• Please review our updated Terms and Rules here

PSU recapping - what's the general consensus on when to do it?

No. You are nitpicking.
I can look at any of my laserdisc players and belts aside it's a crapshoot weather or not they will actually work. Be it the 7820 or the V2600 both have capacitors that have long since past the point the capacitors were expected to still be in use. There are an infinite number of variables as to why one day they will bless me with the ability to watch Die Hard and the next the focus servo doesn't work because it lost its signal decoupling.
The fact that a Lloyds combination Hi-Fi from 1975 will work as-is and not hum too much is nothing short of a miracle, given how that was still the era when Japanese capacitors were absolute garbage. The fact I plugged an all-american 5 into the wall and it promptly smoked a resistor and fried a DC rectifier tube is an understanding those capacitors were long past the point of being capacitors. Yes, indeed there is still a lot of things out there where the capacitors are 40, 50 or even 60 years old and they are still within 5% of spec and will probably last decades more, but I'm replacing capacitors manufactured by Nichicon in 2009 which are already drifted too far out of tolerance.
If I want something to be reliable there is no alternative. If I'm butchering traces, that's my damn fault, but this was going to need to be done anyways.
 
If it was a choice between trying to re-cap my Tek 465 o-scope or letting things ride so long as the thing still worked, I'll choose the latter. I'm sure that working on that thing will affect its operation, which is fine right now.

But as to "shotgunning" replacement, no--there are capacitors that are in good shape that are not in any power-handling role, so they're not going to go boom. I'll choose my replacements when the situation requires it.

Age may have something to do with it, too. If I re-cap something not needing it, there's always the question of the thing outlasting me--i.e. money and time wasted. :)
 
Last edited:
I recap only when, during a visual inspection, I find any traces of electrolyte leaking from the caps - just a slight amount of discoloration on the leads and I remove them. Often, if I'm already there, I'll proceed to recap the entire board (or at least the larger caps). I once proactively recapped an entire functioning Wells-Gardner CRT chassis from a cocktail table arcade game that was around 30 years old, and found the original caps to have better ESR readings than the new ones! Long story short, there was ZERO difference in the CRT's functionality before and after the cap job. The capacitors made in the 1980s were of better quality than later ones based on my findings - but age does take its toll too. Recently, I resurrected a Motorola clock radio with all original components (including a multi-sectional capacitor) and all it needed was an amplifier tube - sounds absolutely perfect, no hum or anything, and it has one of the highest failure-prone capacitors known to mankind! (those old multi-sectional caps are horrendous!). In short, In only recap if I see leakage or if there's a functional deficiency.
 
Another thing for me is if I even SUSPECT a bad capacitor and I need to pull it out of circuit to test it, it's getting replaced. Reason being, if it didn't leak BEFORE being subject to the heat of a soldering iron after 35 years, it'll certainly leak now! If I'm going through the trouble to desolder it, it ain't going back in. This is why I'm working on an ongoing project to stock every known type of capacitor used in the equipment I service regularly.
 
Some of larger machines at the RICM have very large electrolytic capacitors, and lots of them. Our PDP-12 has capacitors that looks like two stacked Foster's beer cans, 160,000 uF @ 20V, and 57,000 uF @ 50V. The cost of replacing them all would have been about $1,000. The distance between the screw terminals on the replacement capacitor is smaller than the originals, so we would have needed to drill new holes in the bus bars. We checked them for electrolyte leakage, slowly reformed them, checked them for electrical leakage, and everything has worked OK since. We would have checked them for capacitance and ESR, but we don't have a LCR meter that will deal with the very large capacitance. I suppose we could have charged them, and measured the discharge time through a precision resistor.
 
Some of larger machines at the RICM have very large electrolytic capacitors, and lots of them. Our PDP-12 has capacitors that looks like two stacked Foster's beer cans, 160,000 uF @ 20V, and 57,000 uF @ 50V. The cost of replacing them all would have been about $1,000. The distance between the screw terminals on the replacement capacitor is smaller than the originals, so we would have needed to drill new holes in the bus bars. We checked them for electrolyte leakage, slowly reformed them, checked them for electrical leakage, and everything has worked OK since. We would have checked them for capacitance and ESR, but we don't have a LCR meter that will deal with the very large capacitance. I suppose we could have charged them, and measured the discharge time through a precision resistor.
Those are possibly PCB capacitors, not likely those would go bad any time soon!
 
Here's another refernce point--being in the US with single-phase power, I have quite a few tools with large motors using start/run capacitors. On those motors that date from pre-1980 or so, things are just fine. On later motors, I find that I'm replacing caps about every 7-10 years.
 
Some of larger machines at the RICM have very large electrolytic capacitors, and lots of them. Our PDP-12 has capacitors that looks like two stacked Foster's beer cans, 160,000 uF @ 20V, and 57,000 uF @ 50V. The cost of replacing them all would have been about $1,000. The distance between the screw terminals on the replacement capacitor is smaller than the originals, so we would have needed to drill new holes in the bus bars. We checked them for electrolyte leakage, slowly reformed them, checked them for electrical leakage, and everything has worked OK since. We would have checked them for capacitance and ESR, but we don't have a LCR meter that will deal with the very large capacitance. I suppose we could have charged them, and measured the discharge time through a precision resistor.
Yeah, we reform the big high-quality stuff when encountered. No reforming for Sangamo caps though :p You can tell they've dried up just by picking it up...
 
Here's another refernce point--being in the US with single-phase power, I have quite a few tools with large motors using start/run capacitors. On those motors that date from pre-1980 or so, things are just fine. On later motors, I find that I'm replacing caps about every 7-10 years.
"Let me expound on how things which are not electrolytics -- the topic of discussion -- not being bad means one shouldn't worry about electrolytics."

Replaced four in a HP 5326B counter from the late 60s/early 70s today. Not what I'd call consumer junk, but the smell of leakage was apparent on removing the top cover. Caps *looked* good though so I suppose I should've just run them, right?
 
I am sure that once design went from hand calculations and fudge factor to doing PS design on computers with little fudge factor (plus bean counters taking parts of till it barely lasted through the warranty) designs were just not that good when they aged.
 
I'm in the 'if it isn't broken, don't fix it category'. If I had a clearly 'developing' situation where something started to take longer and longer to turn on then I'd investigate that and caps would probably be in the first line of checks, but taken to the absolute extreme you could end up on a rollling programme of replacing all the caps in everything you own all of the time, like a kind of electronic Forth Bridge Repaint.

Recapping also requires a certain level of knowledge, for example if you were to recap a typical SMPSU with standard 85 degree electrolytics of no particular pedigree then it might work, but probably not for long. The caps in the high frequency sections of such circuits typically need to be low-ESR high temperature rated types. Plus, there can be situations where a capacitor with a very particular specification was used in one place in the circuit and replacing it with something approximately equal can in the worst case scenario actually stop the circuit from working.

For this reason, if you must recap:-
Take LOTs of digital photos from all angles before you start.
Remove only one capacitor at any time and as you remove it note carefully whether the orientation marked on the PCB actually matches the way the part was positioned. It is not completely unheard of for mistakes to be made on the screen printing on the top side of the PCB - the manufacturers become aware of that and they start fitting the parts the right way around, and not the way around indicated on the PCB. Before you set the removed capacitor aside, mark the capacitor with the circuit position (C123, or whatever) that it came from.

Fit the replacement capacitor and then repeat again for the next capacitor.

Removing only one cap at a time minimises the risk of your getting muddled and not being able to remember which of three capacitors came from which three positions. If that does happen, well, it's a good thing you took all those photos beforehand, isn't it?

Marking the capacitors with the actual position they came from is insurance against the scenario where changing one of the caps and replacing it with something which seems like a reasonable replacement may actually stop the circuit from working. As long as you know exactly which individual caps came from which positions, you always have the option of putting the original ones back in where they came from to see if that gets the unit going again.
 
I've replaced a lot of bulged electrolyic cap's and left others in place to repair pwr. supplies and on radio, computer and other boards. Lots of times I feel it's more damaging to crappy 50+ year old phenolic PCB's to do any more work on them than necessary. It seems to me, that the larger the cap's the less likely they've failed as a rule of thumb. I don't know why that is, but it seems to be often true. Tanalums at least have the courtesy to let the magic smoke out so you know which ones to replace.

Chuck, I'm still (slowly) working on replacing leaking cap's in my Tek scope. So may cap's and so many damaged traces. It did still power up and almost fully function, but not enough to ignore.
 
bulging is never something I check for. It only seems to happen when heat is a factor (like flat screen TV psu boards for instance) I have had plenty of electrolytics that have leaked or didnt bulge and were bad. Bulging doesnt happen often, therefore I dont go looking for it.

Take a Macintosh SE power supply for instance.. Pretty much all the through hole caps are leaky, none are bulging. Why? Well most have been sitting powered off for decades and the seals went on them and they leaked electrolyte all over the psu board. Now if it was a mac se that has been in use for all this time for some reason, sure its possible there could be a bulging cap.. but honestly I think the caps would have just ended up leaking somewhat sooner. Again I dont go looking for bulging caps on vintage hardware.. Maybe on tech from 2005 and newer... but thats it.
 
Older electrolytic capacitors were in pretty strong metal cans. They rarely leaked - they bulged. They also went off like a bomb if you powered them up! They did, however, have a vent plug. We had one go BANG at the electronics club at secondary school when we were repairing an old piece of (very dangerous...) electronics. It may have been an old ex. war department oscilloscope. Needless to say - the repair of old equipment very rapidly became curtailed...

Some of the capacitors on my 11/45 power supplies are bulging. They are saying "go ahead punk if you think you are man enough to power me up"! I am not, those two capacitors are going to be replaced...

Dave
 
Older electrolytic capacitors were in pretty strong metal cans. They rarely leaked - they bulged. They also went off like a bomb if you powered them up! They did, however, have a vent plug.
Had a piece of equipment come in for repair where a large soda can sized electrolytic had blasted its plug out, sprayed its contents into the top of the chassis (clamp mounted by the bottom of the cap), and corroded the anodizing off the aluminum! Left a mound of nasty stuff that hardened and had to be knocked off. Another case of "unable to test" meaning "smoked real hard."

I've never had one fail in the shop, but I hear if a hermetically sealed axial tantalum or paper-in-oil capacitor goes, it's about like a gun shot.
 
This is inside if a Wavetek Sweep Oscillator from the mid-1980s I am working on. The tantalum cap in the middle is open and not causing problems. The tantalum on the left is shorted and killed the -18VDC supply. Both are in backwards courtesy of Wavetek. I will replace all of the tantalum caps in this device, but the electrolytics are OK.

1675214207695.png
 
Their is no correct answer. Everyone will pick different point on the replace/test/wait to fail curve. I've had several failures of the PCB oil filled capacitors so I am replacing them now. Caps that age (60's and early 70's) aren't protected so they rupture and get PCB oil all over. I know of one on a ferroresonant supply may have taken out the transformer when it shorted. New ones may not last as long but likely next failure will be someone else's problem.
http://www.pdp8online.com/shows/vcfe07/pics/dscf0037.shtml?small

I mostly work on old equipment that doesn't have high failure rate of the capacitors so check the big ones and assume the small ones won't cause cascading problems if they fail. Check is reform and check ESR & capacitance if easy to remove or ESR and other tests that I can in circuit if soldered. On one machine caps were powered by 30 AWG wirewrap in a backplane so when it failed it took out the backplane wire and transistor in supply driving it. Luckily it didn't do too much damage to other wires so only wasted some time to find and fix. Replaced the rest of them. Reforming may have been fine but decided replacing better due to damage would be repeated if I was wrong. This ends up being a machine specific worry about these caps.

Had problems with dipped Tantalum caps in tek 4014 but haven't dealt enough with equipment that uses them to decide how to deal with them. First power up of new equipment using them back in the day could be exciting but then they were good after that.

Have more trouble with caps in old electronics we have around than in the vintage computer equipment I normally work on.
 
Actually in my experience small capacitors, especially those placed close to heatsinks or other hot-running components such as high frequency transformers, are at high risk of failure, the reason being that they don't have much electrolyte to lose in the first place and are therefore more prone to being baked dry by nearby hot objects than larger caps which contain more electrolyte.

I hardly ever have to replace the 'big' mains rated electrolytic capacitor on the output of the mains bridge rectifier (thinking specifically of SMPSUs in this case), but if there is a small electrolytic in the start-up chain which supplies a momentary burst of power to the switching controller at switch-on that's the first thing I look at in a 'dead' SMPSU, provided there are no blown fuses and nothing else obviously blown up.
 
I hardly ever have to replace the 'big' mains rated electrolytic capacitor on the output of the mains bridge rectifier (thinking specifically of SMPSUs in this case), but if there is a small electrolytic in the start-up chain which supplies a momentary burst of power to the switching controller at switch-on that's the first thing I look at in a 'dead' SMPSU, provided there are no blown fuses and nothing else obviously blown up.
Yeah a dried out or shorted bootstrap cap is often the source of no-start on switchmode supplies!

I also don't often have to replace the line-side filter capacitors. Pretty sure the handful of times I've had to replace them have been from general severe overheating or mechanical damage.
 
Back
Top