[CuriousMarc] was restoring an old Model 19 TeleType. The design for these dates back to the 1930s, and they are built like tanks (well, except for the ones built during the war with parts using cheaper metals like zinc). Along the way, he restored a hefty tube-based power supply that had two very large electrolytic capacitors. These dated from the 1950s, and common wisdom says you should always replace old electrolytics because they don’t age well and could damage the assembly if powered up. [Marc] didn’t agree with common wisdom, and he made a video to defend his assertion which you can see below.
If you look at the construction of electrolytic capacitors, one plate of the capacitor is actually a thin layer that is formed electrically. In some cases, a capacitor with this plate is damaged can be reformed either by deliberate application of a constant current or possibly even just in normal operation.
Of course, [Marc] agrees sometimes a capacitor is done for and has to be replaced. But he is saying that if you test it and it is good enough, it will get better with use. He also shows a capacitor that is starting to fail in a way such that he did replace it.
We get the idea [Marc] was a bit peeved at the comments on his previous video about his failure to replace the capacitors. He’ll get no sympathy from any Hackaday authors though. We have covered how to reform electrolytics before. We’ve also examined the insides of different capacitors, which is a good start to reasoning about how they might fail.
If I’m taking the engine out of my car at 90,000 miles and the clutch isn’t quite worn out yet, I’m still replacing the damn clutch because it’s easier to do it at that point than at some later date. Who cares if I’m throwing new parts in before the old ones completely failed. It saved me from having to rip the transmission out a second time, and it means that it got replaced before it left me stranded.
I do the same thing with electrolytic capacitors in antique equipment. Replace them all once and you’ll have peace of mind for years to come.
No, I won’t. Somewhere along the way, the world lost the recipe for electrolytic capacitors. When I was repairing things in the 1970s and 1980s, old, old, old electrolytics weren’t the problem – wax-impregnated paper capacitors were. It’s only in recent years that I started hearing about people routinely replacing all of the electrolytics when restoring equipment. There is no basis for the expectation that new aluminum electrolytics will outlast old ones.
This article, https://en.wikipedia.org/wiki/Capacitor_plague, describes one facet of the problem, but even on new-model TVs, the most common failure is power supply capacitors, whether electrolytic or polymer. I’ve seen white papers by capacitor manufacturers that show expected life expectancy as a function of ripple current, or of working voltage. The point here is that equipment is being designed for the capacitors to fail. Whether this is a failure of capacitor design, or just a matter of manufacturers over-rating their products, capacitor failures are a modern problem.
The recipe wasn’t lost…the crappy endurance is a price you pay for the small size and low price of modern caps. Try to find an oldschool cap, that’s about 2mF at, say, 20-40V. Compare the size (and if possible, cost) with a modern one.The old one is huuuge and heavy.
Also, the cap being physically bigger means it will have more parasitic inductance, which is unwanted in switchmode PSUs.
If you want endurance – replace electrolytics with MLCC, but prepare your wallet to take quite a big hit.
Switchmode PSUs also put a lot more load on the capacitors than old linear power supplies.
Imo this is the main reason for electrolytics to fail these times. Although I had one failed in a 1972 sewing machine – one of the first micro processor controlled – and this failed short, not high impedance.
If capacitors are failing due to high pulsed current at the switching rate, then either a) the power supplies aren’t being designed properly, or b) the capacitors are being over-rated. Since electrolytic capacitors are pretty expensive, it’s probably a combination of the two.
Try charging and discharging old caps 100 000 times (or more) per second, and see how long they will last.
Hint: switching power supplies.
You need the right capacitors for that. Otherwise they tend to overheat and boil off.
This is the reason you use electrolytics for bulk storage (i.e., to handle the 50 or 60 Hz ripple) and ceramics or polymer dielectrics in parallel with them to handle the pulsed load. This is not that expensive, and not that difficult. Saying, effectively, “I guess we’re just screwed because of switch mode power supplies” is just exonerating people for bad design.
Capacitor plague really gave electrolytics a bad name.
I understand that you want the same look in antique equipment. It is not uncommon to scoop the contents out of a can capacitor, especially multi-section electrolytics, and stuff modern much smaller replacements inside.
Some Chinese manufacturers do that for you for free on NEW capacitors!
I’ve done that – it’s time consuming, but worth it for a real show piece. I’ll try to re-form first, and if it’s clear it’s not in the cards, I’ll hollow out and re-stuff. On older Tektronix scopes, it seems the higher current supplies like the -150V are more susceptible to capacitor failure than say the much lower current 450V supply. YMMV.
That makes sense, higher ripple currents mean more heating. You can reform from a capacitor sitting for a long time unpowered, but a dried capacitor is gone for good.
I refurbished a Heathkit capacitor bridge, the voltage test function can be used to reform aluminum electrolytics because it is current limited.
A Heathkit IT-28 is exactly what I use for that task ;)
The main reason to replace the large electrolytics, is that they are usually power supply filter caps. If one fails shorted and kills the power transformer….then you have a real problem, and you will really wish you had replaced the electrolytic when you had the chance.
Agreed. That is the one part that is most difficult to replace, and very expensive to have rewound.
…and Marc addresses this in the video. Electrolytic capacitors do not usually fail shorted, and there are fuses to protect the transformer in this device.
See, e.g.: https://www.illinoiscapacitor.com/pdf/Papers/failure_modes_aluminum_electrolytic.pdf
Now tantalums, they’re another story… I could tell you a story about a certain piece of equipment that had tantalums on the outputs of the DC supply, had current limiting (1A) built into the supply, but still burned out the transformer because it was only rated for 250mA, but that’s off-topic :)
Is that as big a deal on radios with tube rectifiers? I would think that a tube rectifier would have enough internal resistance to save the transformer. You could also put something like a 10ma fuse in line with the caps. That would limit the high voltage to about 3 watts assuming a 300V plate supply.
For vintage devices not powered up for a long time this is the rule:
With equipment containing tube rectifiers you have to use external current limited voltage to reformed the capacitors before putting power on it.
The tube rectifiers do not work until a large enough voltage is reach, so starting up an old radio or device with a variable AC power does not help.
So reform the capacitors before applying power for the first time!!
I have made my self a solid state 5u4 before with 2 diodes and the base of a bad tube. Then I was able to bring it up very slowly on a Variac with a dim bulb. It worked great.
well, this COULD be true, but lets face it, you are going to replace a capacitor of very high quality, but old with a much lower quality capacitor which will fail way sooner. of course you don’t use old equipment on a regular basis, so maybe you get the same lifetime out of the new ones.
So, you’re going to leave in archaic capacitors with god knows how many hours of use? That makes no sense. If I’m going to rebuild an engine, I’m not going to put new bearings in. I’m not going to say “well, the rod bearings are worn, but the main bearings look good so I’ll let them ride”. That’s being a bad tech.
Going to put new bearings in*
That’s a poor analogy. You’re comparing a piece of electronic hardware that takes literal minutes to replace with complex, cumbersome procedure that requires hours of labor.
Furthermore, if it has has worked for 70 years and tests well, why are you rushing to replace it?
Peace of mind that now it’ll last a further 70 year?
Also less time consuming to have fewer but bigger services.
no it wont. new capacitors fail way earlier than that. 20 years tops (if treated well) just accept the fact that new ones just aren’t that good.
Absolutely, there’s nothing wrong with reforming a cap if done right. And some caps used in old vacuum tube equipment are still manufactured the same way today, making them far from archaic. Vacuum tube radios is one of my long term hobbies and what gets replaced verses what gets reformed depends on construction. I’ll replace a old paper foil cap without a second thought because of how they fail but a electrolytic can is an entirely different cap. If it’s not leaking and still has it’s electrolyte then you have everything inside needed to reform (remake) the capacitor. And when you reform these caps you are duplicating the process used to make them years ago and is still used to make the same style of capacitor today.
I will certainly not if they are vintage and still look good. Absolutely no point in replacing well made vintage bearings that are still within wear limits with modern chinesium crap.
Not too surprising. The newer mixes they use are better short term but tend to chemically react more. I remember how this ‘replace all the caps’ thing started. Some manufacture ripped off another one and then got the formula wrong. A bunch of OEMs bought those caps. So blow outs and leaks everywhere. Happened in the late 90s. Took years to get rid of them all.
And now, everything is high frequency, smaller capacitors, so smaller form factor to dissipate heat from I^2R losses. I think that is causing some capacitors to fail, and I see a lot of people who think the bad capacitor formula is still going around. When it is really just greater demands put on capacitors in modern electronics.
If heat kills it, maybe aftermarket heatsinks are the answer?
early ’90’s
Dont forget one issue. PCBs they are quite common in old capacitors. If they blow up you got a hazmat issue on your hands.
I as a rule just replace them to be sure.
Just to reiterate: electrolytic capacitors don’t use oil, much less PCB-tainted oil.
As far as I know the PCBs themselves are not the main issue, it’s the Dioxin formed as they degrade that is the actual hazard.
But I’m a belt and suspenders kind of guy so I would replace the old caps anyway.
PCB’s (poly chlorinated bi-phenol’s) WERE used in some capacitors, 100 Kg mass and 2500+ volts…
B^)
Okay, then, I’ll try one more time: ELECTROLYTIC capacitors do not use, and never have used, PCB-containing oil. They don’t use oil at all. Electrolytic capacitors use an ELECTROLYTE, which is a conductive liquid. Electrolytic capacitors form a very thin oxide layer on an aluminum foil, as a result of current flowing through the electrolyte during the forming process. PCB-containing oils s would be completely unsuitable for this, since they are very good insulators.
Also, PCBs (Polychlorinated biphenyls, which are used in transformer cooling oils, and in some high voltage capacitors, not printed circuit boards) are not just toxic, and not due to any compounds they break down into, but are carcinogens in their own right.
You are correct, as an electronics tech in this trade since 1965 when I started an apprenticeship and knew bugger all I have not ever seen so many people offering an opinion on a subject they know very little about. When I first entered the electronics section at the firm I had completed my apprenticeship with there was a sign above the door. “It’s what we learn when we think we know it all that counts”. There are many here that think they know it all and then some.
i was going to comment about how you should replace them anyway BUUUT then i saw it is a malory … so it’s probably going to be good as long as you re-form it very very slowly, the OCV and current-limit need to be adjusted over the course of maybe an hour (or more?)
malory is a very good capacitor formula, and when re-formed every few decades can last WAY longer then that modern crap, and im not referring to that one bad (discontinued) formulation.
ever heard of re-forming modern capacitors? who does it? not many eh? only the good formulations get re-formed over and over and still can be re-formed. but they dont make em (for the domestic market) anymore. time to buy up stock in space-capacitors. they last lol
although “old” capacitors bring to mind shredded paper, shredded metal, eye-injurys, and possibly poisining/ect … some of those old capacitors can be re-formed (like from scratch) and be able to serve many more decades and possibly an additional re-formation as has possibly already been done. but a leak of liquid residue or a crappy formulation means it’s toast, now or soon.
PS: just dont go plugging it in as a way of testing it… if it has PCBs you and your neighbours will not live a long and healthy life if that cap pops. also, your house, belongings, and the soil on your property will be no longer be safe to visit let alone live. and anyone your close to will have elevated levels and pass those down through all future generations with cancer to spare for your co-workers and thier children. this substance is not destroyed by anything in nature execpt maybe a volcano’s high-temperatures. but dont worry, the melting polar ice caps are about to shatter bunkers where it is stored in vast quantities; more then has been released over the substance’s ENTIRE HISTORY. did i mention it is not eliminated by distillation? only 5000 or 6000 celcius can break it down.
PS: i think it is illegal to operate, transport, dis-assemble, or even posses equipment containing PCBs in the united states without a permit and goverment oversight, which may take months or years to get approuved. some people use high-concentration PBC-laced transformer oil for street-vendor deep frying. this is an example of a minor crime (theft of oil from live transformer causing overheat and fire) into a serious crime against humanity (giving cancer to the entire community)
You find PCPs in oil filled equipment like oil filled transformers, not in electrolytic caps. Electrolytic caps are filled with ethylene glycol, a boric acid and water solution or with an organic compound, usually a “secret sauce” that the manufacturer doesn’t want to share. Some of the really old ones are open topped that can be refilled with distilled water.
That’s not how poisonings or genetics works. Your kids’ll be fine. If the father is contaminated he’ll only contribute as much PCB as he can provide during the event, The Big Inning, as it says in the Bible. And most of the father’s contribution will be absorbed by the mother. If it’s the mother she’ll be sharing her contamination but el nino will get a much smaller portion. And what are the stats on PCBs crossing the placenta again?
I am quite sure, that working equipment containing old caps (with PCB oil) does not have to be replaced, can be used until it has to be scraped. You probably need the permit if you want to do scrap disposal or similar activities on PCB contaminated devices. Also PCB is not Plutonium :-) What and where should this bunkers be? And no, it does not need 5000°C A normal chemical waste incineration facility will do it. Yes, when the stuff burns, nasty poisons like dioxins are produced, but even for this stuff incineration at something like 1300°C is enough. But I think you have to specially treat the resulting gasses, as dioxins can re-form when the gases are cooled to slowly, I think some fast quenching is necessary.
Transformer oil, PCB containing or not, is normally based on mineral oil. This is in no case acceptable for human consumption like deep frying. Crazy People!
These capacitors would not pass Mr. Carlson’s capacitor leakage tester. The capacitors are only reforming (or seeming to) because of low run time hours (intermittent use) on the power supply. They won’t last long in actual continual service. These Mallory capacitors are not wax type, nor do they have stamped rubber seals of traditional pcb mount caps. These caps have been mounted inverted so the seals have not been under constant attack from the internal acids. So whether or not the caps pass your test, they will be unreliable in constant use because they have gone past their shelf life.
Electrolytics are inherently more leaky than other caps. The current for reforming appears as leakage current.
Old Hammond amps have oil filed caps, never have had to mess with one.
That ‘lytic formula was wrong on purpose to spoil espionage. So there were 2 lawsuits filed, one against Nippon Chemicon or some one like that for letting it out they didn’t use it. The other of course wasn’t against any one company because that is how industry in Communist China works. They copied it everywhere and probably still do in real crap.
much of the bed rep that electrolytics get is from the period in the 00s when there was that bad electrolyte formula was going around. cpacitors from around that time were exceptionally suicidal. ive replaced caps from those devices with caps salvaged from 70s/80s stereo equipment and they work as well as a new one if not better. survivorship bias was definitely in their favor.
needless to say if a part has stood the test of time then dont replace it because the replacement would be inferior. you dont know if that new fangled cap is going to go tits up on you.
I have several old electrolytics in regular use from the 1950s. One of them clocked 5.000 hours (!) in my personal use already since 2007. Completely reliable.
So if you have measured capacitance and leakage, then there’s no reason to replace them.
Of course i have had several bad electrolytics. You notice that soon enough – tremendously high leakage or lost capacitance. Sometimes the overpressure valve has already acted. Obviously those need to go.
In my opinion many people are way too quick to just replace them.
Thanks for this practical and almost obvious comment! It clearly states what engineers and technicians should go for — clear and measurable parameters as symptoms of failure.
Namely, this “recapping” frenzy is going on and on, and when we think about it in the modern world, no wonder that no single manufacturer, even the reputable ones, write about it critically.
Just one more thing. Measuring the capacitance with a good LC meter or even reliable universal instrument is a simple task. But what would you recommend for a practical and prompt “measurement of the leakage”, that you also mentioned. Of course, if the leakage is great, the capacitor will not hold its voltage when used as DC supply/stabilizer. But what I have encountered up to now were only drastic examples of this, i.e. the total breakage, producing full conductivity and resulting in the short-circuit. Have you measured some in-between states? Also, for the non-polar electrolytic capacitors, used in the signal path, how would you measure the leakage?
Well, the usual process to check for leakage is to take the capacitor out of circuit and apply a current limited voltage that is equal to the maximum rated voltage of the capacitor, and measure the leakage current. Usually, the voltage is brought up to the rated voltage.
If you are fortunate enough to find an old Heathkit capacitor leakage tester, even better. They can test capacitors up to several hundred volts. I just rebuilt a Heathkit IT-11. Ironically, it had a bad capacitor. The circuit required a very high voltage capacitor and so it used two 4.7uF 400V electrolytic capacitors in series, I replaced them with one 2.2uF 900V film capacitor. The previous owner had already upgraded the other capacitors in the circuit.
Back when capacitors were very expensive, people used to reform the dielectric layer with these instruments.
Not my page:
https://www.radiolaguy.com/Showcase/TestEquipment/HeathKit_IT-11.htm
Its a Mallory capacitor
I’m not in favour of the wholesale or shotgun approach to replacing electrolytic capacitors unnecessarily. IMHO this is an amateur approach akin to resolving a fault by replacing everything. My approach is to measure the amplifier to determine if its performance is as specified along with a thorough visual inspection and a critical listening test. I’m of the opinion based on decades of experience working professionally within the industry that if it ain’t broke don’t fix it. If testing reveals a problem then that warrants further investigation. Track down the problem and then fix it. If a power supply electrolytic is failing, losing capacitance and the ESR is rising these problems can be easily identified by an increase in ripple and a drooping of the d.c. supply voltage(s) under load. Failing electrolytics used as coupling or bypass capacitors will be identified as a loss of gain and a premature roll off of the LF end of the audio range (less bass). To verify if the capacitor is suspected be faulty or is failing it should be removed from the circuit and tests for capacitance, ESR, dissipation factor, quality factor, phase angle and leakage should be performed and the results of these parameters compared with the manufacturers’ datasheets (if available). If the capacitor tests fine and there are no other signs of impending failure like bulging, venting or leaking IMO there’s no valid reason to replace it.
I can appreciate why some restorers and repairers just replace all the electrolytics even if they later prove not to be faulty because some gear is difficult to work on and replacing all caps at once means only removing and replacing the board(s) once or they can’t be bothered doing any testing to justify not doing a wholesale replacement. On the flipside I often read on audio and DIY forums as well as see on Youtube where the person replaces a perfectly good capacitor that’s been working faultlessly for 40 – 50 years with an expensive “audio grade” capacitor with a rated working lifespan of only 1000 hours. Please tell me the rationale behind this thinking.