"just replace the caps", factual guidelines for prolonging the life of devices?

[tshlns]

Hello everyone,

I have been dabbling with electronics for a long time but consider myself an utter newbie since it is not my primary hobby nor have I any training in it except for what the Finnish school curriculum and websites offer.

My question relates to old hardware - both computers (Amigas, Commodore 64's, IBM XT/AT PC's) and music equipment (Roland, Yamaha, Korg etc. synthesizers). So, basically hardware from the beginning of 80's up to the end of 90's.

In hobbyist forums there are many rules of thumb particularly regarding some equipment (like C64 PSU's) where the most common guides given for restoring/maintaining the devices are "replace the caps and voltage regulators", as there are some bad designs and real-life examples of PSUs destroying IC's.

But, in some circles, particularly in music hardware, people casually say they "recap" old equipment just to be safe (I'm referring in particular to electrolytic caps), and my question is: what would be an objective metric for deciding when to do this? Particularly with some boards the component density is quite high and the traces are small so replacing dozens of caps is not as simple as with some VIC20 era hardware where everything is big and distances between components are more friendly towards hand-soldering.

Mouser lists 2000 to 10000 hours lifespan for most electrolytic caps I've googled, but OTOH regardless of the use the electrolytics eventually dry out. So, regardless of cap bulging, I am assuming there is a case to be made for potentially recapping equipment that is closing in on being 40 years old?

In addition, many of the often repeated rules of thumb are to replace voltage regulators and transistors due to wear. I've googled some datasheets from e.g. TI and I find no mention of MTBF/lifespan in hours for them. Am I correct in assuming that their aging due to normal rated use is neglible?

Could anyone give me a hint on where to look for reliable (i.e. not just repeated "thruths" online) information on parts that SHOULD be replaced on old electronics in order to protect those components (custom IC's in particular) that cannot be replaced? I am not looking to do unnecessary placebo fixes that risk damaging the equipment during unsoldering/resoldering, but justified preventive care.

ps. I have an ok soldering station (i've managed SMD soldering with it) and an ok 50€ multimeter but no oscilloscope nor a necessary level of skill to properly diagnose a complex board like found on an 90's synth, so I'm more interested in discussion on whether this whole trend of replacing old components with new is more destructive than useful and whether I should continue using the gear and only fix gear reactively WHEN a component has failed rather than proactively.

Thanks in advance for any hints!

[Per Hansson]

Well in your quite specific examples there can be some general guidelines:
Amiga computers that uses SMD capacitors need to be replaced, because they leak electrolyte that destroys the boards!
On the other hand the old Amiga computers with regular through hole (often Rubycon) are just fine.
But then again these old Amigas with through hole caps usually have a NiCad battery that has leaked and destroyed the whole board so...

I have not seen widespread reports of bad caps in old synthesizers, but opening them up and making sure all solder joints near SMD caps look shiny is a good advice.

As for Commodore power supplies AFAIK they are known to fail and overvolt when they do, such a power supply is better replaced by something modern with proper over voltage protection.
I think this is where the logic of replacing the regulators come from: that perhaps they have been exposed to high voltages spikes and are near failure after 40 years...

P.S: You can't compare specs of modern capacitors on Mouser with those 20 to 40 year old caps, they all where rated 2000h back then.
It is a modern development with capacitors with such long lifespans.

[Peabody]

Electronics actually is my primary hobby, but I have the same concerns.  In my limited experience, electrolytic capacitors have been the primary cause of failure.  That has been the case in the last three repairs I've done.  In all three cases, it was the output capacitors in power supplies.  I don't think I've ever needed to replace a voltage regulator, but of course that would depend  on whether it has been exposed to excessive voltage, or allowed to overheat.

I really wonder if it makes sense to replace dozens of capacitors just because they are old.  I suspect in the end the great majority of them will be fine.  But I've been looking at getting an LCR meter to check capacitance and particularly ESR.  I understand that a good meter can be used to check capacitors in circuit, which would be a big help.  Dave has a video on that.  The key to in-circuit testing appears to be having the test signals be low voltage so they are below what would go through a diode or transistor.  I was hoping one of these really cheap $7-20 meters would work, but they all appear to have 5V signals instead of 0.5V or lower.  The cheapest low-voltage meter I've seen is about $60.  Maybe someone here knows of a cheaper one.

[Per Hansson]

I understand that a good meter can be used to check capacitors in circuit, which would be a big help.  Dave has a video on that.  The key to in-circuit testing appears to be having the test signals be low voltage so they are below what would go through a diode or transistor.  I was hoping one of these really cheap $7-20 meters would work, but they all appear to have 5V signals instead of 0.5V or lower.  The cheapest low-voltage meter I've seen is about $60.  Maybe someone here knows of a cheaper one.

If your component tester has been compiled with the menu and option enabled the feature is available:

C+ESR@TP1:3 The additional function "C+ESR@TP1:3" selects a stand-alone capacity measurement
with ESR (Equivalent Series Resistance) measurement at the test pins TP1 and
TP3. Capacities from 2µF up to 50mF can be measured. Because the measurement voltage is
only about 300mV , in most cases the capacitor can be measured "in circuit" without previous
disassembling. If the POWER OFF option is selected in the Makefile, the count of measurements
is limited to 250, but can be started immediately again. The series of measurements can
be finnished with a long key press.

Source: https://github.com/kubi48/TransistorTester-documentation/blob/main/pdftex/english/ttester.pdf

Edit: scope shots of the 300mV test mode vs default 5v test mode: https://www.eevblog.com/forum/repair/ultra-cheap-lcr-meter-to-test-capacitors-in-circuit/msg4339822/#msg4339822

[Peabody]

Thanks very much.  It looks like I have a lot of reading to do.  Is this the product of the 300+ page thread here?

[Per Hansson]

Thanks very much.  It looks like I have a lot of reading to do.  Is this the product of the 300+ page thread here?

Yes: https://www.eevblog.com/forum/testgear/$20-lcr-esr-transistor-checker-project/

[tkamiya]

Listed lifespan for electrolytic caps are only valid when using said cap at rated voltage.  If you are using 6.3V rated cap at 3.3V; for example, lifespan will be much longer.  How much longer is usually not listed, though.

I don't usually replace caps unless I have a reason to do so.

[Per Hansson]

Listed lifespan for electrolytic caps are only valid when using said cap at rated voltage.  If you are using 6.3V rated cap at 3.3V; for example, lifespan will be much longer.  How much longer is usually not listed, though.

I don't usually replace caps unless I have a reason to do so.

No, rated lifespan is at rated temperature and ripple rating, and the general rule of thumb is that for each reduction of 10°C you double the capacitors lifetime.
So a 105°C 2000h rated capacitor running at 55°C has an estimated lifetime of: 64000h (at max rated ripple!)

[Microdoser]

If you're dealing with computers of that age, make sure to replace any batteries you find. Battery leakage is one of the things that will damage old boards the most.

It's possible that over time, some through hole components will have separated their holes from the boards because of physical trauma, or the solder may have dislodged from the hole because of a dry joint that was just holding on enough to pass QC but not good enough to last 30 years. This can cause odd, intermittent problems that are hard to trace. A physical check under a scope and a continuity check of any suspicious joints would be prudent. Generally, reflowing those holes does the trick.

If your goal is to essentially give the product as much of a trouble free lifespan as you can, I personally would replace the caps too, especially 'can' type or tantalum caps.

There are some types of C64 power brick which are known as 'machine killers', it would be worth googling which type they are, I can't remember off the top of my head. I feel certain a modern power supply of the right type would help prolong machine life.

Getting a boardview + schematic etc will help if you need to diagnose issues.

Good Luck!

[tshlns]

Thank you all for excellent responses and discussion! Per was right that my initial post was a bit rambly and not the clearest one possible to start off.

What you described (e.g. leaking batteries and leaking caps) is something I've experienced (and repaired) myself and I consider these the "obvious cases" which are easy to both diagnose and repair - of course assuming that the leaking has not caused excessive damage.

What I am more worried about are latent problems that could manifest in a sudden failure and destruction of sensitive components but show no signs before the failure. I.e. faults that let too much voltage or current through. Thus the question in my initial post boils down to:

"are there components that should in general be replaced on ~30-40 year old hardware in order to preventively minimize the chances of a destructive failure and maximize the lifespan of the device AND does a general hobbyist with just a basic multimeter stand a chance in diagnosing these components to judge their state and urgency to replace?"

Considering that with most vintage hardware we have no idea of the conditions the device has been subjected to, my initial guesses have been electrolytic caps (can leakage current increase gradually throughout years? can a sudden burst damage components downstream?) and voltage regulators (what are their general failure modes? increasing or decreasing voltage?). So in essence I'm trying to understand, which are the components that fail gracefully and which do not

[Microdoser]

Generally, the only thing that will fail in such a way that it destroys other components is the power supply. If it suddenly starts putting out too much voltage, or the current limiting fails etc that is the sort of thing that will fry boards.

On the motherboards, the most common thing to fail will be caps, and they fail short circuit. This sort of acts like a board protector and lowers the voltage down to almost nothing. Of course, they get hot, and they may even burn or explode if the power going through them is enough.

Watch some of the videos from this guy. He seems to be doing exactly what you plan to do. You might pick up some tips. In this video he is fixing a C64.



I'd say things will be far harder to diagnose if you don't have some sort of scope, even one of those really cheap ones, so you can see if data lines have sensible things going on.

[mariush]

A lot of surface mount electrolytic capacitors have a bad reputation for leaking, much more than through hole electrolytic capacitors.

If you have a device with lots of such surface mounted electrolytic capacitors, it's worth checking them with an esr meter (after desoldering a lead from the board, because there may be multiple capacitors in parallel and that affects the measurement)

Norcal on Youtube has a bunch of videos where the cause of failures were these surface mounted capacitors, for example see the video below where he replaces 100+ such capacitors with through hole versions because these surface mount went bad due to overheating

https://youtu.be/p1zhfYHUIIY?t=565

Someone already told you about the 10 degree less = double lifetime .. keep in mind that applies only for electrolytic capacitors. For solid (polymer) capacitors, there's a different formula... basically their life is much much bigger.

Look for cracked solder joints and bad solder joints as well, for example another of his videos shows both bad/unreliable solder joints and dead surface mount electrolytic capacitors :

[wizard69]

Thank you all for excellent responses and discussion! Per was right that my initial post was a bit rambly and not the clearest one possible to start off.

Ramble away.    However stating questions concisely always helps.   In any event just so you know where I'm coming from, my experience is industrial maintenance focused.   As such I've seen a lot of power supply failures over the years.   When we see a failure, especially on packaged and supported hardware, we just upgrade the power supply to the manufactures latest rev.   For controllers and other automation hardware, this gives you a supported and hopefully improved hardware.    For everything else we often do our own upgrades and seldom repair power supplies.   The exception is custom supplies, often from manufactures that have gone under or don;t support, that are repaired if possible (crowbars can be very damaging).

What you described (e.g. leaking batteries and leaking caps) is something I've experienced (and repaired) myself and I consider these the "obvious cases" which are easy to both diagnose and repair - of course assuming that the leaking has not caused excessive damage.

There are well known obvious things to look for.   Bad solder joints being one.

What I am more worried about are latent problems that could manifest in a sudden failure and destruction of sensitive components but show no signs before the failure. I.e. faults that let too much voltage or current through. Thus the question in my initial post boils down to:

There is not a good solution to address bad design.    As somebody mentioned above, sometimes a modern well designed power supply is called for.

"are there components that should in general be replaced on ~30-40 year old hardware in order to preventively minimize the chances of a destructive failure and maximize the lifespan of the device AND does a general hobbyist with just a basic multimeter stand a chance in diagnosing these components to judge their state and urgency to replace?"

A multimeter can solve plenty of problems.    As for the details of what should be looked at and what can fail, I'd suggest spending some time with youtube videos from "MrCarlson's Lab".   I believe that is the name of the youtube site.   In any event there is a lot to learn.

As for "recaps" sometimes it is highly suggested and seems to be a thing with Guitar amps.

Considering that with most vintage hardware we have no idea of the conditions the device has been subjected to, my initial guesses have been electrolytic caps (can leakage current increase gradually throughout years? can a sudden burst damage components downstream?) and voltage regulators (what are their general failure modes? increasing or decreasing voltage?). So in essence I'm trying to understand, which are the components that fail gracefully and which do not

The best answer here is a qualified yes.   A cap failure can cause other components to fail.   There are many factors to consider, such as the position of the cap in the circuit, any safety built in.   In a nut shell all electrolytic caps fail eventually, but modern caps and better engineering often means very long run times.  However that doesn't mean everything is designed well and has high quality components installed.  Just a couple of months ago I was involved in pulling some hardware from manufacturing, in this case PC motherboards with obviously bad caps.    So even with relatively recent hardware caps are still a problem.

[El Rubio]

Recapping vintage audio gear is common. I think it is a bit of cheap insurance in an ampnor receiver from the mid 1970’s. Some people think it’s too invasive and may damage the traces on the board etc. i have recapped many amps and receivers and found many caps way out of specs. The larger power supply filter caps are not usually the ones that fail ( at least what I have seen in audio gear), but the smaller voltage value electrolytics. Some folks will replace voltGe regulators with new ones rated for higher current because they increased the filter capacitor values in the P.A.

I follow a few basic rules when recapping, not hard rules, but my preferences. I use high ripple caps in power supplies, low leakage in phono preamps, general caps or low leakage in most other circuit applications. Small electrolytics in the audio signal path get replaced with film caps, and I tend to replace all tantalum caps as they often short when they fail whereas electrolytics and film caps more likely to go open.

One thing I find in audio circles is the assumption by some that a full recap will fix whatever is wrong with the device. Almost never is the case. Forst rule is fix or isolate any faults before recapping.

I don’t buy into audio grade caps like Elna Simic II. If somebody wants those put in, sure I will put them in, but i don’t believe the “magic” audio properties associated with them.

[Per Hansson]

Thank you all for excellent responses and discussion! Per was right that my initial post was a bit rambly and not the clearest one possible to start off.

I don't know if I said that? I may have been thinking it though but I hope you can't hear my thoughts? 

What I am more worried about are latent problems that could manifest in a sudden failure and destruction of sensitive components but show no signs before the failure. I.e. faults that let too much voltage or current through. Thus the question in my initial post boils down to:

"are there components that should in general be replaced on ~30-40 year old hardware in order to preventively minimize the chances of a destructive failure and maximize the lifespan of the device AND does a general hobbyist with just a basic multimeter stand a chance in diagnosing these components to judge their state and urgency to replace?"

Considering that with most vintage hardware we have no idea of the conditions the device has been subjected to, my initial guesses have been electrolytic caps (can leakage current increase gradually throughout years? can a sudden burst damage components downstream?) and voltage regulators (what are their general failure modes? increasing or decreasing voltage?). So in essence I'm trying to understand, which are the components that fail gracefully and which do not

I never owned a Commodore or Amiga but I fond it therapeutical to watch Adrian's Digital basement restorations of such (linked by Microdoser above).
You can learn allot from his videos, a specific type of "MT" brand RAM is very often bad in C64's, along with glue logic (74-series chips) built in their own FAB: MOS Technology.
This does not cause destructive problems though, and neither do capacitors in general, the only thing that is destructive is the failure of a power supply, but I covered that in my previous post.
P.S: There are also several other great Youtube channels if this is your main interest, like Jan Beta, The 8-Bit Guy & RMC - The Cave to name just a few.