TK Capacitors

[Pentium4]

Okay I have a couple questions.

Is there a way to tell if TK caps are resleeved OST, or ones that were actually made in Japan? I'm curious because this Asus P5GC-MX/1333 motherboard has all TK caps except on the VRM + a few pannies by the RAM. The system has 44,165 running hours on it (It was used to scrub mail 24/7) and none of the TK caps are bulged. The PSU feeding it had bad caps on the 12V rail, and the 3.3V rail. The failing Rubycon cap on the 5VSB also read 0.62Ω (Keep in mind this PSU was manufactured in Feb, 2000) So I guess these were sourced from Japan, or maybe got lucky? Since this PSU has been in use since 2000, and this board was installed in 2008, I have no idea when those JPCE-TUR caps bulged, it could have been the entire time! Since the caps on the 5V rail were just fine (Still tested 0.03Ω) this is the only rail that would have had clean ripple the entire time of use. There's 17 TK caps and they're all 820uF 6.3V, which leads me to my next question:

When a motherboard, such as this one is being used with a power supply with bad capacitors, clearly the PSU will create lots of ripple. So with all this ripple going into the motherboard, are the capacitors the only thing that are seriously stressed by the ripple? Or does this damage the chipset as well? I ask because this thing already has 44,000 hours on it so if the ripple has damaged more than the caps, I don't know if I want to spend the time, or caps (assuming those TK's are resleeved OST)


Appreciate any replies or input

[c_hegge]

Re: TK Capacitors

TK usually have a slightly different vent stamp to OST. OSTs usually have a + vent, where all lines are the same length, whereas on TK, two of the lines are slightly longer (a bit like -+-). That said, though, there is no such thing as a Japanese made TK cap. OST re-sleeves or not, they are all junk. I've seen the ones with the non-OST vent stamps fail after only 12 months in service before (and that's with a good PSU and not 24/7 use).

[Pentium4]

Re: TK Capacitors

TK usually have a slightly different vent stamp to OST. OSTs usually have a + vent, where all lines are the same length, whereas on TK, two of the lines are slightly longer (a bit like -+-). That said, though, there is no such thing as a Japanese made TK cap. OST re-sleeves or not, they are all junk. I've seen the ones with the non-OST vent stamps fail after only 12 months in service before (and that's with a good PSU and not 24/7 use).

Cool, thanks for the info. I guess it must have helped a lot that it was in an air conditioned server room the entire time, so 18C ambient temperature. Still not sure if it's worth recapping. It's a nice board, but I wonder how worn the chemi con polymers on the VRM are after all that ripple from the PSU. The two 2200uF 16V JPCE-TUR on the 12V rail read 25.0Ω

[tmiha71]

Re: TK Capacitors

When a motherboard, such as this one is being used with a power supply with bad capacitors, clearly the PSU will create lots of ripple.

Not always true, if there's enough capacitance (and resultant ESR is low enough) on motherboard - they (caps on mobo on that rail - thinking on 3.3V rail) will act as source for loads (on that rail) on motherboard, and PSU is asked to VERY slowly raise output current (even if it has dead caps, it will succeed in this - ripple will be low (1), and mobo will work /will POST, and maybe boot/... But how long will they survive (caps on mobo) ?

To put it simple, switching regulators, and LDO's, will pull current from where it is available (maybe wrong approach, they should pull current from PSU always)...

Watch for your hard disk(s) if you use SATA power connector (and hd uses 3.3V :-))...

(1) excessive ripple = NO BOOT :-))

This is not recommendation to keep things as they are, caps should be changed...

By the way PSU is (make and model, maybe not suited for mobo's which power CPU from 12V ), and CPU is (model) ?

[momaka]

Re: TK Capacitors

Board should be fine. As tmiha71 explained, even if the caps in PSU went bad, the caps on the motherboard probably "took" a larger portion of the ripple to make things work, but that likely didn't significantly shorten their life (unless they were running extremely hot). I'd say replace the TK caps and be done with it.

Since the caps in the PSU that filter the 12V rail were bad, the HDD probably took the most damage as it probably ran quite hot.

[Pentium4]

Re: TK Capacitors

Cool, good to know! Yeah, I know that the motherboard caps take a beating when PSU caps go bad, just didn't know if anything else could have been damaged. I was extremely impressed with the HDD's attached to that PSU, it was three Seagate Barracuda 7200.10 80GB drives, all tested good and with no bad sectors. Once again though, cold server room had to have helped quite a bit. Looks like I'll be saving this one, and probably reapply thermal paste to the chipset

Another question: What is the likeliness of small value caps on motherboards to fail? I know that I've seen small value caps in PSU's (Even down to 1uF) fail, I know that motherboards work differently. Brands such as Lelon/Teapo, etc. should I worry about values such as 100uF 16V, 470uF 6.3V, etc? If they're general purpose caps, should they be okay?

[momaka]

Re: TK Capacitors

Another question: What is the likeliness of small value caps on motherboards to fail?

I would guess somewhat low, as they don't see a lot of stress unless they are near a hot component.

should I worry about values such as 100uF 16V, 470uF 6.3V, etc? If they're general purpose caps, should they be okay?

Most of the time, these caps filter outputs of small linear regulators (like for example, for audio IC, LAN, I/O IC, etc.). Therefore, they don't see a lot of stress. If I see a motherboard that appears to be flaky even after all of its major capacitors have been replaced, I then replace the small ones, too, as they can cause erratic behavior. Usually, however, there are at least 2 of them in parallel, so even if they start going bad, they are still okay for the linear regulator to keep voltages in spec.

The 470uF 6.3V ones may be filtering power going to the PCI slots, USB ports, or some other device that may draw more power, so I advise you check what they are connected to (first check if they are connected to 3.3V or 5V rail on PSU, if not them check for connections to linear regulators).

[tmiha71]

Re: TK Capacitors

Again, partly true...

Yeah, I know that the motherboard caps take a beating when PSU caps go bad,...

As seen on this mobo, PSU caps were beaten by mobo caps, probably too low esr, or too much capacitance (more likely), or both (on motherboard)...

Some of better mobo manufacturers, are givig their best to keep capacitance on +3,3V rail and +5V rail as low as possible. Some manufacturers just patch everything with capacitance...

[PeteS in CA]

Re: TK Capacitors

Again, partly true...



As seen on this mobo, PSU caps were beaten by mobo caps, probably too low esr, or too much capacitance (more likely), or both (on motherboard)...

Some of better mobo manufacturers, are givig their best to keep capacitance on +3,3V rail and +5V rail as low as possible. Some manufacturers just patch everything with capacitance...

If you think of the components from the output of the P/S rectifiers to the pins of load (i.e. the various ICs) as a filter subsystem - CPU and MB designers view it that way - the failure of any component in that subsystem to perform its "job" increases stress on the other components.

[tmiha71]

Re: TK Capacitors

^
Agree with that, but as said caps on mobo took dominant rule on that rail, also killing PSU caps -> most stressed components are/were PSU caps (weakest link - they died, presuming there's no manufacturing defect), also in that process other components on that rail were under increased stress from beginning, and also in time while PSU caps were good...

Question is, does mobo need TK (CPU-vrm output) grade of caps directly on some rail - if it needs probably mobo has a lot of other problems...

Apologize for pretty lose definition of excessive ripple (no boot)...

[Pentium4]

Re: TK Capacitors

I wish I knew when the caps in the PSU went bad, but I have no way of knowing. Since the PSU has over 100,000 running hours on it, it could have been any time. Around the VRM, it smells burnt. Could just be all the dust around it though, all the FETs tested good.

[Pentium4]

Re: TK Capacitors

Not a single TK cap was that bad ESR wise, although 0.06Ω is starting to get up there for a motherboard (Only two read that high) Most read 0.04Ω
On the other hand, most of their capacitance was out of spec on the high side. They all ranged from 964uF-1,106uF with 14/17 being 1,032uF or higher

[PeteS in CA]

Re: TK Capacitors

If my mental math is correct, the max capacitance for those 820uF caps should be 984uF. That almost all were out of spec high suggests - if their electrolyte is water-based - that the electrolyte is attacking the aluminum oxide layer on the anode foil. That means that the dielectric is thinning, increasing the capacitance, and may thin to the point that the circuit voltage will break down some weak point in the dielectric, which will result in swelling and venting. If this MB is something you want to use, I urge replacing all of them, even the one(s) that is/are barely in spec.

[Pentium4]

Re: TK Capacitors

If my mental math is correct, the max capacitance for those 820uF caps should be 984uF. That almost all were out of spec high suggests - if their electrolyte is water-based - that the electrolyte is attacking the aluminum oxide layer on the anode foil. That means that the dielectric is thinning, increasing the capacitance, and may thin to the point that the circuit voltage will break down some weak point in the dielectric, which will result in swelling and venting.

Thanks for the detailed description. I did not know that. I guess that explains how some cheap caps I've seen read over 5KuF but not high ESR. Isn't a water based electrolyte the cheaper method? My guess is they didn't have much time left till they started popping, especially running 24/7 with a bad PSU!

If this MB is something you want to use, I urge replacing all of them, even the one(s) that is/are barely in spec.

I will. Whether they were out of spec or not, they'd be going into the scrap pile I was debating whether or not I want to replace those Panny's at the top but I think they will be fine. Going to replace all of them with nichicon HN. I'll post a picture when it's finished.

[tmiha71]

Re: TK Capacitors

That almost all were out of spec high suggests - if their electrolyte is water-based - that the electrolyte is attacking the aluminum oxide layer on the anode foil.

So, they were killed by too low voltage ? Or they were killed by too high ripple current ? Or by both ?

[Pentium4]

Re: TK Capacitors

Ahhh what a relief, it looks so much better I just ran a 2 hour memtest and it hasn't had any errors or freezing, I will test it in Windows 7 later.

[PeteS in CA]

Re: TK Capacitors

So, they were killed by too low voltage ? Or they were killed by too high ripple current ? Or by both ?

No and no. Water-based electrolytes are used in ultra-low impedance parts, but require extreme purity of materials, precision in manufacture, and special additives that neutralize that type of electrolyte's natural tendency to attack the aluminum oxide. I'm no chemist, but over time the neutralizer gets "consumed" (or otherwise loses its effectiveness), and eventually the electrolyte attacks the aluminum oxide anyway, leading to catastrophic failure. With parts from better manufacturers - e.g. Rubycon, Nichicon, Panasonic, some UCC - and good component selection by board/equipment designer-manufacturers, that happens after the whatever-it-is has been replaced due to obsolescence. Parts that have really bad materials and/or wrong recipes can actually fail in sealed bags without being installed in electronic equipment, so operating voltage or ripple current need not be a factor (though heat can accelerate chemical reactions).