junky Jetway P4MDPT board

[zunasthegreat]

Sorry for hijacking OP. Will the NCC PSA series 2.5V 680uf Solidcap work for the VRM area on a junky Jetway P4MDPT board ? Originals are "Evercom" 1500uf 6.3V and are crap, 16V Highside VRM has same Evercom 16V 1500uf that I'm gonna replace with some Nichicon HD series @ 2200uf.

[momaka]

Originally Posted by zunasthegreat

Sorry for hijacking OP. Will the NCC PSA series 2.5V 680uf Solidcap work for the VRM area on a junky Jetway P4MDPT board ?

Looking at pictures of the board online... Most likely, I'd say the answer is YES.

Given how many caps it has around the CPU, converting it to all-polymer will still probably give good enough capacity on the CPU VRM low side. This can be important for older boards (and especially budget ones like PC Chips and ECS) that used relatively low VRM switching frequency. But your board's CPU VRM appears as relatively medium frequency design, so solid/poly caps should work OK.

That said, you don't have to go all solid or all wet electrolytic caps. You can mix and match too to get both higher capacitance and better ESR in the circuit. That's what I did with my Jetway N2PAP Lite, detailed here:
https://www.badcaps.net/forum/showthread.php?t=63041

In case of your board, you can probably do like 2/3b of the caps on the CPU VRM low side as polymers, and 1/3 or so wet electrolytic with high capacitance.

Also, as "junky" as this Jetway may appear, notice they used a through-hole socket for the CPU. That actually makes the board quite bullet-proof, especially when subjected to the horrible warping from the stock Intel heatsink mounting mechanism.

[zunasthegreat]

That jetway on the link looks so OP with proper caps. I'm awaiting some nichicon HD's to arrive so I'll consider mix and max like u said. Btw I still have around a Lelon-Modded Soltek P4 board that still goes strong from ~2012.

[momaka]

Originally Posted by zunasthegreat

Btw I still have around a Lelon-Modded Soltek P4 board that still goes strong from ~2012.

Nice.
A lot of older boards didn't care as much about low-ESR caps (particularly early Pentium 4 boards and ones older than that.)

The arrangement of those caps look similar to an ECS P4VXASD2+ I have. That one came stock with OST RLX and all caps are still doing OK, despite the board abusing the VRM high-side caps with a ton of heat. Then again, that board doesn't care about low-ESR caps either. Its CPU VRM is one of those inefficient 2-phase designs, done with Schottky diodes as the free-wheeling devices and a KA7500 IC for PWM control (read; relatively slow switching frequency.) Oh, and the funniest part is that it uses the 5V rail to power the CPU instead of 12V. The board is quite good for abusing/testing the 5V rail on old PSUs.

[zunasthegreat]

For lowside VRM cap bank I'll mix solid 820uf 2.5V (sepc series) + some nichicon 2200uf 6.3V. The 16V ones will be replaced with Nichicon HD's 1500uf 16V.

*My cheap tester shows Vloss above ~5% on the NCC solid 680uf 2.5V caps.
Crappy reading or whole batch of semi-working caps ? Other caps on tester show the usual 0.5-1% Vloss so I guess its not tester error.

[momaka]

IDK, I think I've seen poly caps with high Vloss like that before. I guess just connect that cap to a 2-2.5V source and see if it holds voltage for, say, over 10 minutes. If yes, it's probably not as leaky as the meter shows. I honestly don't pay too much attention to Vloss on my meter, as some of my lytic caps (and I think a few polies too) easily show 3-4%.

If you have a small low-voltage DC motor (like out of a CD drive eject mechanism or tape player or similar), you can also connect the charged cap across it to see if the motor spins for an instant. If it does, then the ESR and capacitance are probably reading correct and cap should be OK.

[zunasthegreat]

KZG Job done ... 6.3V 1500uf and 3x Nichicon 1500uf 16V. Mainboard posts normal so far. The old GSC 1500uf caps are a mystery, 6.3V ones show very high Vloss on meter ~5% but their capacity is epic at around ~2600uf.

[momaka]

Originally Posted by zunasthegreat

KZG Job done ... 6.3V 1500uf and 3x Nichicon 1500uf 16V. Mainboard posts normal so far.

Nice work!

Keep an eye on those KZG caps, though. Generally, they've had a lot less issues with newer batches than older ones (from like 2005-2007), but they still may fail occasionally. By far, their biggest enemy seems to be sitting on the shelf without seeing any voltage / use. So the more you use that motherboard, probably the better off they will likely be.

On that note, I just recently posted about a socket 775 ASUS P5GC-MX motherboard with KZG caps, where pretty much every KZG failed. Of course, note that this is an older motherboard from 2006-2007 era, when KZG really was problematic, so that's probably why so many of them failed. Newer KZG seem to be doing OK for the most part and I tend to leave them alone (but still keep an eye on them when there's a chance! )

Originally Posted by zunasthegreat

The old GSC 1500uf caps are a mystery, 6.3V ones show very high Vloss on meter ~5% but their capacity is epic at around ~2600uf.

Ah, classic GSC / Evercon failure mode before they start bulging. Looks like you caught them right on time!

The high capacitance means their electrolyte is breaking down the aluminum foil layer and increasing the capacitance at the expense of decreasing the maximum voltage they can tolerate. As a result, their internal leakage current also goes up, making the cap more "lossy", especially when operated at voltages closer to the cap's rating. Eventually, this will keep getting worse until the cap "just loses it" one day and partially begins to conduct/short out, making it bulge and then leak.

So again, you caught them on time.

Though IME, GSC / Evercon tend to break down rather slowly over time, so they sometimes allow the device to continue working for quite a bit before they finally fail completely.

On a related note, the GSC caps from the Jetway N2PAP-LITE motherboard I linked to above in post #2 have all started failing in the same exact way - extremely high capacitance and high Vloss... though none have bulged yet. I've used them as test caps in a few low-stress circuits, and they are just about adequate for that - but never going to use them permanently in anything again, because they will get worse further and further over time.

That being said, if a capacitor is showing a capacitance that is more than 20-25% higher than what the cap is rated for, then suspect a possible near future failure. The old UCC KZG also actually fail in this manner (as do many other unstable caps.) So checking the capacitance of loose caps in your stock from time to time can sometimes tell you if any cap may be failure pending to happen.

[zunasthegreat]

Great tips there thanks a lot. I recently tried to repair a very bizzare LC-8460BTX PSU that mad a horrible buzz sound, the amount of cost savings the maker did was just nuts. PSU had in total 3 filtering caps for 12V/3.3V/5V voltages, one each and half the pcb was empty. The 12V cap gave a reading of 35pF ...

[momaka]

Originally Posted by zunasthegreat

I recently tried to repair a very bizzare LC-8460BTX PSU that mad a horrible buzz sound, the amount of cost savings the maker did was just nuts. PSU had in total 3 filtering caps for 12V/3.3V/5V voltages, one each and half the pcb was empty. The 12V cap gave a reading of 35pF ...

I think I have a pretty good idea about the PSU you're dealing with. LC-84xxBTX is a model number for Solytech / Deer / L&C / Allied PSUs. Probably somewhat similar to these:
https://www.badcaps.net/forum/showpo...&postcount=491
https://www.badcaps.net/forum/showpo...&postcount=492

Sometimes they are total crap (many parts missing and PCB not having enough space to put decent parts), and sometimes they could be rebuilt with a few spare parts from other PSUs and also good caps.

They key, in general, to getting these cheapo PSUs working somewhat half-decent is to install 2x 2200 uF at least on each of the major output rails (that is, 3.3V, 5V, and 12V). 5VSB will do OK with 2x 1000 uF. There is also one or two 50V, 47 uF caps in there - those must be replaced too (one will be on the primary side, close to the 5VSB transfomer/circuit and is very important so that the 5VSB doesn't overheat and possibly take the motherboard with it.) If you add the above caps, the PSU should be a lot better, though may still need other work - for example, if the PI inductors are missing and replaced with jumper wire, it may be a good idea to install those. Likewise, the input EMI/RFI filtering components are also often missing on the cheaper units. Getting those installed should make the PSU passable (though not great by any stretch of the imagination.) Of course, only do it if you have the spare parts from other PSUs and electronics. It's probably not worth to buy all these parts for that cheap PSU (well, the electrolytic caps are cheap, but the EMI/RFI filter parts will not be, usually.)

On that note, I just finished rebuilding a somewhat similar cheapo PSU unit from another company (CWT). I've had it for a while and figured I can use it with some older PCs. Here's the thread with that:
https://www.badcaps.net/forum/showthread.php?t=49205

[zunasthegreat]

Playing with such terribly made PSU's is risky, it doesn't worth it even if you use the crappiest hardware. For the record I found a good deal on a FSP-300W-85%, will post at some point the internal part quality.

[momaka]

Originally Posted by zunasthegreat

Playing with such terribly made PSU's is risky, it doesn't worth it even if you use the crappiest hardware.

Yeah, I agree.
In general, you'll probably spend more time and money on parts than what a new and well-built PSU is worth. But some people live in parts of the world where this isn't always the case. So sometimes you just have to make do with what you have. In my case, I like fixing these cheap PSUs out of curiosity. Some of them aren't too terrible and can be made into basic functional PSUs again that aren't dangerous. On that note, the cheap Solytech / Deer / Allied PSUs tend to have really robust and reliable design that's based on the ancient (but well-understood) half-bridge topology. So with the right parts in there, they can be reliable.

Here's one I rebuilt about 12 years ago and still use occasionally in an off-line Pentium 3 computer:
https://www.badcaps.net/forum/showpo...&postcount=490

Originally Posted by zunasthegreat

For the record I found a good deal on a FSP-300W-85%, will post at some point the internal part quality.

Sounds good.

And yes, here I can get used or NOS (new old stock) PSUs for $14 or less shipped to my door, and I'm talking about name-brands like Delta, LiteOn, HiPro, FPS, and Bestec. Sure some of the used ones may need new caps, but it's well worth the effort. I recapped three HiPro 300/350 Watt units just a few months ago, posted here:
https://www.badcaps.net/forum/showthread.php?t=81905
^ These are rock-solid with new capacitors. They don't have APFC and efficiency is only around 75-80% at best, probably. But that's also why they tend to be so reliable.

[zunasthegreat]

I opened up a Nitrox 550W psu with a bloated 16V 2200uf Nippon something cap, had to band-aid it for now with a Lelon 25V 2200uf. From the following caps which one is the ideal : Nichicon PW 2200uf 16V / HD 2200uf 16V or a leftover 3300uf VY 16V ? Rest 1000uf caps on the PSU 5V/3.3V rail seem fine although I'm gonna replace them with some spare nichicon VY 16V. PSU is group regulated and I guess has passive pfc.

[momaka]

Originally Posted by zunasthegreat

From the following caps which one is the ideal : Nichicon PW 2200uf 16V / HD 2200uf 16V or a leftover 3300uf VY 16V ?

Either the Nichicon PW or the HD will do just fine. Given the name of the PSU - Nitrox 550W - it sounds to me like a cheap PSU, meaning probably a half-bridge topology (as that's all the Chinese really do for those.) So I'd probably put the PW on the 12V rail and the HD on the 5V or 3.3V rail... or better yet, see if you can squeeze both of these on the 12V rail - that will give it better filtering for a 12V-heavy modern PC.

As for Nichicon VY - that's a general purpose 85C -rated only, so not really suitable for SMPS unless you really have nothing else and need a temporary fix (though, on an old half-bridge design, that VY may still last a few years.) Thus, if you can, try not to use the VY and use HD or PW instead.

[zunasthegreat]

I forgot I had already done 3.3V filter caps with some Su'scon 10V 1000uf. The 5VSB seem fine atm, the PSU originally had "NICON" caps and a melted 4PIN CPU. I re-soldered an 8-pin EPS using AWG 18 instead of 20. Will polymers work for filtering ? Have a ton of polymers and really not great quantity of lytics. My HD/PW are not 2200uf, for the 12V line I can use ->

OCRZ polymer 16V 1000uf
Beryl polymer 16V 470uf
Nichicon (VY) 16V 3300uf
Nichicon (HD) 16V 1500uf
JAKEC 25V 2200uf

or keep Lelon 25V 2200uf

[zunasthegreat]

Jakec show lower ESR than Lelon but both are c[R]ap. PSU ate the 2200V 16V NICON cap within 4-5 years, system use to run a 750Ti then a 1050Ti and a stock AMD FX 8320. Can't be over 220-250W full load.

[momaka]

Originally Posted by zunasthegreat

I forgot I had already done 3.3V filter caps with some Su'scon 10V 1000uf.

Yeah, that's not going to do too well for filtering ripple. A lot of cheap crap PSUs come with just 2x 1000 uF on each rail (3.3V, 5V, and 12V), but that's *not* anywhere enough to filter ripple properly, especially because these PSUs are based on half-bridge design, which uses relatively low switching frequency and thus needs bigger caps on the output for proper filtering. Therefore, such PSUs need, again, *at least* 1x 1000 uF and 1x 2200 uF caps on each rail (so total of *at least* 3000 uF on each rail.) Any less, and the PSU will be spewing out abnormal ripple that then the hardware will have to cope with. This can lead to shortened lives of hard drives and motherboard components.

Originally Posted by zunasthegreat

Will polymers work for filtering ?

Short answer: No

Long answer:
Polymers will likely make the ripple output even worse. Too low of an ESR, especially on these older designs, is a really bad thing - the combination of low capacitance and low ESR will produce a ton of ringing noise on the output rails (ringing is noise created by the filter itself due to it being an LRC circuit, naturally.)

The only exception is the 5VSB - there you can probably use something like 1x 1000 uF and 1x 470 uF polymers or just 2x 1000 uF polymers. But even then, some PWM-FET controllers for the 5VSB (if the PSU uses one) may also not like the polymers too much or may not be compensated for it, so you may also get a terrible whining noise from the 5VSB too. So polymers and 5VSB: MAYBE OK. But polymers and the rest of the PSU: NO.

To keep it simple, just remember this about recapping PSUs: you can always go up in the capacity of the capacitors, but never lower. On cheap PSUs that use the 2x 1000 uF cap treatment, just completely ignore what they had and go with 2x 2200 uF for each rail. ESR-wise, caps should be entry-level to mid-range low ESR. Ultra-low ESR and polymers are not recommended and may not work (or even make the PSU oscillate wildly and blow.) On half-bridge designs, polymers should never be used due to low switching frequency (typically under 30 KHz.)

Originally Posted by zunasthegreat

My HD/PW are not 2200uf, for the 12V line I can use ->

OCRZ polymer 16V 1000uf
Beryl polymer 16V 470uf
Nichicon (VY) 16V 3300uf
Nichicon (HD) 16V 1500uf
JAKEC 25V 2200uf

or keep Lelon 25V 2200uf

Oh, I see you have Nichion HV series, not HD as you stated - those are actually very durable and dependable for PSU use. IIRC, a few models of the PS3 power supplies use them.

How many cap spots does the 12V rail have? One or two? If two, go with 2x 1500 uF Nichicon HV or 1x 1500 uF HV and 1x 3300 uF VY. Do the same for the 3.3V and 5V rails if they have 2 spots. And if any of these has only 1 cap spot, go with the 3300 uF VY. Yes, I said VY are GP and 85-C only, but I honestly trust them more than JAKEC. And your Lelon is RGA series, which is GP 105-C - i.e. same stuff as the Nichicon VY, but rated for higher temperature... which probably won't matter much, because I still think Nichicon VY is of higher quality than any Lelon.

Of course, if I had to pick between Lelon and JAKEC (or any other major no-name crap cap brand like JAKEC), then I'd surely pick Lelon.

But clearly you have some OK caps on your hands. Both the VY and HV should do.

By the way, your PSU above looks somewhat similar to the one I posted here:
https://www.badcaps.net/forum/showthread.php?t=90392

[zunasthegreat]

Thanks for the help, unfortunately PSU has just one spot for 12V cap. I saw the cap mark on VY and also checked their site and it seems VY is a 105 C rated cap. The 3300uf VY might not be a bad idea at all.

If sus'con 1000uf 10V are underrated or crap for 3.3V filter then what about KZG 1500uf 6.3V which I have in massive stock ? If 6.3V cap is OK I can even use Panasonic FL 6.3V 1500uf.

3.3V -> 2X 1500uf FL or KZG 6.3V
5V -> 2X Nichicon HV 1500uf 16V
12V -> 1X Nichicon VY 3300uf 16V

*Pretty sure even 100$ PSU doesn't use that kind of CAP anymore its just Teapo, ChongX, ASIA-X etc

Can't get any better with current market state, every order is delayed and stock is low.

[ChaosLegionnaire]

Originally Posted by zunasthegreat

then what about KZG 1500uf 6.3V which I have in massive stock ? If 6.3V cap is OK I can even use Panasonic FL 6.3V 1500uf.

neither of those are suitable because kzg is a very low esr series and FL is an ultra low esr series like FJ. kzgs are also a defective failure prone series unless their date codes are later than 2007 or 2008.

[RJARRRPCGP]

I've been lately suspecting 2004 and 2005 KZGs being the worst.