will only burn if it has a large current across it, if it's not across a power rail it will usually not burn,
and even if it is, if the psu is protected against shorts then it still wont burn.

too many stupid scare stories are going around about tants.
i have on the other hand, seen some pretty spectacular electrolytic failures caused by long term storage,
i wouldnt ever lean over any pcb when first powering it up after several years - i like my face with 2 eyes on it!

[CkRtech]

Happy to hear these responses. I don't mind replacing things when necessary. I also like both my eye balls. Some of the stuff I run on my test bench and not inside a case. I also expect to get some older stuff closer to that 1977-1983 range before too long, and I am just curious about some of the stuff I have watched/read over the last week after never having given it a second thought.

Some guys talk about powering up stuff every few years or so (and they are collectors before electronics hobbyists). I am more likely to be acquiring stuff directly from garages and attics after long periods of time, moisture, heat, etc., so I thought I would run it by you guys.

So don't overthink it, eh?

[stj]

if it's been in questionable storage, and that means someone elses storage,

clean it, visually inspect it for damage,
then leave it someplace warm for a day to get rid of any moisture you cant see.

then i like to power up stuff i dont trust with a sheet of acrylic from an old lcd panel laying over it - it contains any surprises!!

as for the "every few years" type of comments, i suspect that's to do with electrolytic caps de-polarising.
unfortunatly, if a cap is drying up inside, it will go anyway.
hence the acrylic sheet 

[Kjelt]

Normal electrolytes have to be reformed when they aren't used for 10s of years, how about these (wet) tantalums ?
They can self heal but is that the optimal way to bring them back from storage ?

[tggzzz]

Tantalum capacitors are "self healing", meaning that they regrow the barrier layer if there is a crack in it. The problem is that this healing process releases oxygen and heat, which in large amounts can allow combustion. It only happens if there are cracks in the dielectric, and those are not normally present simply because the assembly is old: age does not contribute to tantalum failures.
The short of it is that if you wouldn't replace all the tantalums on a brand new card, there is no reason to do so on an old one.

Really?

Based on experience and having just cleaned a PCB where two electrolytics had corroded and destroyed some tracks on a PCB, your statement seems a little, um, dogmatic. They were high quality "silver" metal cased tants in military equipment, and had discharged their contents while in storage for quite a few years.

[stj]

we are talking about SOLID tantalum beads, not the old wet ones.

i would trust a wet tantalum about as much as a wax-coated paper cap.
which is not a lot!

[amyk]

we are talking about SOLID tantalum beads, not the old wet ones.

i would trust a wet tantalum about as much as a wax-coated paper cap.
which is not a lot!

Wet tantalum caps, the hermetically sealed type, are actually high-reliability parts commonly used in military/aerospace. They are also extremely expensive.

[stj]

only as good as the casing and seal.

[tggzzz]

we are talking about SOLID tantalum beads, not the old wet ones.

You might be, but that isn't in the thread title, nor did you bother to mention that caveat.

i would trust a wet tantalum about as much as a wax-coated paper cap.
which is not a lot!

Others have pointed out just how mistaken you are. I even stated they were "high quality" and "military" grade.

The equipment in question contains a >2GHz CRT oscilloscope which runs on batteries and can be submerged in water! One of the service manuals indicates it was specially written for use with a Patriot missile defence system. Hence it doesn't piss around with crap components.

[MosherIV]

Hi

I have asked my colleagues at The National Museum of Computing what the process is for dealing with tantalum capacitors when restoring machines. They said
1. Visually check for damage, replace any that have signs of physical damage
2. Do basic checks for short on psu etc
3. Power up briefly to see if anything goes bang or phut
3a repair anything that went  phut or bang.
4 Power system up and let it boot.
They said there is no need to replace tants unless they go bang.

Hope this helps.

[helius]

Really?

Based on experience and having just cleaned a PCB where two electrolytics had corroded and destroyed some tracks on a PCB, your statement seems a little, um, dogmatic. They were high quality "silver" metal cased tants in military equipment, and had discharged their contents while in storage for quite a few years.

I know the type you're talking about; the hi-rel types have a sulfuric acid electrolyte, the case is solid tantalum, and they use hermetic glass frit seals. Replacing them is extremely expensive 
This type does not fail by exploding, but I have also seen instances where the acid has eaten through the case and corroded traces. This is an aging related failure mode