• Please review our updated Terms and Rules here

Replacing capacitors in >40-year old boards.

D

DavidWien

Member
Joined
Apr 30, 2026
Messages
48
Location
Vienna, Austria
Some old capacitors are known to self-destruct. As far as the Kaypro is concerned, these are often on PSU or video boards. But what about the 50 or so 0.1µF Cs that are attached to almost every IC? Is it advisable or worth it to replace them all before putting such a mainboard back into service, or are they generally still in good shape?

David
 
mark0x01 said:
Disc ceramic types at low voltage are usually very reliable.
Click to expand...
Indeed. Electrolytic capacitors are the primary area of concern due to the use of a liquid electrolyte. Then tantalum capacitors that weren't sized with enough voltage "headroom" (generally 2.5x the in-circuit working voltage). Then there are RIFA capacitors on the power inlet that crack and go (smelly) boom.
 
Last edited:
I am glad to have this confirmation. Indeed, it seems to me, from my experience with other old equipment that I have, that Cs in general, such as those in the PSU, are more likely to die when the equipment is not fired up for a long time, rather than when they are in frequent use.
David
 
Certainly my experience is that every single RIFA X2 capacitor must be changed before powering up any machine. Just not worth the 90% failure rate.

Electrolytics, I would change those obviously bulging, leaking or otherwise showing distress, otherwise give it a go. I haven't even used my variac on really big electrolytics and haven't had a big failure (yet).
Tants are difficult. PC/XT, they seem to have been undersized. MZ80K PSU they failed but in most other machines, I haven't had many problems. Ceramics, only seen mechanical problems and even then they seem to work.
Then you have the myriad of other types, but they are better diagnosed with power normally than replacing adhoc rather than whipping out the soldering iron.
Of course, its ones handling big power that are the risky ones. PSU and VDU caps. Most others just stop something working properly ie in Apple LC's era computers, everyone of the surface mounted capacitors have failed and leaked so need replacing (get some hot tongs)
 
Forget (in general) about the ceramic decoupling capacitors.

RIFA, tantalum, and electrolytic capacitors are the ones to go for.

As capacitors age, they can change their capacitance value. As a result, capacitors used for 'special' purposes (e.g. timing) may end up being wrong.

If this is a reset circuit, who cares. However, if these are used to generate specific pulse widths (as on the TV Typewriter) you can end up with a malfunctioning board (see the thread we are currently working on for some examples!).

Dave
 
daver2 is spot on as usual.
There is documentation out there on 'expected lifespans' and 'MTBF' for various components.
These are serious engineering documents; compiled by professionals, not hobbyists.
Other than mechanical components, the capacitors mentioned all have a low MTBF.

As an aside; in the vintage radio crowd, it is common to 're-cap' a vintage unit as part of the restoration process.
're-cap' meaning replace all the capacitors.

-J
 
Nothing quite like the unexpected bangs from tantalum caps.

I have quite a few S100 cards that I'm going to try conditioning, by appling an initial very low voltage to the power rails and raising it over time, to allow them to reform.

I believe this is how they were created during manufacture and could save s significant cost if successful.
 
I do believe that as a general rule, tantalum capacitors have a chance to explode if the inrush current is too high at first power up after manufacture. I do wonder if they have been sitting for decades unpowered if they go back to some state that mimics that of initial post-production. If so, bringing voltage up slowly as mark says above should help save them from shorting.
 
I was involved with a project at work to remanufacture 1,300 cards as replacements for what was running in one of our plants. All of these cards had tantalum bead capacitors on the + and - 24V rails.

We used brand new capacitors from a reputable manufacturer.

We had a few failures when the cards came off the production line.

We then performed a 400 hour burn-in for each card. We had more failures either when we initially powered up the cards within the burn-in rig or during the burn-in process.

Most of the faults were --- drum roll --- shorted tantalum bead capacitors on either the + or - 24V rails causing the power supply to crowbar.

When we checked the manufacturer's datasheets, we were still within the quoted failure rate for infant mortality...

Even brand new tantalum bead capacitors have a (relatively) high infant mortality failure rate...

The quoted MTBF is only valid during the 'normal operating life' of the component.

The bathtub curve applies here... You get infant mortality until you 'weed out' the poor specimens; then we enter a period of 'normal, stable operating life'; followed by 'wear out' where the failure rate increases dramatically.

All of our vintage equipment falls into the latter category!

It is also true that tantalum capacitors generally fail when the power is initially applied. This is where the capacitor undergoes the maximum amount of stress. However, they can also just fail whilst powered on...

Interestingly, whether a tantalum bead capacitor 'explodes' or not is dependent upon the failure mode. The maximum power dissipation occurs in the tantalum bead capacitor when the internal fault impedance matches that of the internal impedance of the power supply that is feeding it. I am indebted to @Hugo Holden for pointing this out a few years ago, and I validated it for myself.

A faulty tantalum can cause a large current to flow that does not exceed the power rating of the capacitor (or at least causing it to go supernova). In this scenario, if the power supply does not crowbar (or shutdown), the increased current can damage the PCB tracks. This is how I obtained my first S100 backplane! The -18V PCB track had evaporated due to a fault...

Dave
 
johnx993 said:
As an aside; in the vintage radio crowd, it is common to 're-cap' a vintage unit as part of the restoration process.
're-cap' meaning replace all the capacitors.

-J
Click to expand...

One of my other hobbies is vintage radios and TVs. Recapping has been standard practice with those for decades because up until about 1960 paper capacitors were the most common type used in consumer electronics. They absorb moisture, and become leaky. Since vacuum tube electronics usually runs at several hundred volts, even a small amount of leakage causes big problems. It's not even worth plugging in something until the paper capacitors have been replaced. The 50+ year old electrolytic capacitors can be dried out, or leaky by now, although I usually only replace electrolytics on an as needed basis. I've been doing it that way for 25+ years, and I've had very few failures.

After a few decades of dormancy the whole bathtub curve seems to reset. If an old cap is going to fail, it usually happens when it's still on the workbench. After burn in they tend to be quite reliable. Of course, that's after weeding out the obviously bad caps (leaking, dried out, shorted, or know troublesome brand/type). I know it's trendy to do a full recap on anything over 10 years old, but it's just not necessary in most cases. Yes, I might have to occasionally open something back up, but I doubt I would live long enough to do a full recap on every piece of old electronics I own! Also, people assume 100% reliability and long life of the new caps is a guarantee. We just don't know how today's caps will age.

With experience, I've come to know which old caps tend to leak, or fail, and which are reliable. I often open up a computer power supply and find all of the Nichicons leaking, but the other brands are all fine. Nichicon is considered a top brand, but even they have had some bad batches. There have been plenty of supposedly high quality caps that turned out to be absolute disasters. I'm thinking of the surface mount electrolytics from the 80's and early 90's, RIFA caps, the "capacitor plague" of the early 2000's, and the cheap caps that are still being used in things like power adapters.
 
You must log in or register to reply here.
Back
Top