Importance of replacing the electrolytic capacitors in any vintage gears

[MAB]

If someone replaces a capacitor and the new part fails early due to a defect, how would they know?

Given that it is most often very difficult to tell if a cap is bad while in service without directly measuring it, and in most consumer audio applications the penalty for a bad cap is so incredibly small (most often a buzz of supply harmonics, often close to -90 or -100dB), are we all really so sure that all of the caps that we replaced are working perfectly? After all, most of us with vintage audio gear have at least a few caps bad in them, and yet we all can't hear that unless it is one of the critical few.

Same for professional service and repair, who's goal is to get the part past the 30-90 day service warranty. They don't know, their customers don't know, unless the repaired part was absolutely critical. Unless the end user has an audio analyzer or goes through the tedious task of remeasuring all of the replaced items, they just don't know.

For aerospace and military, they do a variety of early life failure verification. Some audio manufacturers do this, like Bryston who release a post Burn-In Final Check report with each piece, but their customers pay a premium.

The consequences of capacitors going bad are typically very mild, and detecting them in the field is done so seldom (where is the post-service-remeasure crowd?), it's hard to say that doing a recap extends the life of a piece, when the consequences of many of the parts failing is typically negligible for both the old part and the part it was replaced with.

 

[levimax]

If someone replaces a capacitor and the new part fails early due to a defect, how would they know?

Given that it is most often very difficult to tell if a cap is bad while in service without directly measuring it, and in most consumer audio applications the penalty for a bad cap is so incredibly small (most often a buzz of supply harmonics, often close to -90 or -100dB), are we all really so sure that all of the caps that we replaced are working perfectly? After all, most of us with vintage audio gear have at least a few caps bad in them, and yet we all can't hear that unless it is one of the critical few.

Same for professional service and repair, who's goal is to get the part past the 30-90 day service warranty. They don't know, their customers don't know, unless the repaired part was absolutely critical. Unless the end user has an audio analyzer or goes through the tedious task of remeasuring all of the replaced items, they just don't know.

For aerospace and military, they do a variety of early life failure verification. Some audio manufacturers do this, like Bryston who release a post Burn-In Final Check report with each piece, but their customers pay a premium.

The consequences of capacitors going bad are typically very mild, and detecting them in the field is done so seldom (where is the post-service-remeasure crowd?), it's hard to say that doing a recap extends the life of a piece, when the consequences of many of the parts failing is typically negligible for both the old part and the part it was replaced with.

I agree with this.

Another thing I see is that the location of the capacitor on the board can often cause it to fail early, usually one or two located in areas of the PCB that are much warmer than other areas. If these measure way out of spec (ESR) then replacing them makes sense but if other caps in other locations on the board measure in spec I don't think replacing them makes sense in most cases.

 

[JeffS7444]

With recapping there's always the risk of board damage, especially with equipment that uses lead - free solder.

AFAIK, lead-free solder will mostly be found in equipment dating from 2005 or newer.

 

[DanielT]

If someone replaces a capacitor and the new part fails early due to a defect, how would they know?

Given that it is most often very difficult to tell if a cap is bad while in service without directly measuring it, and in most consumer audio applications the penalty for a bad cap is so incredibly small (most often a buzz of supply harmonics, often close to -90 or -100dB), are we all really so sure that all of the caps that we replaced are working perfectly? After all, most of us with vintage audio gear have at least a few caps bad in them, and yet we all can't hear that unless it is one of the critical few.

Same for professional service and repair, who's goal is to get the part past the 30-90 day service warranty. They don't know, their customers don't know, unless the repaired part was absolutely critical. Unless the end user has an audio analyzer or goes through the tedious task of remeasuring all of the replaced items, they just don't know.

For aerospace and military, they do a variety of early life failure verification. Some audio manufacturers do this, like Bryston who release a post Burn-In Final Check report with each piece, but their customers pay a premium.

The consequences of capacitors going bad are typically very mild, and detecting them in the field is done so seldom (where is the post-service-remeasure crowd?), it's hard to say that doing a recap extends the life of a piece, when the consequences of many of the parts failing is typically negligible for both the old part and the part it was replaced with.

A slightly bigger annoyance is mechanically humming transformers. Humming sound from the amplifier/trafo , not humming that is heard in speakers/headphones that is. That hum disappears, or is drowned out by the sound of the music but at low volume or no music at all but the amp on it can be irritating. I say can because it is probably highly personal how you experience that humming sound.

I tested, due to Hifi history, a vintage NAD3020 amplifier. It had too much mechanical humming from the transformer for my taste. I sold it to my friend ,the electronics engineer, who... I don't remember what he did with it. He maybe plugged in a DC blocker, he maybe changed... if that helps? , some capacitors. I don't know, but he did something. I'll ask him about it. It was a number of years ago.

My Luxor 7082A amp , same problem, although not as loud. Cut to fit rubber sheet placed between transformer and chassis plus properly tightened screws. It worked well. The humming transformer sound disappeared. However, it did not with the same operation on my NAD3020.

 

[nanook]

If someone replaces a capacitor and the new part fails early due to a defect, how would they know?

Many caps have a date code or it's a series that is no longer in production. .

and in most consumer audio applications the penalty for a bad cap is so incredibly small

I agree, but this holds true for maybe 1 out of 3 caps.

Same for professional service and repair, who's goal is to get the part past the 30-90 day service warranty.

In my case my internal customers (post-docs and doctoral students) would just bring in the unit again, so I rather make sure it works flawlessly until I'm retired ;-)

Last not least, disassembling, assembling a unit and doing the final testing often consumes more time than replacing a couple of caps with ones that will live "forever".
If you compare the 1000 to 2000h @ 85°C that were available decades ago to the 7000h @ 105°C that are available nowadays, you know that this is the last time it needs exchanging the liquid electrolytics.

 

[MAB]

Many caps have a date code or it's a series that is no longer in production. .

I'm asking a slightly different question. How would the user know? Given most capacitors in consumer audio products have a purpose that is inaudible, and need sensitive measurements to detect. It's more likely that the person listening to the 'repaired' item has no idea that some of the new capacitors have a defect that will fail in the first few weeks or months. I think most people are unaware of the actual early life fail rate of electrolytic. It used to be several percent, is better these days, but not zero.

I agree, but this holds true for maybe 1 out of 3 caps.

Edward Muntz would disagree, more like 90% serve purposes not directly observable to most.

Muntzing - Wikipedia

en.wikipedia.org

The truth is more likely somewhere in the middle. Many audio pieces have lots of performative capacitors, unlike other more value-oriented consumer electronics. Some of my vintage (and new) gear has so much unnecessary decap on each rail, I recall fondly the page-long advertisements describing the incredible measures manufacturers took on power supplies. It's certainly nice to fix these and get that last bit of original performance back. I just think we have to accept that a measurable percentage of the replacement parts will show an early life fail. Back in the 1970s, this was a pretty large number. Now I think it is <<1%, but still measurable unless the number of replacements is small. This assumes flawless execution, and some of the vintage units have poor manufacturing, but also some are hard to remanufacture to the same standard. Either the boards suck, or the original solder was excellent and highly uniform and I suck.

In my case my internal customers (post-docs and doctoral students) would just bring in the unit again, so I rather make sure it works flawlessly until I'm retired ;-)

That's my question, most of our friends aren't getting out analyzers and LCR meters and exhaustively investigating every replaced item. So won't notice unless one of the few key components fails. Not saying this is absolute, just that we trade a certain end of life risk that may be 10 or 20 or 1 years in the future, for a small but real early life risk. 'Risk' is quite low in most consumer electronics, according to Muntz.

Last not least, disassembling, assembling a unit and doing the final testing often consumes more time than replacing a couple of caps with ones that will live "forever".
If you compare the 1000 to 2000h @ 85°C that were available decades ago to the 7000h @ 105°C that are available nowadays, you know that this is the last time it needs exchanging the liquid electrolytics.

Well said. It takes a long time to clean up and verify the work, no matter which approach you take.
I tend to take the don't-fix-unless-broken approach, identify key issues by test, use my limited experience, and telltales like temperature. I land in a different place on the bathtub curve but am pretty happy there. It is a fun hobby, either way.

 

[nanook]

@MAB: I agree, that most people will not notice degraded caps unless it's one in a crucial circuit location. The start up reservoir for the IC in a SMPS is such a candidate - the unit will just not turn on anymore.

I just asked ChatGPT about early failure rate in aluminum electrolytic caps.
This infant-mortality-rate seems to be extremely small because the "quality" manufacturers do 100% burn-in. Manufacturers do not publish numbers however.
ChatGPT found one source: Young Electronics Group: 6 manufacturing related defects in 129 Mio. parts:

> Title: High reliability electrolytic capacitors
Autor: Jan Matthews, Technical Director, Young Electronics Group
Veröffentlicht am: 3. April 2013
Wichtige Passage:
„During the past six years YEG has shipped over 129 million pieces and experienced just six returns for a manufacturing defect. This issue was due to some microscopic scratching within the aluminium can of a particular capacitor.“

This underlines my personal experience.
I never came across a "new" capacitor in a unit that was faulty.
When recapping I probably wouldn't have noticed an "open" or degraded new cap, but degrading is very much related to drying out. I would however have noticed a short circuit.
Edit: And short circuit should be the dominant failure mode for infant-mortality since it's a manufacturing related consequence of defects in the oxide.

This burn-in probably is similar to the "gate-stress-test" in DMOS manufacturing. There you apply about 2x the voltage stated in the maximum ratings for a short time. This will very slightly degrade the oxide, but in case the oxide has a defect, it will break down, the gate current is way too high and the device is sorted "fail".

One can still debate if it's worth the effort to recap "all", but doing it will very likely not compromise reliability as proposed in an earlier post.
Edit: From the thoughts above I actually expect increased reliability and vastly increased lifetime.

 

[nanook]

Here is a nice paper from Cornell-Dubilier:

https://www.cde.com/resources/technical-papers/reliability.pdf

Here they give a number for infant - mortality under stress conditions and repeated stress conditions.

"During the manufacture of the capacitor, rated
voltage and temperature are applied in the aging
process. Capacitance, leakage current, and ESR are
tested on a 100% basis. Usually we employ addi-
tional screening techniques to attempt to weed out
infant mortalities. Such methods include burn-in,
surge voltage test, and hot DC leakage test. At CDE
at the present time, these screening methods may
weed out an additional 0.1-0.5% of weak capaci-
tors which may fail early in the field. The yield of
large high-voltage capacitors has increased from
92% in 1990 to over 98% today. If the burn-in or
other high--stress-screening processes are repeated,
a small percentage (0.02-0.2%) may be expected
to fail each time."

This statement suggests that if you keep away from these stress conditions, the infant-mortality-rate should be lower (probably a lot lower) than the 0.02 - 0.2% stated.

 

[levimax]

Here is a nice paper from Cornell-Dubilier:

https://www.cde.com/resources/technical-papers/reliability.pdf

Here they give a number for infant - mortality under stress conditions and repeated stress conditions.

"During the manufacture of the capacitor, rated
voltage and temperature are applied in the aging
process. Capacitance, leakage current, and ESR are
tested on a 100% basis. Usually we employ addi-
tional screening techniques to attempt to weed out
infant mortalities. Such methods include burn-in,
surge voltage test, and hot DC leakage test. At CDE
at the present time, these screening methods may
weed out an additional 0.1-0.5% of weak capaci-
tors which may fail early in the field. The yield of
large high-voltage capacitors has increased from
92% in 1990 to over 98% today. If the burn-in or
other high--stress-screening processes are repeated,
a small percentage (0.02-0.2%) may be expected
to fail each time."

This statement suggests that if you keep away from these stress conditions, the infant-mortality-rate should be lower (probably a lot lower) than the 0.02 - 0.2% stated.

Good quality new caps are no doubt good but if you have a cap on a board that you can measure for ESR and it is within spec it is also going to be very reliable (if not as long lived as a new cap). There are two other issues with replacing "good installed caps" with new ones. Often times new caps aren't available with the same ratings and especially the same size and shape so it my be hard to fit them in. In addition the chance of damaging something during the un-soldering and re-soldering process is going to be orders of magnitude higher than either a "old good" cap or "new" cap failing. Since it is a hobby it really doesn't matter either way but for me "If it isn't broken don't fix it" has served me better than "replace everything to try to make something old new again" which has gotten me into some serious trouble.

 

[nanook]

I don't want to turn down your and MAB's approach, I just hate to rework a unit a second time. And if it's about the gear at work that runs 24/7 I don't want to bother my colleagues with unnecessary down-time.

Often times new caps aren't available with the same ratings and especially the same size and shape so it my be hard to fit them

Today's caps have a higher CV per volume. I usually replace with the same diameter which is good for lifetime (although smaller caps with the same ratings would be available) and end up with a significantly higher voltage rating (which is good for reliability).

In addition the chance of damaging something during the un-soldering and re-soldering process

I do anyway have a thorough look at the solder joints and re-solder those that look suspicious. Components that get hot often have deteriorated solder joints.
For double-sided PCBs (especially if they forgot to use thermals in the planes) the risk is indeed to be considered.

I think both "philosophies" are equally fine, I mainly want to correct contradict the statements that warn about a high early fail risk of aluminum electrolytic caps.
To my opinion this is a solved problem - unless we stumble into something like the "capacitor plague" again (wrong formula for the electrolyte).

Capacitor plague - Wikipedia

en.m.wikipedia.org

 

[levimax]

For double-sided PCBs (especially if they forgot to use thermals in the planes) the risk is indeed to be considered.

Does anyone know the proper procedures (temperature, tip size, other tricks) for desoldering and re-soldering components on PCB's using "non-lead" solder? I have had a lot of problems with this.

 

[DVDdoug]

Does anyone know the proper procedures (temperature, tip size, other tricks) for desoldering and re-soldering components on PCB's using "non-lead" solder? I have had a lot of problems with this.

Too many variables...

Different solders have different melting points.

I don't have a lot of experience with lead-free solders, but with what I've tired I still like tin/lead, and that's what I use at home. I don't solder every day so I'm not afraid of lead poisoning.

Usually the soldering iron has to be quite a bit hotter than the melting point of the solder because once you touch the connection, heat is dissipated. A more massive tip holds the heat better as long as it's not too big for whatever space you have on the board.

Extra flux helps. I like the water soluble flux. But it does have to be cleaned-off because it's corrosive.

I also helps to keep the tip clean with a wet sponge and something like this.

I have solder wick and a solder sucker for desoldering, and sometimes I use a toothpick (or similar) to help clean-out hole after the component is removed.

I've also used some special low temperature solder. It's not really for soldering (it's brittle) but it has some "magic properties" and when it melts and it's touching regular solder, it mixes with the regular solder, making the "new alloy" that melts at a lower temperature. I used it with a surface mount rework station that that was something like a heat-gun "warming-up" a fairly large area of the board. When you add the low-melt solder you can remove a component without anything else on the board getting hot enough to come unsoldered.

 

[levimax]

I've also used some special low temperature solder. It's not really for soldering (it's brittle) but it has some "magic properties" and when it melts and it's touching regular solder, it mixes with the regular solder, making the "new alloy" that melts at a lower temperature. I used it with a surface mount rework station that that was something like a heat-gun "warming-up" a fairly large area of the board. When you add the low-melt solder you can remove a component without anything else on the board getting hot enough to come unsoldered.

This sounds like it would be very helpful, thank you!

 

[nanook]

Does anyone know the proper procedures (temperature, tip size, other tricks) for desoldering and re-soldering components on PCB's using "non-lead" solder?

I'm guess there are people around in this forum that have more experience than I have.
I have been working with lead-free solder for a couple of years at work but at home I usually use lead based solder. Edit: At work we usually are working on old equipment, so we use lead-based as well.

Here are my recommendations (and I'm looking forward to different opinions .
- avoid mixing lead-free and lead based
The composition of a mixture may be brittle afaik.
Use lead-free solder for unit being produced with lead-free solder (or remove the solder from the solder joint and use lead based solder)
- Temperature, lead-free: ca. 360°C
- Wattage: at least 80W (50W is fine for lead based, but marginal for lead-free)
- Tip shape: use a short tip when possible (better heat transfer to keep the time to a minimum)
-Tip size: use rather a thicker tip (better heat transfer to keep the time to a minomum)
- Desoldering of caps: apply additional solder and use a broad tip that covers both solder joints (only feasible up to 5mm pin spacing). Pull out the capacitor with some tool (will get too hot to use fingers). If legs are bent, try to push them straight with the soldering iron one by one before proceeding as described.
- For difficult cases: try to remove lead-free solder using good (fresh) solder-wick and "substitute" the lead-free solder with lead-based one (lower melting point). This procedure may need 2 rounds.
- Take care of your soldering iron tip. If it is not in good condition, you will likely damage the PCB because you are tempted to apply too much pressure to the solder joint (should ideally work almost without pressure). You may use "Tippy" which helps to keep the tip in good condition when working with lead-free solder.

For single-sided PCBs I strongly recommend a de-soldering iron (the ones with a hole in the tip that can apply vakuum once the solder is liquid).
But they are only reasonably affordable if you plan to utilize it a lot.

Good luck!

 

[nanook]

I've also used some special low temperature solder. It's not really for soldering (it's brittle) but it has some "magic properties" and when it melts and it's touching regular solder, it mixes with the regular solder, making the "new alloy" that melts at a lower temperature.

Thanks for this hint!
That sounds more promising than my strategy to "substitute" the lead-free solder with lead-based one

 

[Audiofire]

Some people get confused by the term cold solder joint, and outdated use of a standard soldering iron temperature (that is from the vacuum tube era). The correct approach is to understanding wetting of flux and liquidus temperature of solder. A more accurate or convenient term is dry solder joints.

 

[nanook]

@levimax: What exactly do you have problems with?
- SMD or THT (through-hole)?
- single-sided or dual - sided (or multilayer) PCB (plated through holes?)?

 

[JeffS7444]

Does anyone know the proper procedures (temperature, tip size, other tricks) for desoldering and re-soldering components on PCB's using "non-lead" solder? I have had a lot of problems with this.

Liquid flux is your friend.

 

[Dimitri]

Liquid flux is your friend.

Liquid flux and solder wick will absorb solder like a sponge much easier than solder wick alone. As for solder, I have no plans to use lead-free solder

 

[levimax]

@levimax: What exactly do you have problems with?
- SMD or THT (through-hole)?
- single-sided or dual - sided (or multilayer) PCB (plated through holes?)?

I have only messed around with through hole and find the solder hard to melt to remove a component without damaging the board. I think the "low temp solder" and some flux will go a long way. I don't use no lead solder but most devices made during the last 20 years or so seem to.