Rubycon MBZ leaks in storage!

[mockingbird]

Re: Rubycon MBZ leaks in storage!

Don't know about FL but in accordance to an old post by PCBONEZ, they should be equal to FJ

Did you see anything about FL possibly being lower in ESR than FJ? I can distinctly remember reading that somewhere.

Actually, premature failures of Samxon GC/GD definitely exist...

Wow! Thanks for posting that. First time I'm seeing failures. I've had good experience with 3300uF GC, ran them in a pretty hot environment, none of them bulged after a couple of years (That Foxconnn board I posted earlier... Now I'm starting to think that maybe my PSU is fine and the motherboard was the problem, I'll have to re-test this PSU).

And we all know how unreliable Samxon GF and GK are, and they are equal to HE and HD respectively, so I don't think that can bode well for them either.

What about GT and RS? I have RS in the SP-450 that I suspect of failure, but the caps aren't bulged or anything, nor have there been any reports of failures with RS AFAIK.

[Wester547]

Re: Rubycon MBZ leaks in storage!

Did you see anything about FL possibly being lower in ESR than FJ? I can distinctly remember reading that somewhere.

Here you go:

https://www.badcaps.net/forum/showpo...8&postcount=22

Wow! Thanks for posting that. First time I'm seeing failures. I've had good experience with 3300uF GC, ran them in a pretty hot environment, none of them bulged after a couple of years (That Foxconnn board I posted earlier... Now I'm starting to think that maybe my PSU is fine and the motherboard was the problem, I'll have to re-test this PSU).

Samxon has released notices about fake capacitors on the market like Rubycon, Chemi-con, and Nichicon have so it could be simply that. I've always suspected that's the real reason for those huge bad batches of HMs and HNs. I noticed the stenciling quality on their sleeves is significantly worse than that which is found on genuine capacitors ordered from Nichicon and authorized distributors. Might be the reason for the prematurely failed MCZs, MFZs, and MBZs as well (like what the OP found in this thread), and maybe even some of the worst KZGs and KZJs. I honestly think that the Chinese have become skilled enough to fake them well enough that you wouldn't know the difference just by looking at them. See this thread:

https://www.badcaps.net/forum/showth...hp?t=556&pp=15

Foxconn supposedly gets their capacitors directly from the factory, but I wouldn't be surprised if they sometimes sourced their components from "trusted" Chinese and Taiwanese suppliers in order to accelerate production times.

What about GT and RS? I have RS in the SP-450 that I suspect of failure, but the caps aren't bulged or anything, nor have there been any reports of failures with RS AFAIK.

Only other unreliable capacitor that I'm aware of from Samxon is the GL series. No datasheet for those, but those are notorious for going open without any signs whatsoever so it could simply be that the bungs aren't secured well on the bottom but with potentially aqueous electrolyte that the electrolyte could evaporate so fast that there simply isn't enough left to leak from the bung (just a guess...).

As far as I know, those Smart Powers have a very weak fan controller by default so the fan spins so slow in them that they will eventually cook any lytic...

[LLLlllou]

Re: Rubycon MBZ leaks in storage!

Here you go:

https://www.badcaps.net/forum/showpo...8&postcount=22

Contrary to what PCBONES said, it just doesn't make any sense to me that Panasonic would make two different series with identical specs. It's MUCH more likely that the FLs are equivalent to HN/MCZ. The tolerance of my ESR meter is insufficient to verify with any certainty, just as it can't reliably distinguish HMs from HNs.

DAMN YOU PANASONIC FOR NOT RELEASING THE FL DATA SHEET!!!!!

Either way, FJ and FL are just about as good as you can get with wet electrolytics. Even Dell USFFs can't kill them.

[Wester547]

Re: Rubycon MBZ leaks in storage!

Contrary to what PCBONES said, it just doesn't make any sense to me that Panasonic would make two different series with identical specs. It's MUCH more likely that the FLs are equivalent to HN/MCZ. The tolerance of my ESR meter is insufficient to verify with any certainty, just as it can't reliably distinguish HMs from HNs.

Assuming that the leads are of the same length, those kinds of ESR meters do seem to have a 0.005 +- ohms limitation.

I don't understand it either but it makes sense to me because... on some of those Asus P4SD boards and P4x-800 boards they interchangeably use 1500uF 6.3V 8x20 KZGs, FJs, and 1800uF 6.3V 8x20 MCZs and FLs in the VRM output along with either Fujitsu RE polymers or Sanyo OS-CON SP polymers (or SEPC or NCC PSA IIRC). 8x20 MCZs and 8x20 KZGs are already of a different range ESR wise. But that could just be ASUS being ASUS...

And on some 865 series Intel/Foxconn boards, they use 2200uF 10V Rubycon MHZ (10x23) instead of MBZ interchangeably with 2200uF 10V Nichicon HM (10x25 but those tend to be blown very often since they are of that "age"). Why not just use MBZ instead? Or on some ASUS boards where they use 1000uF 6.3V TMZs instead of KZGs (once again another custom order).... manufacturers probably have reasons for doing such things that don't seem logical to us because we didn't engineer or design the board. And we don't have datasheets for these custom orders either so all we can do is conjecture. With that in mind, I could believe that FL is equal to KZG, especially since FL is as durable as FJ.

Either way, FJ and FL are just about as good as you can get with wet electrolytics. Even Dell USFFs can't kill them.

Dell USFFs do eventually kill them, it just takes longer. And sometimes they do have a tendency to dry up without any signs, whereas MCZ are more likely to physically spill their guts (but they can sometimes go open without signs too).

More to the original topic, while I agree that the pair of MBZs are quite "aged" at this point, I never thought I would see a MBZ bloat in storage. I would have thought I'd sooner see that happen to MCZ...

Rubycon, I am disappoint.

[mockingbird]

Re: Rubycon MBZ leaks in storage!

Speaking of the unreliability of Samxon Ultra Low ESR, I just happened upon this pic from this post over at Vogons:



I'd like to get Topcat's take on this. He must have re-capped dozens of boards with them.

[stj]

Re: Rubycon MBZ leaks in storage!

i think pana FL is equal to MCZ because i'v seen ASuS boards with either.

i suspect they simply switched to FL because panasonic is cheaper than rubycon.

[Wester547]

Re: Rubycon MBZ leaks in storage!

It looks like this isn't the first time something like this has happened...

https://www.badcaps.net/forum/showthread.php?t=8312

Looks like it happened to a Panasonic FJ as well! But since that one was acquired from eBay, it could have been fake despite looking authentic. Actually, it could definitely be fake because FJ/FL are only sold directly to OEMs so the fact that they still retain their full leads is suspicious, unless someone acquired them from some OEM who Panasonic sold them to. And further down the thread, there's a KZJ bulging without having ever been soldered into a circuit - go figure! Seems like those ultra low ESR, highly aqueous capacitors are bound to do that sooner than later.

[mockingbird]

Re: Rubycon MBZ leaks in storage!

Here's another report of reputable capacitors leaking in storage. This time Nichicon HD, date code H0137.









I've had these for a while, took them out of an old motherboard I think. I've have some Nichicon HD from the same factory ("H") and very similar datecodes, but they're 10mm, and none of them show any signs of bulging whatsoever. So I wonder if nichicon has or had the same problems with the 8mm version of the series, like Teapo has with 8mm SC until today.

[Wester547]

Re: Rubycon MBZ leaks in storage!

^ Those 8mm 1500uF 6.3V Nichicon HDs are just more fragile. I suspect that the 8x20 case size pushes the chemistry a bit too far, though theoretically the highly etched aluminum foil should preclude that (having to resort to thinning the dielectric too much in order to increase capacitance in smaller case sizes). See here:

https://www.badcaps.net/forum/showthread.php?t=189

And here:

https://www.badcaps.net/forum/showthread.php?t=3242

And here:

http://www.slotware.net/blog/tag/vc-31l

All the failed capacitors are 8x20 1500uF 6.3V Nichicon HDs. But go figure - the 2200uF 6.3V 10x25 and 1200uF 16V 10x25 Nichicon HDs in the second URL are just fine (visibly anyway)! On the other hand, I noticed all these custom HDs (if they aren't counterfeits - I take it those HDs have the proper bungs) aren't afar from 14 years old in terms of shelf life, so maybe it's just their time. And I've come to the opinion over the years that all Teapos are unreliable if not most of them, they just fail without showing it many a time - several times too many.

The "H" prefix stands for Ohno, Japan. I wonder if it has to do with the fact that Nichicon were having some problems around that time period (with their HM and HN series but possibly all their Hx series of the timeframes).

[mockingbird]

Re: Rubycon MBZ leaks in storage!

You make some excellent points, as usual...

I don't have any caps that are direct series equivalents to compare too (From the same year and in the same diameter). AFAIK, UCC KZE, Nichicon HD, and IC KXM have an almost identical if not identical spec. Panasonic FM is a bit different IIRC, eventhough it's on the same level (speaking of which, I don't ever see any bad Panasonic caps, not even old 85C ones in 90s equipment). So if UCC or IC ever made an equivalent cap that year, it would be interesting to gauge their performance.

I take it those HDs have the proper bungs) aren't afar from 14 years old in terms of shelf life, so maybe it's just their time. And I've come to the opinion over the years that all Teapos are unreliable if not most of them, they just fail without showing it many a time - several times too many.

Yup, I got them off a motherboard. I wish I could remember which one. I do have some busted 10mm HDs in my box of bad caps, but they were killed in circuit because of the conditions, and not in storage. I also have some of those unsleeved HCs that were killed on a very hot Radeon. But HC is only a 1000-2000 hour series, unlike HD which is 3000 for 8mm and 4000 hours for 10mm.

Come to think of it, I wouldn't mind at all if Nichicon reps (and other manufacturer reps) came on this forum once in a while and gave us all a little inside scoop

The "H" prefix stands for Ohno, Japan. I wonder if it has to do with the fact that Nichicon were having some problems around that time period (with their HM and HN series but possibly all their Hx series of the timeframes).

The HM overfill issue came to my mind right away. But then I looked at my 10mm HDs which are still fine. I need to find an inexpensive method of reforming my capacitors. I was thinking of using different wall warts of varied voltage along with some sort of load. I can't afford a bench PSU, nor do I want to be stuck with the bulk if I ever had to move. Do you have any ideas for a poor man's cap reformer?

[Wester547]

Re: Rubycon MBZ leaks in storage!

You make some excellent points, as usual...

Thanks.

AFAIK, UCC KZE, Nichicon HD, and IC KXM have an almost identical if not identical spec.

Nichicon HD, Chemi-con KZE, Rubycon ZL, Panasonic FF, and Sanyo WX are all direct crosses to one another. Of them all, I have found KZE and ZL to be the most durable. HD a little less so as this thread is partially proof as to, and WX seem to be a little too sensitive to heat for their own good (more so than WG, oddly enough, but not as bad as WF) to put it in one manner (but certainly not notorious for failing like KZGs). I don't know about FF because I haven't seen them on motherboards much but I'd guess they're very reliable.

Panasonic FM is a bit different IIRC, eventhough it's on the same level (speaking of which, I don't ever see any bad Panasonic caps, not even old 85C ones in 90s equipment). So if UCC or IC ever made an equivalent cap that year, it would be interesting to gauge their performance.

I've only ever heard of bad Panasonic FJx and FL being mentioned as far as alarming failures go (late 2006 datecodes and more recently, IIRC) - some premature failures have been documented here and on two other forums. The issue with this is that because we only have one datasheet for FJ, and because it's 12-13 years old, there's really nothing except anecdotal experiences to determine the reliability of these "ultra low ESR" capacitor series. Without some earnest equipment we can only speculate the real reasons as to why they fail prematurely when they aren't overheated or abused.

Yup, I got them off a motherboard. I wish I could remember which one.

Possibly an ASUS board. Those custom HDs in the last two URLs failed on ASUS boards.

I do have some busted 10mm HDs in my box of bad caps, but they were killed in circuit because of the conditions, and not in storage. I also have some of those unsleeved HCs that were killed on a very hot Radeon. But HC is only a 1000-2000 hour series, unlike HD which is 3000 for 8mm and 4000 hours for 10mm.

Yes, I remember that thread. They were in laminated cases, indeed, which does not help their heat dissipation at all. They overheated on that card. Though I would guess that Rubycon ZA is more durable than HC if I absolutely had to - ZA is rated up to 3,000 hours.

Come to think of it, I wouldn't mind at all if Nichicon reps (and other manufacturer reps) came on this forum once in a while and gave us all a little inside scoop

It would be nice, but I'm pretty sure that sort of information isn't something they'd let us be privy to. Such secrets are closely guarded for the same reasons that these companies won't sell their materials to any other companies with the exception of some of their subsidiaries. If they were to offer any such information, I'm pretty sure they would be more inclined to post it publicly to their websites.

The HM overfill issue came to my mind right away.

Actually, the "overfill" issue was just a rumor that was never supported or proven to be true. It is a common misconception that Nichicon came out, conceded to the issue, and pulled the faulty capacitors. No such thing ever happened. That announcement came from a university in Iowa, IIRC:

https://www.badcaps.net/forum/showpo...3&postcount=24

It did not come from Nichicon, and it can be surmised that they simply saw the capacitors were leaking heavily and came to the inference that they might be "filled" with too much liquid - easy for the neophyte to conclude. The only announcement that came from Nichicon at the time came from Rubycon and Chemi-con as well, and it has to do with using certain disinfectants in cleaning the capacitors in order to avert SARS - no mention or relation to HM or HN failures whatsoever. It was Dell and HP, as I recall, that announced that the motherboards shipped after April 2004 wouldn't have faulty capacitors... but Dell has been caught intentionally lying to their customers before, so it's safe to say they didn't know what they were talking about.

Furthermore, premature HM and HN failures with 2005 datecodes were still documented here and sometimes even 2006 and 2007 datecodes for HM here. One possible indicator that the issue was resolved was that the early HN and HZ datasheets show that the electrolyte is rated for -40C to +105C, but the datasheets show -25C to +105C after August 2004 (the HM ratings remained the same, though), so that could mean they switched to another water-based solvent. It could have also just been an error in the datasheet, however.

I am still willing to give Nichicon the benefit of the doubt because the HM and HN failure rate dramatically decreased over the years and there was always the chance of fakes.

As to cap reforming... do you have some sort of battery or adapter that you could use in connection with the capacitors in question, if they were connected to a resistor of no more than 100 kiloohms? It would heavily depend on what voltage the original capacitors were rated for.

[Wester547]

Re: Rubycon MBZ leaks in storage!

My mistake, the last URL shows a FIC motherboard, so those HDs possibly came off a FIC or ASUS motherboard.

[mockingbird]

Re: Rubycon MBZ leaks in storage!

As to cap reforming... do you have some sort of battery or adapter that you could use in connection with the capacitors in question, if they were connected to a resistor of no more than 100 kiloohms? It would heavily depend on what voltage the original capacitors were rated for.

I have to read up on cap reforming again. I'm pretty sure some PDF by NIC or one of the other manufacturers was posted here a while back, I'll have to dig that up. I've got hundreds and hundreds of caps. Speaking of which, I desoldered some 2002-era Panasonic FJs this week. They still look to be in good shape.

As for voltage source, IIRC, reforming calls for raising the voltage incrementally. So starting off with 6.3V to 16V caps, I'm thinking of using 5V switched PSUs of which I have plenty, and then migrating to 7V and 12V adapters. So those would be the latent 3 stages. The question is how much current to load the cap with.

Also, what's the best way to step down a switched 5V supply to say anywhere between 1-3V which is what I'd like to start the cap at? Would something like a 1117 regulator be sufficient, and would I have to design a circuit around it or can I just use it by itself?

[Wester547]

Re: Rubycon MBZ leaks in storage!

I have to read up on cap reforming again. I'm pretty sure some PDF by NIC or one of the other manufacturers was posted here a while back, I'll have to dig that up. I've got hundreds and hundreds of caps. Speaking of which, I desoldered some 2002-era Panasonic FJs this week. They still look to be in good shape.

Yes, those 2002-2003 Panasonic FJs are extremely reliable. I've never seen one fail for an example on those original Xbox motherboards. Those would usually have 3300uF 10V Panasonic FJs in 12.5x20mm (VRM input) and 680uF 16V Panasonic FJs in 8x15mm. The ones I saw had July - August 2003 date codes. The KZEs (1500uF 6.3V 10x20, August to September 2003 date codes) that were present on the board were also fine along with the Sanyo WGs (3300uF 6.3V, July - September 2003 datecodes) in the VRM output.

The only capacitors I've seen blown on those older Xbox motherboards are the 3300uF 6.3V Nichicon HMs in the VRM input and output, 2002-2005 date codes, and sometimes the 1F 2.5V Nichicon UC clock capacitors (I think it's because they're ran right at their maximum voltage, though, and double layer electrolytics don't take to that well). And of course the Aerogel caps don't hesitate to leak from the bottom but usually with no formation of hydrogen gas, probably because of dreck bungs.

As for voltage source, IIRC, reforming calls for raising the voltage incrementally. So starting off with 6.3V to 16V caps, I'm thinking of using 5V switched PSUs of which I have plenty, and then migrating to 7V and 12V adapters. So those would be the latent 3 stages. The question is how much current to load the cap with.

Yes, for about 30 minutes or more at a time, it would help to apply a voltage in 0.5V increments or thereabouts. The current limiting is determined by ohms ratting (the watts it will be dissipating) on the resistor.

Also, what's the best way to step down a switched 5V supply to say anywhere between 1-3V which is what I'd like to start the cap at? Would something like a 1117 regulator be sufficient, and would I have to design a circuit around it or can I just use it by itself?

IIRC, the fixed 1117 regulators from the commonly known manufactures come with the ability to step the voltage down to +3.3V, +2.85V, +2.5V, +1.8V, and/or +1.2V. Would that suit your purposes?

[momaka]

Re: Rubycon MBZ leaks in storage!

Do you have any ideas for a poor man's cap reformer?

1 KOhm potentiometer + 10-100 KOhm resistor will do fine.

Connect pot fixed pins between ground and your power supply. Adjust so that voltage on the variable resistance pin is right at the maximum voltage the cap is rated for (i.e. something like 15.8V for a 16V cap). Connect variable resistance pin to 10-100 KOhm resistor. Connect other end of resistor to the caps (+). Let caps sit for a few hours (but as little as 30 minutes will reform them fairly well too). Don't worry about stepping the voltage. The 10-100 KOhm resistance will limit the current going to the caps, which in turn will drop the voltage across them. The voltage across them, however, will slowly increase as they reform to higher and higher voltage. So it is a very simple process and you don't need to do anything.

For power supply, use 6-9V for 6.3V caps, and 16-20V for 10V and 16V caps.
As for the 10-100 KOhm resistor - you pick the resistance based on how many caps you are going to charge and what their capacitance is. For large capacitances (over 3000 uF), I'd say use something around 47 KOhm. For higher capacitance, use 10-47 KOhms, and for lower capacitance (<470 uF) use above 47 KOhms (100 KOhm will be okay).

[mockingbird]

Re: Rubycon MBZ leaks in storage!

Wow, brilliant post as usual momaka.

Do you have any advice on which voltage to start off at for each voltage rating (i.e. 6.3v 10v, 16v, etc...)

Also, is it safe to parallel this setup, and how much amperage does it draw on average?

[mockingbird]

Re: Rubycon MBZ leaks in storage!

Ok, I found this adapter lying around. After opening it up, I can see that the way this works is that the transformer has many taps, each tap is for a different voltage. At first I assumed that the switch was some sort of pot.



I replaced the old rotting "Nova" brand 25V 1000uF with a brand new 2014 datecode 25V 1000uF KYA (Which I bought from our great forum member Behemot).

So this is perfect for my needs. I'm going to go up to the appropriate voltage which is just under the cap voltage limit (And this will either kill caps which have had their etching conform to a lower voltage after use in circuit or raise them to their labeled voltage).

I'm also planning on using a breadboard of some sort to hook up a jig where I can do several caps at a time.

My question is this:

The adapter says 300mA maximum. Is this for all the voltage taps, or only the 12V (In which case, the lower taps can handle more). Also, based on the adapter's amperage rating, how many caps can I do in parallel per voltage?

[momaka]

Re: Rubycon MBZ leaks in storage!

Wow, brilliant post as usual momaka.

Do you have any advice on which voltage to start off at for each voltage rating (i.e. 6.3v 10v, 16v, etc...)

Thank you.

You can start right at the maximum voltage for each cap (i.e. about 6V for 6.3V caps), if you use a series resistor. The 10-100 KOhm resistor in series with the cap(s) (that are to be reformed) will limit the current, so the cap(s) will not get nuked, even if they came from a circuit where they were sitting on a much lower voltage.

If you don't have the series resistor, then YES, you will need to start at a low voltage and slowly increase it (like 0.5-1V increments) in 30-60 minute intervals. But I don't recommend that. The whole idea behind the series resistor is to make the reforming process easier. You put the max cap voltage, plug in the caps, and set it and forget it for a few hours.

And you can have multiple caps in parallel with the setup I mentioned in my previous post. However, the more caps you add in parallel, the lower the series resistance to use. Don't go under 10 KOhms, though. Not that anything bad will happen if you do, as it is not precise rocket science.

As for current draw: it will be very LOW with the series resistor. If you use a series resistor with the cap(s) you are reforming, the maximum current will be the power supply voltage, V, divided by the resistance of the series resistor, R. Or I = V/R, good old Ohm's law. So with a power supply outputting 6V and a 10 KOhm series resistance, the current will be 0.6 mA max. And that's only when the cap(s) is (or are) fully discharged. Once the cap(s) start charging, the current will drop even more.

Let me know if this is still not clear. I can draw a circuit diagram for you tomorrow if you like.

[LDSisHere]

Re: Rubycon MBZ leaks in storage!

FWIW the output voltage on those adapters is usually higher then the setting until it is under some sort of load and even then it is not precise. You may want to keep a multimeter on it until while you are using it to see what it is actually putting out voltage wise.

[mockingbird]

Re: Rubycon MBZ leaks in storage!

Thank you.
If you don't have the series resistor, then YES, you will need to start at a low voltage and slowly increase it (like 0.5-1V increments) in 30-60 minute intervals. But I don't recommend that. The whole idea behind the series resistor is to make the reforming process easier. You put the max cap voltage, plug in the caps, and set it and forget it for a few hours.

Are you saying that overvolting caps with very low current will not harm the cap? In that case, there's no reason to reform the cap at a lower voltage, just at a low current, correct?

As for current draw: it will be very LOW with the series resistor. If you use a series resistor with the cap(s) you are reforming, the maximum current will be the power supply voltage, V, divided by the resistance of the series resistor, R. Or I = V/R, good old Ohm's law. So with a power supply outputting 6V and a 10 KOhm series resistance, the current will be 0.6 mA max. And that's only when the cap(s) is (or are) fully discharged. Once the cap(s) start charging, the current will drop even more.

Thanks, this makes perfect sense. So maybe I should set up some sort of decade resistance array hooked up to some kind of switching system on my jig. Each step will draw more and more current until it's up to the maximum current the cap can handle. I'm thinking a computer PSU would be good for this. I don't think I have any 12V adapters that can do more than 1A. But with a PC PSU, I can go all the way to the maximum of the 12V rail, or at least up to the 2A - 5A of current some of these caps handle.

Let me know if this is still not clear. I can draw a circuit diagram for you tomorrow if you like.

That's fine. I drew something up in LT Spice. I'm still trying to figure out how to measure amperage in the simulation so I can quickly test different amperage draws with different resistors by running the simulation.

FWIW the output voltage on those adapters is usually higher then the setting until it is under some sort of load and even then it is not precise. You may want to keep a multimeter on it until while you are using it to see what it is actually putting out voltage wise.

Right. And LT Spice has a provision for this in the voltage source settings. I just don't know what parameters to put in because I'm such a noob. And if momaka says it's necessary or even ok to reform it with the full current load, I'll probably end up using a PC PSU as the voltage source.