LC 475 prone to cap leakage as other LCs

[alectrona2988]

View attachment 58712


PowerMac G4 logic board with leaky capacitors.

The capacitors in Quicksilver PSUs are also garbage and prone to failure/leakage. Computers from that era are by no means problem-free and it's a bit silly to assume otherwise, honestly. They're twenty years old and electrolytics have comparatively short service lives.

I should go ahead and check mine. Ouch!!

 

[Forrest]

I bought some automotive grade thru hole capacitors recently. The specification sheet for many of these capacitors shows a working life of 1000 to 10000 hours at 105C. Automotive grade is extended temperature range compared to consume electronics and typically rated at 85C. My point is most older computers have already survived well in excess of their rated lifetime, but they won’t last forever. 10000 hours ain’t that much - 4 hours of use per day and they’ll last just over 10 years.

 

[joshc]

My point is most older computers have already survived well in excess of their rated lifetime, but they won’t last forever. 10000 hours ain’t that much - 4 hours of use per day and they’ll last just over 10 years.

This is the thing. These components all have a lifetime by design - and the usage and storage conditions affect it. It’s not as simple as saying post 1995 machines are less likely to leak. Depending on parts used, even a 10-15 year old machines may need a recap if someone wants to keep them going for a long time. I don’t think that many computer motherboards use electrolytics these days but power supplies and other applications absolutely do.

 

[3lectr1cPPC]

I agree 100%. As I’ve said I’m all for recapping later systems. I’ve recapped my 1997 PowerBook 3400c, my 1995 5300c, I’m prepping to recap my 1400c, have recapped a handful of 2000s era desktop motherboards, etc, and my Sawtooth is on the list. My only point was that failure is less common in many cases in these later systems, and physical leaking is much less common. Purely as an observation of how these systems are luckily seemingly less likely to start dying due to cap leakage breaking traces and such. As with failure itself, it depends on time, use, conditions, like you’ve said. We’re the weird ones using ancient technology far past its expiration date after all.
Sorry if I wasn’t clear enough earlier!

 

[pizzigri]

I would not at all say that only vintage computer fans are using ancient technology - not at all actually. Hifi nerds are much more common, and some of the 70-80’s stuff is built with incredibly crappy stuff, there once was a “computer grade” quality of caps after all. Let’s not even delve in tube audio. But ancient technology comes up in the most surprising places (for the layman), such as electronics labs, in the forms of analog and even digital meters of all kinds, frequency meters, DMMs, etc - some of which are incredibly accurate and extremely popular, even for today standards, look at the HP 34401A, it’s late 80s technology, and it’s a 6.5 digit DMM…
what is interesting is that many of these meters, even the ones from the 70s, rarely need their electrolytic caps to be replaced. I had an HP frequency meter that was built in 1983 that had better performing caps (in a quite expensive LCR meter) that the ones I bought to replace them.

 

[CircuitBored]

I didn’t say they were problem free, I said they don’t leak as often. I mentioned PSU issues too.
Anyway, that image is strange. Looks to me more like something was spilled there, the gunk on the top and sides doesn’t look like any cap leakage I’ve seen. Looks like it was stored in a wet environment and the metal just corroded.

The image looks a bit strange because it's a close-up taken with an iPhone so there's a lot of "AI" sharpening going on. The aluminium has split around the edge of the cap and the contents have leaked. You may be correct that it was caused by external factors but that doesn't change the fact that the contents of the cap are now elsewhere.

 

[CircuitBored]

I would not at all say that only vintage computer fans are using ancient technology - not at all actually. Hifi nerds are much more common, and some of the 70-80’s stuff is built with incredibly crappy stuff, there once was a “computer grade” quality of caps after all. Let’s not even delve in tube audio. But ancient technology comes up in the most surprising places (for the layman), such as electronics labs, in the forms of analog and even digital meters of all kinds, frequency meters, DMMs, etc - some of which are incredibly accurate and extremely popular, even for today standards, look at the HP 34401A, it’s late 80s technology, and it’s a 6.5 digit DMM…
what is interesting is that many of these meters, even the ones from the 70s, rarely need their electrolytic caps to be replaced. I had an HP frequency meter that was built in 1983 that had better performing caps (in a quite expensive LCR meter) that the ones I bought to replace them.

HP is almost the exception that proves the rule... I have an unmodified HP oscilloscope that still reads accurately and it's from 1989. Their equipment from that era is legendarily reliable. It helps that it was all very expensive stuff, even compared to an Apple computer from the late 80s. My 'scope has gold traces instead of copper, for example. That said, there will still come a day when those electrolytic caps just don't work well enough any more.

 

[Phipli]

My 'scope has gold traces instead of copper, for example.

Gold plated, they're still copper underneath.

 

[Phipli]

That white stuff on the cap is the main thing - I've never seen a failure like that before, in images or otherwise.

The white is aluminium oxide. If seen it form on caps fitted to boards that have been badly stored in sheds and basements. The metal can used by those electrolytics is aluminium.

Its rust Jim, but not as we know it.

 

[s_pupp]

I bought some automotive grade thru hole capacitors recently. The specification sheet for many of these capacitors shows a working life of 1000 to 10000 hours at 105C. Automotive grade is extended temperature range compared to consume electronics and typically rated at 85C. My point is most older computers have already survived well in excess of their rated lifetime, but they won’t last forever. 10000 hours ain’t that much - 4 hours of use per day and they’ll last just over 10 years.

Bear in mind that the rated lifetime is based on use at the rated temp. For every 10 degree Celcius drop in actual temp, life expectancy doubles. A 10,000 hour capacitor will likely die a natural death from age-related deterioration before reaching its use-based life expectancy.

 

[feltel]

I started renovating the machines today. I started with the PSUs as I'm waiting for parts for the boards itself. Two TDK PSUs had visual cap leakage going on. The PCBs were coated with a shiny sticky film. After removing all caps on the low voltage side I soaked them in IPA. After a short time and a little bit brushing the IPA turned brownish. A clear indication that caps have leaked.



After recapping both PSUs (used Panasonic caps with 105° rating and slightly higher voltage ratings) one started working right away. All voltages look healthy. The second TDK PSU has issues even with new caps. The voltages are jumping between reasonable values and 2-3 V within seconds. Also the PSU ticks in a slow to medium rate. After switching of mains the voltages look okay. 5V are 5V and 12V is slighly below 12V. It lasts for a few seconds than the voltages are dropping as expected. I checked my soldering work and orientation of the caps. It seems okay. Could the possible fault be on the high voltage side? I did not change the big cap and the small one next to it.

 

[Phipli]

I did not change the big cap and the small one next to it.

Swap the small one in from the other PSU to see if it helps? Just to test.

 

[feltel]

Swap the small one in from the other PSU to see if it helps? Just to test.

I replaced the 8.2uf Cap with an 10uf Cap, which stabelized the PSU somewhat. The voltages are now in the expected ranges (except -5V, which is around -3,6V), but they are changeing by .1V ever so often. The other TDK PSU does not show this behaviour. I expect to receive my Apple to VGA adapter every other day so I can test the PSUs with an actual motherboard.

 

[Phipli]

but they are changeing by .1V ever so often

0.1v variation doesn't sound like an issue to me. They're not that well regulated new. The voltages on most of my old macs vary enough for the fan speed to waver up and down.

-3.6 is more of an issue. Is this loaded?

 

[feltel]

Measured while nothing connected to the PSU.

 

[Phipli]

Measured while nothing connected to the PSU.

You can't rely on readings taken with no load. Its sort of meaningless. Some PSUs won't even be stable enough to switch on like that. For an unloaded PSU, that is really good performance.

You need to plug it in to a logic board, or a dummy load.

 

[feltel]

Hmmm, okay. The measurements on the other TDK are fine and "spot on" while nothing is connected.

 

[Phipli]

Hmmm, okay. The measurements on the other TDK are fine and "spot on" while nothing is connected.

But that doesn't really mean anything. PSUs have a minimum as well as a maximum load specification.