Hmm... are you using primary-side ground to measure these?
Primary-side ground is the negative side on one of the two big 200V caps that are connected in series. To know which one it is, find the cap whose positive lead is connected to the negative lead of the other big cap. The negative lead of this cap (whose positive is connected to the other's negative lead) is primary-side ground.
Yes, absolutely.
The easiest way is to measure the resistance between each rail output and ground. Just take into consideration that sometimes older PSUs like this have large...
momaka
-
I don't think the caps should be the main problem here. The PSU appears to use a forward converter design and the output toroid is Micrometals -52 type, which suggests it should be using a little higher switching frequency compared to oldschool half-bridge stuff... so the lower ESR of the caps should be OK. But maybe not?!
So let's start with the basics here. First, did you test the PSU with a proper load? If you have an old motherboard that draws more from the 12V rail, try that first. Alternatively, 12V incandescent light bulbs / auto bulbs would also make a good load - preferably...Last edited by momaka; 12-26-2025, 11:03 AM.Leave a comment:
-
Yup, Panny FR's should be just fine for any Pentium 3 board.
The Fujitsu cap is likely from their function polymer (FP) family, and unlikely to be bad (I've seen a few of these fail, but only when heavily abused from heat and/or physically.)
DON'T replace the small SMD electrolytics with Tantalums! A few of these are likely for the audio circuits, and that's once place where Tantalums don't belong.
If it was my board, I probably wouldn't bother to replace any of these smaller (100 uF and under) caps....Leave a comment:
-
Yup... bad MCP it is.... well, probably. :p
Since the issue seems to come and go when messing with the RAM, it could also be bad BGA under the CPU socket. Some 939/AM2/AM3 coolers are pretty bad about warping the board, and I have seen a few boards loose a RAM slot or too due to this. But if that was the case with your board, I imagine it would either full-not-POST or POST and work fine. The fact that there are graphics glitches, however, pretty positively sways this towards bad MCP.
I suppose before you part it out, give the MCP a good re-heating with hot air, just for...Leave a comment:
-
The board has solid polymers (and Tantalums for the CPU VRM output), so very unlikely for these to be the problem. If anything, Tantalums almost always fry themselves violently if something is off about them. Since that isn't or hasn't happened on your board, I think we can rule these out.
MOSFETs or drivers? - It's a pretty high-end board with a complex multi-phase CPU VRM... so again, if any MOSFET was failed or bad, I'd imagine it would get fried pretty quickly by the other MOSFET phases (at least on the CPU VRM). Of course, do a simple thermal check (OK to do with a finger) and...Leave a comment:
-
If this was my board, I probably wouldn't waste too much time/money/effort with it. Reason why: the caps that you have on your board are mostly OK (except possibly for the Nichicon HM's and UCC KZG's... but I'll elaborate on that below) and it's an nForce 4 chipset, which aren't great when it comes to reliability. There's actually a good chance the caps might outlast your chipset. So again, it may not be worth it to go all poly or anything like that.
In regards to the caps:
- For the Nichicon HM caps, check their date (production) code. If it's anything that starts with H01,...Leave a comment:
-
You can leave the big 200V caps alone, as in, not buy new ones. The old ones measure fine and won't give you issues. On non-APFC PSUs, those large HV caps almost never fail. So save yourself the money and hassle with trying to replace those. From your list of the caps that you posted above, the one 3300 uF (C27) cap that measures 307.4 uF is definitely bad... unless that was a typo as well?? Is that all of the caps in the PSU? I suspect there should be lots more small ones... and usually these are the ones to fail "silently" (ironic, given the PSU's symptoms, isn't it
) without bulging...
Leave a comment:
-
I doubt those "heat mats" used for phones and tablets can get hot enough to assist you enough with replacing the socket.... though I haven't used one (or even seen one) so I don't know how hot they get.
In any case, you will need a lot more heat, probably.
If you have an electric (but NOT induction type) or gas stove / cooktop / burner (almost every household does), you can use that as your bottom heater. It's not a neat solution, but it does work and I have used it many times to remove large CPU sockets even without the need for hot air, and the boards were...Last edited by momaka; 07-30-2025, 01:00 PM.Leave a comment:
-
I 2nd what lotas mentioned in regards to the buttons - sometimes they start to partially short over time, and the symptoms could be completely intermittent too. I've had a few monitors suffer due to this and randomly pop up menus or change their brightness and etc. In some cases, it's temperature-dependent and nothing happens until the monitor heats up. Nearly pulled out all my hair with a 24" Dell some many years ago due to this. Just because the buttons appear to read OK when checked once with a multimeter does not mean they are OK. The best but probably most painful way to deal with this...Leave a comment:
-
Just to shed some light on the original issue of the PSU here...
If you have a PSU that turns On and appears to output the proper voltages on *all* of the output rails, but the connected motherboard does NOT turn On, then the issue is likely with the PG signal - either it's not there or not coming on at the correct time.
But now with the shorted MOSFET(s), that's a completely different issue and probably unrelated to what was going on originally. Just putting this here for anyone else who reads this thread so they can keep track of what's going on.Leave a comment:
-
I have not tried it before and only read a little bit on the matter... but I think the resolution capability is just programmed in the firmware. Now, as for weather the specific Genesis chip in the other, lower-end 17" or 19" monitors has the bandwidth capability to be pushed at higher resolutions, that I do not know. In any case, if you do take the logic board from one of the lower-sized monitors, you would need to make sure the LVDS signals going to the t-con board match the layout to that of your 2007FPb first... and then, perhaps by using a firmware dump from the 2007FPb or just...Leave a comment:
-
Interesting fix.
Just a curious question about that capacitor circled in red above - is that the original or is that the replacement?
From your post, I'm understanding it's the former. If so, that looks like a polymer capacitor and I'm curious if you actually checked it to see if/how it failed. Would be good to know if that really was the issue or if the heat in the area caused something else to temporarily get "fixed".Leave a comment:
-
The "HC" marks indicate that these are Nichicon HC series - a low impedance type of capacitor. So you need to get the same type - i.e. match the impedance/ESR and ripple current to generally within +/-30%. The brands and series I listed in my previous post should all be close enough. So just look for, for example, Rubycon ZLH or ZLQ on Digikey, and you should be able to find good-matching replacements....Leave a comment:
-
Yes, I have found several sellers on ebay that are the same way - most of the caps they sell are NOS (5-10 years old, and a few even older). Have not had any issues with using those.
I have nothing against buying from Mouser and Digikey. The only issue is, both of these tend to be NA-centric (US / Canada.) So if you live elsewhere in the world, options can indeed get more limited. And in that case, it's good to know that there genuine Japanese caps too floating around on platforms like ebay and Aliexpress. One just has to be more careful to avoid the counterfeits. But other...Leave a comment:
-
Their electrolyte (Rubycon MFZ) starts to attack the foil / plates, making the cap unable to withstand the voltage that it is rated for. The leakage current also greatly increases. In some cases, this can cause the cap to short-circuit internally and then rupture. Or, if the cap sits unused for a long time, then the electrolyte will still consume the foil/plates, producing gas inside the can and make the cap vent eventually.
In either case, you can observe this on the cheap ESR meters when the capacity of the cap starts going close or above its tolerance - e.g. the 2700 uF Rubycon MFZ will...Leave a comment:
-
socketa,
I must say, I'm really impressed by your dedication, both in terms of troubleshooting and keeping your threads updated
So it looks like it's boiling down to a bad solder joint, possibly.
I don't think (or at least I would hope so) that the nForce 2 Southbrdige chip is the issue here. It's a wirebonded chip, and these won't (or should not) have the same issues like flip-chip technology chips can... unless the chip was somehow severely overheated and for a long time. So I think it's either a BGA under the Southbridge... *OR* could also be a bad connection...Leave a comment:
-
@ Kabir85: I think those are legitimate.... or at least I have not seen fake/counterfeit ones that can copy the original Nichicon rubber bung.
The datecode also looks real. Looks like these were produced in week 23 of year 2020. The letter in front indicated the COO (Country of Origin - i.e. where the cap was made.) I forgot what all of them are, but I think "C" is for China and B is for Malaysia (can't find the datasheet where I previously found this info, so if someone does, please correct me if I'm wrong.) The one I know for sure is "H" is for Japan.
In...Leave a comment:
-
The cap that scottl31 posted above is Nichicon HC series, as lti correctly identified.
EB series from Panasonic is not quite adequate in that case, if ESR and ripple current capability are to be observed.
Relatively close replacements would be United Chemicon KZE and KZH or Rubycon ZLH or ZLQ. All of these are readily available on both Mouser and Dikigey ATM. Just match the capacitance and the voltage (10V, 470 uF), along with the diameter. Height-wise, UCC KZH and Rubycon ZLH and ZLQ will likely both be shorter, since they are newer miniaturized series.
*edit*...Last edited by momaka; 07-02-2025, 06:37 AM.Leave a comment:
-
I don't think the chipset makes that much of a difference (it at all) on newer systems like that. But the RAM type and speed definitely does... and also, whether you have regular desktop type or Registered / Reg. + ECC. Any type of registered RAM tends to add quite A BIT of latency... so for gaming -type of workloads, these systems tend to do rather poorly.
I'm not sure where compiling stands with that, though I suspect it's more like video encoding/decoding - CPU is just number-crunching through a bunch of data, so possibly the high RAM latency is not that big of a deal.
[/quote]...Last edited by momaka; 06-21-2025, 08:15 AM.Leave a comment:
-
I don't see how storing a capacitor charged can damage its dielectric and I've done it as a test on quite a number of caps I have reformed too. What's interesting is I didn't find much difference between the caps that were stored fully charged (after reforming) vs. the ones that were not stored with charge. Only the ones that weren't reformed and stored without a charge seemed to go bad quicker - in my case, I was (re)using Rubycon MFZ 6.3V 2700 uF caps pulled from scrap Xbox 360 motherboards. As I have found over the years, all of these are unstable and bound to go bad, much like UCC KZG 6.3V...Leave a comment:
No activity results to display
Leave a comment: