The Hall of Shame - Badcaps Photo Montage

[Behemot]

Not good idea, it will be leaking through the bad cap and making feedback crazy.

I got good deals on good caps all right, search no more

[eccerr0r]

It actually works now. The GPU crashes like mad with the bad caps, but with that parallel cap even with the longer wire ESL, it's sufficient to allow the GPU to stop crashing.

These caps originally were 1mF 6.3V 8mm units IIRC... I stuck another "bad" (phony/counterfeit) 3.3mF 16V cap in parallel.

[Behemot]

I got the Rubycons MBZ 1000/6.3 of fleebay. Not from datasheet, but look genuine allright, I guess I got lucky and actually got some real deal-old MoBo manufacturer custom stock…IIRC made in 2006, still well in spec, every single piece is fine.

[RukyCon]

This capacitor is from a cheap LED lamp i drowned in salt water.


Brand: GKE [Garbage Korroded Electrolytic]

[Behemot]

Wanna D8 Rubycons LLE 3.3/400? I've stocked them especially for lamps with close to no space inside.

[RukyCon]

It was one of those dirt cheap bulbs from ebay.
I wanted to see how they made it only 0.75 cents, and now i know.
And the LEDs i do get already have rubycons in them. [Well, most of them.]

[Behemot]

I don't actually have a photo cause I am too lazy for that, but got Foxconn G31MX (-S, -K) here with nine custom Rubycons MFZ 680/4 in D8×8 mm or so. All of them bloated. Gonna swap these for polymers…the rest are KZGs and some OST craps, will replace those too. I wonder if the board actually ever saw some use, or it was just kept as a backup, it is almost factory clean.

Has anybody ever seen bad caps around the audio codec, mostly 100/16? I usually replace them too when in the mood for full recap, but I wonder I have ever sen failed one so when I lack the time, I tend to leave those be. There can be a dozen of them and they are usually under no stress at all so even Evercon, G-LUXON and similar crap survives decades fine, no?

[Per Hansson]

Originally Posted by Behemot

I don't actually have a photo cause I am too lazy for that, but got Foxconn G31MX (-S, -K) here with nine custom Rubycons MFZ 680/4 in D8×8 mm or so. All of them bloated. Gonna swap these for polymers…the rest are KZGs and some OST craps, will replace those too. I wonder if the board actually ever saw some use, or it was just kept as a backup, it is almost factory clean.

All electrolytic caps with ultra low ESR are inherently unstable.
Seeing them fail after so many years, even when not used is not so strange.
Actually having them never powered on may be even worse for the caps.
Because they have no way to reform then...

Originally Posted by Behemot

Has anybody ever seen bad caps around the audio codec, mostly 100/16? I usually replace them too when in the mood for full recap, but I wonder I have ever sen failed one so when I lack the time, I tend to leave those be. There can be a dozen of them and they are usually under no stress at all so even Evercon, G-LUXON and similar crap survives decades fine, no?

Yea they do dry out eventually, but it takes a long long time...
I've had one such case:
http://www.badcaps.net/forum/showthr...912#post316912
But then back in the days I didn't use the boards built-in soundcard anyway because they where always crap so...

[Wester547]

Originally Posted by Per Hansson

All electrolytic caps with ultra low ESR are inherently unstable.
Seeing them fail after so many years, even when not used is not so strange.
Actually having them never powered on may be even worse for the caps.
Because they have no way to reform then...

Not sure I agree with that.

Panasonic FJ/FJS, FL, and Rubycon MBZ/MHZ, for an example, are ultra-low-Z capacitors, but generally have a very good track record and aren't known to die in NOS equipment or to bulge whilst still retaining their full leads (unsoldered), on their original looms, in a sealed bag, at room temperature. I've never seen or heard of a NOS Rubycon MCZ dying either, nor a NOS Samxon GD or GC (or GA).

As for storage time being worse for the caps, it's the oxide layer (the porous anodic one) that's supposed to deteriorate in storage, not the electrolyte. There are generally two reasons that the electrolyte would break down and produce excessive H2 (a process heat would only accelerate, even taking into account that phase change is more physical than chemical mechanism):

1) An unstable electrolytic composition. There is probably a happy medium to strike where using the right amount of additives is possible while still retaining conductivity and allowing for a reasonable breakdown voltage, but only a couple manufacturers seemed to find it.
2) High ripple current, especially when the electrolyte lacks the necessary depolarizers, oxidizers, and H2 absorbers to deal with electrolysis that occurs as the direct result of ripple current. It's high (AC) ripple current that will cause the chemistry to change and possibly even decompose. The bias applied to the plates has little effect in comparison, except for the hydrolysis which occurs when that happens.

Originally Posted by Behemot

I don't actually have a photo cause I am too lazy for that, but got Foxconn G31MX (-S, -K) here with nine custom Rubycons MFZ 680/4 in D8×8 mm or so. All of them bloated. Gonna swap these for polymers…the rest are KZGs and some OST craps, will replace those too. I wonder if the board actually ever saw some use, or it was just kept as a backup, it is almost factory clean.

Rubycon MFZ, AFAIK, aren't known to vent on the shelf, under good conditions, like KZG (especially those made between 2004-2007), KZJ, YEC, or Fuhjyyu and GSC/Sacon/Evercon. These are right under the CPU heatsink, right? That doesn't help. They probably died after exposure to surfeits of heat blowing down directly on them from the HSF. Still better than those 680µF 4V OST RLAs which were known to die within a matter of 6 months under the same conditions (many years ago). The only 680µF 4V 8x8 ultra-low-Z caps that seem truly reliable are those of the Panasonic FL variant, both the sleeveless and sleeved type.

Has anybody ever seen bad caps around the audio codec, mostly 100/16? I usually replace them too when in the mood for full recap, but I wonder I have ever sen failed one so when I lack the time, I tend to leave those be. There can be a dozen of them and they are usually under no stress at all so even Evercon, G-LUXON and similar crap survives decades fine, no?

The smaller caps usually dry out first. Cheap caps use cheap rubber seals whose properties allow the electrolyte to evaporate faster.

[Behemot]

Nope, not all; one of them was by the southbridge. And as it was really crystal clean (you can hardly ever clean dirty MoBo to a state it is as clean as new), including the CPU cooler, I think it was mounted to a case and put aside for years, till they ditched it for recycling.

[Per Hansson]

Wester547: I agree with you not agreeing with me
I must have been overly pessimistic when I wrote that
That said most caps are only rated for something like 1000 hours storage.
So it makes it difficult to go back to the manufacturer and claim a new one at least :P

Behemot: Reading your post again you stated:
"with nine custom Rubycons MFZ 680/4 in D8×8 mm or so. All of them bloated."
Actually that size of electrolytic capacitor seems very strange to me.
The height is obviously to counter polymer caps.
Yet at a size like that the ESR must be around 0.020 or so: Hardly "polymer range"
Found another post here with pictures of them:
http://www.badcaps.net/forum/showthr...023#post343023

[Wester547]

Originally Posted by Per Hansson

Wester547: I agree with you not agreeing with me
I must have been overly pessimistic when I wrote that
That said most caps are only rated for something like 1000 hours storage.
So it makes it difficult to go back to the manufacturer and claim a new one at least :P

That's at 105°C (or at the rated temperature for other capacitors), with no voltage applied and without conducting any ripple current. At room temperature, they should last much longer in storage. That shelf life specification pertains more to leakage current and the dielectric than anything else. Assuming that they aren't abused or that their internal chemical composition isn't askew, electrolytics should not outgas all on their own, even on the shelf.

Behemot: Reading your post again you stated:
"with nine custom Rubycons MFZ 680/4 in D8×8 mm or so. All of them bloated."
Actually that size of electrolytic capacitor seems very strange to me.
The height is obviously to counter polymer caps.
Yet at a size like that the ESR must be around 0.020 or so: Hardly "polymer range"
Found another post here with pictures of them:
http://www.badcaps.net/forum/showthr...023#post343023

Panasonic FL and Nichicon HN exist in those case sizes as well, in those same values. Same goes for NCC TMV and TMJ, and OST RLA (the RLA datasheet rates that case size for about 0.030Ω). I think the ripple ratings must be much lower for those electrolytics than comparable polymers, and the impedance values higher too, which probably means they are dissipating significantly more internal heat than their polymer counterparts (the high ripple current doesn't help), hence their higher failure rate.

[Per Hansson]

Originally Posted by Wester547

That's at 105°C (or at the rated temperature for other capacitors), with no voltage applied and without conducting any ripple current. At room temperature, they should last much longer in storage.

Ohh thanks, I didn't realize it referred to storage time at the capacitors rated temperature.
Now it makes allot more sense!

Originally Posted by Wester547

Panasonic FL and Nichicon HN exist in those case sizes as well, in those same values. Same goes for NCC TMV and TMJ, and OST RLA (the RLA datasheet rates that case size for 0.030Ω). I think the ripple ratings must be much lower for those electrolytics than comparable polymers, and the impedance values higher too, which probably means they are dissipating significantly more internal heat than their polymer counterparts (the high ripple current doesn't help), hence their higher failure rate.

Yes I would infer that a Rubycon MFZ 4v 680µF cap is only rated at something like 0.020ESR and 1000mA ripple current.
But without a datasheet we will never know for sure.
It's weird these caps where ever produced, as solid polymer caps with better specs existed back then too.

[Behemot]

Old engineers, or they just wanted to save few pennies Anyway, that board is running fine, as well as another G31 ASUS and others. Now only need to process some PSUs to run all that, that's gonna be a challenge with all these GLN craps which resist to getting repaired…

[momaka]

Originally Posted by Wester547

That's at 105°C (or at the rated temperature for other capacitors), with no voltage applied and without conducting any ripple current. At room temperature, they should last much longer in storage. That shelf life specification pertains more to leakage current and the dielectric than anything else.

And to add to that...

The shelf life is the *minimum* guaranteed time that a cap can do in storage at rated temperature. After this time, the datasheets mentions nothing in regard to how quickly the cap will fail afterwards or any deterioration. So even at maximum rated temperature, many caps will last longer than their rated shelf life.

Now lower the temperature to normal room temperature, and you can see why a cap rated for only 1000-2000 hours storage can make it more than a full year (about 8700 hours) without going bad. (Typically even 2-3 years will not damage a cap.)

That said, older (bigger) caps tend to have a much longer shelf life. Part of it likely has to do with the fact that older caps have thicker foil and oxide layer, so it takes more time to break that down than a modern small/compact cap. And that's why you can still find caps from the 70's and 80's still being okay.

[Behemot]

I have old stock of small ultra-low ESR Rubycons, some as old as a decade, all of them fine. Than old big can caps, some of them as old as 60 years (used though), fine too. Not sure what that has in common with size than…

[Wester547]

Originally Posted by momaka

And to add to that...

The shelf life is the *minimum* guaranteed time that a cap can do in storage at rated temperature. After this time, the datasheets mentions nothing in regard to how quickly the cap will fail afterwards or any deterioration. So even at maximum rated temperature, many caps will last longer than their rated shelf life.

Their specifications will likely drift further at that point. If a capacitor is rated for 1,000 hours at 105ºC with no bias, and with a 20% +/- tolerance for capacitance and 200% for dissipation factor, that tolerance will likely increase the longer the capacitors are left in storage at such high temperatures.

Now lower the temperature to normal room temperature, and you can see why a cap rated for only 1000-2000 hours storage can make it more than a full year (about 8700 hours) without going bad. (Typically even 2-3 years will not damage a cap.)

The shelf life spec ranges from 500 - 2000 hours at the rated temperature. I do wonder if the variation in the hour spec has more to do with the solderability of the wire terminals (and obviously the anodic layer of porous oxide film, as aforementioned) than the electrolytic composition.

That said, older (bigger) caps tend to have a much longer shelf life. Part of it likely has to do with the fact that older caps have thicker foil and oxide layer, so it takes more time to break that down than a modern small/compact cap. And that's why you can still find caps from the 70's and 80's still being okay.

Capacitance and breakdown voltage is inversely proportional to the thickness of the oxide layer. Higher capacitance, lower breakdown voltage, thinner oxide layer, higher leakage current. Lower capacitance, higher breakdown voltage, thicker oxide layer, lower leakage current. The formula for calculating leakage current in the datasheets also shows that leakage current rises as capacitance does. Capacitance also depends on the surface area of the aluminum foil and the pits etched into the aluminum foil in order to increase said surface area and thereby capacitance. As far as I know, the process of etching is rather controlled, and has improved over time as capacitors become more miniaturized.

Originally Posted by Behemot

I have old stock of small ultra-low ESR Rubycons, some as old as a decade, all of them fine. Than old big can caps, some of them as old as 60 years (used though), fine too. Not sure what that has in common with size than…

I'd guess those 60 year old caps are likely screw-terminal capacitors. I could be wrong, the large can snap-in capacitors seem to be rather formidable. Still, 60 year old caps (at least those smaller) likely have epoxy-coated rubber bungs or hermetic seals (such as metal, glass-metal, ceramic, etc) to have lasted that long without drying up or leaking. The rubber (and plastic) bungs on most electrolytics are bound to age and degrade eventually, hence the 15 year max life spec given by most manufacturers for caps sealed with rubber and/or plastic.

[momaka]

Originally Posted by Wester547

The rubber (and plastic) bungs on most electrolytics are bound to age and degrade eventually, hence the 15 year max life spec given by most manufacturers for caps sealed with rubber and/or plastic.

I've never seen a manufacturer mention a maximum life expectancy anywhere. Could you point me to where you found this?

IMO, 15 years is definitely NOT the maximum. Not even close. At least not what field tests show (i.e. caps in consumer gear). Even many crap caps will last more than that in general use locations (like speaker/amplifier rail smoothing caps, for example - in amplifiers with non-switching PSUs, of course).

I think 20-25 years is where a cap may start to degrade. And that still largely depends if the cap is/was abused or not. If not abused, I'd say expect 25+ years, especially from the quality Japanese brands.

Now as far as motherboard-grade ultra low ESR caps - yes, those may not last that long due to their electrolyte being more unstable. But still... 15 years just seems like a very pessimistic figure.

[Wester547]

https://www.chemi-con.co.jp/e/catalo...p-e-170401.pdf

http://www.rubycon.co.jp/en/products/alumi/pdf/Life.pdf

https://www.illinoiscapacitor.com/pd...capacitors.pdf

Scroll down to bulletin 3 on the NCC URL, the bottom of page 9 on the Rubycon URL, and page 2 on the Illinois URL. 15 years or 131,400 hours, whether that many hours has passed in storage or in use. Nichicon's capacitor lifetime calculator doesn't go beyond 131,400 hours either. Again, that only applies to rubber-sealed electrolytics, liquid or otherwise. I'd give 20 years (or 175,200 hours) to those epoxy-sealed and probably much longer than that for those hermetically sealed.

Liquid electrolytics aren't super long life components to begin with (like fans). Due to the fact that they rely on a liquid solution to achieve conductivity and retain their specs, with a polymeric (not hermetic) seal, they have a finite life. The electrolyte gradually evaporates and escapes, either by consumption of the self-healing mechanism or through rubber-bung diffusion. Nothing lasts forever.

That 15 year spec does not mean the caps are guaranteed to fail or go out-of-spec after 15 years - it simply means that their failure rate may hit a bathtub curve where the failure rate goes beyond that of the estimation of the manufacturer (IE, 1% after 15 years, 2% after 16 years, 4% after 17 years, 8% after 18 years, 16% after 19 years, 32% after 20 years, 64% after 21 years... you get the idea).

And that 15 year spec doesn't allude that the caps will outgas, leak, or bulge after 15 years either, only that they will very gradually dry out and very slowly go out-of-spec.

EDIT: Looking at the SMD capacitors in the link below that Per Hansson provided, SMD capacitors are the worst... those are known to dry out much faster than their through-hole next-of-kin. I'd give those wet SMD capacitors less than 15 years before they dry out.

[Per Hansson]

Very good links Wester, thanks!
Here is the view from a component integrator:
http://dl.mitsubishielectric.com/dl/.../sh030006j.pdf

With age, the electrolytic capacitor of the servo amplifier will deteriorate. To prevent a secondary accident
due to a fault, it is recommended to replace the electrolytic capacitor every 10 years when used in general
environment.

Also see the table on page: 9-1

This lifetimes spec seems quite accurate from what I've seen in the field too:
http://www.badcaps.net/forum/showthr...ighlight=servo