View Full Version : Re-capping power supplies

IBM Portable PC
October 19th, 2013, 03:03 PM
I've over capacitors failing, and the wife is a little tired of the resulting fumes and smoke (usually in the kitchen!). Last night, a tantalum on my shiny new/old 5161 Expansion Unit planer exploded, the resulting pyrotechnic display missing my head by about a foot :dodgeball:

From now on, I'm replacing electrolytic, foil and tantalum capacitors in everything I obtain.:bomb:

This will have the added benefit of reducing ebay purchases as replacing capacitors is not exactly fun, well for me at least. Especially in something like my IBM Portable, the power supply is a nightmare, 3 boards pop riveted to the steel box; I've been putting that job off for about 2 years :fishing:

What do others do? :confused:

Of course I DOI NOT recommend this for those who are not suitably trained/skilled as repairing switch mode power supplies, even when they are not energized, CAN KILL YOU!

October 19th, 2013, 05:02 PM
I don't replace caps until they fail. I've got stuff from the 1940s with the original capacitors and they work just fine.

OTOH, the Great Capacitor Plauge of the 1990s has necessitated quite a bit of replacement work.

October 19th, 2013, 10:11 PM
I have to recap the PSU for my 5162... someday... tomorrow I swear.

October 20th, 2013, 03:10 AM

Usually, i replace tantalum capacitors with electrolytics capacitors and it works very well.
It's a good idea to short the big capacitors near the power plug of the power supply BEFORE trying to repair it : these nasty cylinders can keep high voltage for a long time, and if you forget it, it might be quite painful...

But i still replace failed tantallum with tantallum onto the old "real" IBM motherboards : i like to keep it original as far as i can... But i can't assure you it's a good idea.

The "wife problem", it's a quite common and well known problem, around the world : wifes hate exploding electrical thingies. I cannot explain why... It's quite funny...:rofl:
The kitchen must be avoided at all cost, it's their very own kingdom...
When i test an old motherboard, i try to do it when she's out, or i close the door, open the windows and listen loud music.
Sometimes, the music is a bad idea, because she thinks (incredible, isn't it ?) that i want to hide something from her...
In fact, it's easier for me to replace failed capacitors than dealing with suspicious wife.. that's life...:roll:

October 20th, 2013, 06:32 AM
I work in my renovated basement. That keeps her away most of the time. :)

October 20th, 2013, 08:24 AM
Replacing tantalums with electrolytics is really only a good idea if the value or ESR of the capacitor isn't critical. Old tantalum caps used a liquid electrolyte that would cause the "bang", since the failure mechanism was almost always a short circuit.

In contrast, today's modern tantalum capacitors use a dry powder electrolyte that's not subject to drying out and subsequent catastrophic failure. In theory, if kept within ratings, there's no reason why a modern tantalum capacitor shouldn't last nearly forever.

Tantalums do have two characteristics that distinguish them from standard wet electrolytic capacitors--extremely low ESR and very stable value. These may be critical to an application, so substitution should be done carefully.

In short, other than cost, there's no good reason to substitute a standard aluminum electrolytic cap for an older tantalum device.

October 20th, 2013, 10:51 AM
Thank you for the valuable informations about today's tantalum : i don't think i will replace failed ones with electrolytics anymore.

October 20th, 2013, 10:55 AM
Out of curiosity, about when did manufacturers start using the dry powder tantalum capacitors?

October 20th, 2013, 11:19 AM
I don't know exactly, but probably since SMT has taken over the manufacturing business. WikiPedia says only that:

Because most tantalum capacitors being made as of 2012 are solid (no liquid electrolyte), there is no known wear-out mechanism. Solid-bodied tantalum electrolytic capacitors are less prone to "drying out" than aluminum capacitors, which tend to decrease in capacitance (and increase in ESR) particularly when used in hot environments. When operated within their design limits, tantalum capacitors can maintain their designed capacitance under such conditions for decades.

October 20th, 2013, 01:37 PM
That is until they decide it's their time and just go boom :), scares the piss out of the cat if it's near by when it happens :D

October 26th, 2013, 10:11 AM
In the museum, I often used to start old PCs with a bang and then replace the broken capacitor. (Or sometimes just not bother.)

However a collogue almost never blows anything up. The techniques is to soak the machine with 110 volts for a couple of hours. That is enough to form the capacitors.

Now being in a US lab in the UK means that we have lots of 230/110 transformers. In the US, you need to supply the machine from a power source with a variable output. Try soaking the machine in about 50 or 60 volts for few hours.

Old Thrashbarg
October 26th, 2013, 10:33 AM
In the US, you need to supply the machine from a power source with a variable output. Try soaking the machine in about 50 or 60 volts for few hours.

A variac is fine for linear supplies, but it's a very bad idea on switch-mode power supplies like most PCs use. Not only will it not do much (if anything) to re-form the caps, it can also cause other damage.

November 4th, 2013, 10:34 AM
A variac is fine for linear supplies, but it's a very bad idea on switch-mode power supplies like most PCs use. Not only will it not do much (if anything) to re-form the caps, it can also cause other damage.
We have never blown anything using this technique. I blow a lot more by not doing it. So maybe that is theoretical problem?

November 4th, 2013, 10:43 AM
It depends to a great extent what the load is on the PSU and the design of the PSU itself. Some SMPSUs have automatic brown-out detection and will simply cease functioning if the AC input voltage falls below a certain level. Absent that, if the SMPSU manages to start at a too-low input voltage and with a substantial load, it's possible to subject some components to over-current stresses. In practice, this almost never happens--the PSU most often simply refuses to start until the input voltage has reached some minimum level.

But then, using the variac-controlled input method doesn't help anything either.

November 4th, 2013, 10:49 AM
I generally don't bulk replace caps until a board has show to be stuffed with capacitors that like to fail. Perhaps the key to long-lived capacitors is using them often -- the computers that I use frequently almost never have problems (with the exception of modernish machines with bad capacitors from around 2004-2005). My S-100 and Ohio Scientific stuff has a mix of electrolytics, tantalums in hermetically sealed axial tubes, dip tantalums, ceramic disc, and MLC capacitors. Failures after initial check-out and repair are rare for me.

Some systems seem to be made with capacitors that have a higher failure rate. I had a clone XT board that ended up with all of its tantalums and ceramic capacitors replaced as it would usually smoke at least one every time it was powered on after a period of inactivity. I've also had a disproportionate number of North Star boards with exploding tantalum dip caps.

November 4th, 2013, 11:15 AM
Heat is the big enemy of electrolytics. If your box is well-ventilated and the layout isn't absolutely stupid, then you're usually in pretty fine shape. On the other hand, there are hordes of LCD monitors, set-top boxes and small-profile computer systems that are guaranteed to--and do--cook the juice right out of the capacitors.

If you're in the USA, you probably remember DTV converter boxes from 2006 or thereabouts. The government subsidized the public with special "coupons" (actually looked like credit cards), but manufacturers had to play by the FCCs rules for them--one of those rules was that no external power supplies (i.e. wall warts) were allowed. So manufacturers stuffed a SMPSU into closed boxes with no heat management. Perhaps mandated by design, but nonetheless has resulted in massive capacitor failure in those boxes.

IBM Portable PC
November 4th, 2013, 12:29 PM
Preventative replacement is not much fun, however it avoids potential damage should they fail. Also, remember their purpose to begin with, the cleaner the DC that computer gear receives the better, especially old kit which already has numerous issues we need to work around etc, why introduce more potentially serious issues which may also be difficult to diagnose?

November 4th, 2013, 01:34 PM
I'd agree with you if capacitor technology has progressed in favor of reliability. But, if one looks at the last 20 years or so, I suspect that more effort has been toward getting things into smaller packages, rather than making more reliable capacitors.

November 4th, 2013, 02:49 PM
I get the feeling I will end up doing a cap replacement somewhere down the line on something, so I might as well ask for future reference:

What is the best method of removing the charge from a capacitor? Providing a short circuit doesn't seem so wise, so what is the preferred alternative?

November 4th, 2013, 03:24 PM
Assuming that there is charge left, a simple bleeder resistor will do the job quickly and safely.

Say, 500 ohms for a 12V supply rail. Current will be only 24 milliamps--not enough for a flash-bang, but enough to get the job done in less than a minute.

November 16th, 2013, 09:32 PM
I've got stuff from the 1940s with the original capacitors and they work just fine.

I find that hard to believe. Every piece of electronics that I have come across before the mid to late 1960s has had bad electrolytic capacitors.

November 17th, 2013, 08:12 AM
Just pulled out an old Western Electric power supply with *wet* electrolytics. Comes right up without a trace of ripple in the output. The old telco folks made things to last.

November 17th, 2013, 08:51 AM
The old metal-encased electrolytics last a lot longer than the ones in cardboard tubes. Here's a radio from 1927 that is still working with original capacitors: