FSP400-60GLN worth fixing?

[Behemot]

Re: FSP400-60GLN worth fixing?

Yes, it is open-source design so they just make some cheap crap in a garage from the reference circuit, put firmware downloaded from the maker onto it and send it away. That's also why all of them look the same.

Some ppl who actually know a little bit about it make some changes, like with the Malish I have a sample of. Nobody made changes to the program itself which could easily greatly improve the device (like that you could select what component you want to measure!), AFAIK, though.

[beel1]

Re: FSP400-60GLN worth fixing?

Hello Behemot,
don't bother to unsolder your transformer, I think I found the culprit: Q61, it's neither open nor shorted that's why I didn't find earlier...

[Behemot]

Re: FSP400-60GLN worth fixing?

One of the switchers? Interesting. Better test it in series with incandescent lightbulb after you replace it.

[beel1]

Re: FSP400-60GLN worth fixing?

Yes, in my PSU Q60 and Q61 are Fairchild FQPF13N50C which cannot be found anymore.
Does anybody know what other MOS are mounted on this kind of PCB (>=400W configurations)? Or a good cross-reference suggestion?

[Behemot]

Re: FSP400-60GLN worth fixing?

Any 13N50 should do, or you can upgrade it to 18N50, quite common in other PSUs too so you can scavenge them from some dead half-decent units.

[beel1]

Re: FSP400-60GLN worth fixing?

Update: I put a pair of TK17A65W and tried with a 100W light bulb in series: after I turn on the PSU the light is very bright even with no load, maybe this is due to the PFC? Output voltages are OK, a small 12V fan is OK, but if I add a HDD the PSU turns off.
Actually I remembered that was the same behaviour before the PSU died the first time: it didn't start in a full system, but the PSU fan was OK and failed only with a HDD, until Q61 died.

Then I removed R8B to disable the PFC and now the PSU is OK with the HDD, there is only a quick slight glow of the light bulb when the HDD starts spinning and that's all. A power meter after the light bulb measured 220V (230V mains), PF=0.65, 17W (about 110mA).

So now I have to try with bigger loads and find what's wrong with the PFC.

[Behemot]

Re: FSP400-60GLN worth fixing?

It should only blink when the cap is charging, than dim. I think that fan or HDD makes almost no difference, it should still work.

I'd guess there is still some problem which made the switchers go bad before…

[beel1]

Re: FSP400-60GLN worth fixing?

Well actually the PFC was doing its job
In my PSU the target boosted VB set is 380V and the CM6800 switches OFF with VB below 180V.
Even with very light load (the PSU fan for instance) the PFC tries to boost VB drawing too much current with the light bulb in series. So mains input voltage drops, VB decreases down to 180V and PSU switches OFF. I think the same happens with the broken Q61.
Without any secondary load, VB reached 380V slowly (after several seconds), then when light was OFF I added loads and it was OK, the light remained OFF meaning there are no shorts.
So after checking the PFC section driving signal I made tests without the light bulb and it ran perfectly, with better efficiency than an Antec SmartPower 300W: so to me, my PSU is fixed. I used it with a complete system loading the PSU up to 230W without problems.

So in the process I've replaced the switchers pair and 2 damaged 1000uF/16V caps. I've also unsoldered/resoldered almost every components to check their values.

My guess is the switcher failed because of its fullpack package which cannot dissipate much power. Together with the ageing thermal paste and hot room temperature it has broken up.

[Behemot]

Re: FSP400-60GLN worth fixing?

Better do complete recap, every single cap in it is PoS and will fail, it is only a matter of time. I have seen units where ALL or 90+ % of caps were bloated. Wonder how it managed to run for so long so all the caps failed, it often takes just a couple of them to go bad so the unit does not start anymore. Maybe it was under continuous operation…

If you need caps, I already have stock of 1200/16 and 2200/6.3 Rubycons in D8 which are especially handy for these FSP units where they use custom D8 craps with values none of the quality brands offered until recently.

[Stefan Payne]

Re: FSP400-60GLN worth fixing?

Better do complete recap, every single cap in it is PoS and will fail, it is only a matter of time. I have seen units where ALL or 90+ % of caps were bloated. Wonder how it managed to run for so long so all the caps failed

Sometimes a bloated cap can still measure inside spec for whatever reason...

[Elitist]

Re: FSP400-60GLN worth fixing?

Not only do some bulging caps continue working, presumably temporarily, but when removed, the vents can kick in and the bulging sometimes disappears. Since gaseous components must've been lost, undesirable that the now flat-again units should be re-deployed as their lifetime would be short and possibly spectacular. Notwithstanding, only ever seen/heard two of thousands expire in pyrotechnics.
Furthermore, I had stored a small number of new, expensive Rubycons for future for several years but they failed as soon as deployed. The chemistry of electrolytics suggests that, apart from tantalums, all liquid types best not be stored unconnected. A small continuous charge would be preferable, but not easily or readily applied in most practical situations. The smaller, high capacity, thinner film types probably most vulnerable - the film is dynamic!

[Behemot]

Re: FSP400-60GLN worth fixing?

I currently store over 50000 pcs of quality capacitors and nothing fails on me.

Most likely bad device, putting insane stress on the caps.

[Wester547]

Re: FSP400-60GLN worth fixing?

Not only do some bulging caps continue working, presumably temporarily, but when removed, the vents can kick in and the bulging sometimes disappears. Since gaseous components must've been lost, undesirable that the now flat-again units should be re-deployed as their lifetime would be short and possibly spectacular. Notwithstanding, only ever seen/heard two of thousands expire in pyrotechnics.
Furthermore, I had stored a small number of new, expensive Rubycons for future for several years but they failed as soon as deployed. The chemistry of electrolytics suggests that, apart from tantalums, all liquid types best not be stored unconnected. A small continuous charge would be preferable, but not easily or readily applied in most practical situations. The smaller, high capacity, thinner film types probably most vulnerable - the film is dynamic!

Ick, that doesn’t sound good at all. What were the series and datecodes of the Rubycons that failed, and the capacitance / voltage values? Maybe they needed to be reformed properly, or they were sleeved backwards. Leaving capacitors in storage isn’t ideal, but this is the first I’ve heard of genuine? Rubycons failing the second bias is applied to the plates. What output (or input?) voltage were they filtering?

[Elitist]

Re: FSP400-60GLN worth fixing?

"insane stress" - nope, 1-4-1 m/b replacement, mainly 2200/6.3 some 3300/6.3 - long since disposed of, so now in the Al remelt pot. The film on such is very thin, guessing, ideally best stored under minimal bias. No such thing as 100% coherency. Then there's the electrolyte...

[Wester547]

Re: FSP400-60GLN worth fixing?

Did you try to install them “backwards” on a motherboard whose polarity markings were backwards (the + sign on the shaded half of the polarity circle)? Of course something like that would be obvious, but sometimes even the best techs may make simple mistakes such as that. It could certainly explain why they suddenly blew.

[Elitist]

Re: FSP400-60GLN worth fixing?

Extremely unlikely as I usually replace one at a time into the same configuration. Bit puzzled why there's such doubt about storage deterioration? Makes perfect sense to me! They didn't blow, they failed fairly quietly if I remember and replacement with new stock fixed the problem.

[Wester547]

Re: FSP400-60GLN worth fixing?

So you’re saying they didn’t bulge or leak but that they silently died? I don’t doubt that the oxide film deteriorates in storage and must be restored using the proper precautions (applying the rated voltage every 18-24 months at least).

[Elitist]

Re: FSP400-60GLN worth fixing?

they silently died

- yup, in storage. I believe the recommended storage protocol is to ramp up to rated every so often (but nobody does this except, perhaps, a few manufacturers with excess stock). A lower level continuous 'soak' also preserves the film without risking the organics in the electrolyte, aqueous solvent or not. Again, rarely done in practice. No wonder the trade is transferring to tantalum, but (i) not much option for mobiles (ii) tantalum is rare, very expensive/difficult to extract & refine (iii) Coltan mineral source in DRC and slave labour raises ethical concerns.

[Behemot]

Re: FSP400-60GLN worth fixing?

Sir, I do not know where you live, but tantalum is becoming dead (in mainstream electronics anyway) is it can not cope with MLCC in terms of capacity density, price and reliability. Plus ceramics usuall do not tend to catch on fire

As for recommended storage, no datasheet talks about anything like that. Either the caps were way too old (remember the ultra-low ESR caps are way more unstable and less reliable than other types, the have short lifetime even new), got frozen during transport (or similarly damaged) or they were fakes.

I do store about 1500 Rubycon MBZ and MFZ (with 500 already in use - aprox.) and none of them failed yet. Same with couple hundred Pannies or Samxons GC. So the statistics is clear - you are single exception, or there was problem with the caps even before you got them. Interesting how you jump to conclusions and creating wild theories after half a dozen bad caps.

[momaka]

Re: FSP400-60GLN worth fixing?

It is true that the ultra low ESR electrolytic caps are more unstable and some may develop problems after a good few years in storage... but overall, I too have rarely seen failure. Like Behemot, I have a stock of about 200 or so total Rubycon MFZ and MCZ as well as Nichicon HN and HZ (all pulled from Xbox 360 consoles - i.e. used caps). I just opened my glass jars of them the other day as I needed a few, and noted that I had 2x Nichicon HZ 6.3V/2200 uF that had bulged by themselves. Upon closer inspection, both of these turned out to have a date code of H05xx, which is known to have some problems (though not as bad as H01-H04 datecodes). Thus, I re-tested all my H05 HZ stock and found 2 more caps with border-line in spec high capacity (2700 uF) and one out of spec (3800 uF). This indicates some of these H05 HZs have started to break down. What's interesting is that I reformed and checked all of them roughly 2 years ago, and none had a problem. Some of them, I stored with full plate charge, while others I stored discharged in order to see if that would make a difference. But it didn't - one of the failed HZs was stored with a charge and the other was not. Same for the HZs that read high capacity.

But other than those Nichicon HZ caps with H05 datecodes, all my other ultra-low ESR caps read in spec, including all H06 and later Nichicon HZ caps. Now, I did notice that from these caps, most of the 16V caps that I stored with a full plate charge (15.5V) still retained some of it (ranging from 5 to 14V). Meanwhile, none of the 6.3V caps that I stored with a full plate charge (of 6V) retained anything - most read between -0.5 to 0V. Yes, you read that right, some of the 6.3V caps had negative charge on their plates from sitting in storage.

This indeed indicates that lower-voltage higher capacity caps are indeed probably more prone to going bad in storage due to thinner oxide layer (or at least aging faster than their 10V and 16V counterparts). But again, after a quick reform on some of these caps, I had no problem using them again. And as I noted above, except for the problematic Nichicon HZ with H05 datecode, all other caps read in spec.