ECS P6STMT with OST and G-LUXONS caps

[trodas]

It died on me 7. 8. 2006
Symptoms - when I power it on, it power on, but quickly turn off. So some voltages or at least spikes of them are out of limit, I quess... After repair it working as my DL server again 24/7 stably and reliably since 7. 27. 2006.

I use this mobo as fanless server that run 24/7 and folding + downloading + P2P sharing machine. I invented it back in late 2003, as you can check there:
http://forums.silentpcreview.com/viewtopic.php?t=9704
And some more pictures of building this fanless computer are there:
http://ax2.old-cans.com/g.php?p=silentpc&d=1&v=v2
Yet after about 16 months my first moded fanless PSU died, so I made another design, much better, the heatsink is now hardly warm, etc. All in the galery linked before... Now I use better PSU and by checking it, the caps seems to be fine, no burnout, no smell, nothing...

Anyway, I inspected the mobo and - on the very first sight:

These G-Luxons are gone by...

The mobo use G-Luxons everywhere, except the Vcore caps:

Rest is, as I said, G-Luxons:

The overal mobo design aren't bad, tough it probably won't withstand todays 30 - 32°C in my room - running fanless...

Tough on second hand, since the died caps are so close to the PSU connector, then a suspiction is, that the PSU might be generating excess ripple there...!

Luckily, I got all the caps need for this one, Pannyes, ready:

ECS P6STMT
------------
6x 2200uF 6.3V d10 - P12344-ND
11x 1000uF 6.3V d8 - P12342-ND -> 1200uF 6.3V
7x 560uF 6.3V d6 - P12348-ND
6x 120uF 16V d6 - P12922-ND
5x 33uF 16V d4 - P12925-ND - 35V
1x 22uF 25V d4 - P12927-ND - 50V
10x 10uF 16V d4 - P11212-ND - FC, 25V

Caps in the PSU are brand "HEC" - hmmm, never heard about them, yet they are in the warning list of bad brands, damn. But in PSUs one mostly find such crappy ones, so... So I thought it could last a little bit longer or show more warning symptoms, but it did not... Eurocase PSU suxx...

So I recapped the board completely:

And yep, traditionaly I removed LPT and COM port (getting pretty skilled at it!) - the second blue "COM" port is VGA output - no removing planed, hehe:

And this is the picture of the mosfet I need to get!

...if I get there, I fill the empty spaces and we see how quality regulation we got then

I recapped even the 10uF caps, hehe. No, really, look:

Everything G-LUXON is gone, yet the board won't power anymore.

Now what went wrong? The 5V is shortcuted. I though - caps. Desoldered them, and checked them - and they aren't shortcuted. So, I soldered all the Pannyes and still - shortcut. Then I desoldered the top mosfet (the alone one on the second phase design, or what the hell this is...) and - 12, 32, 48, 62 ... all the way to around 300 ohms I getting as the caps charging. Correct now So, it is the damn mosfet that give up when the caps give up.

And the caps give up because they suxx and also because the PSU use HEC caps witch are giving up after year of running fanless 24/7 and therefore recapping the Eurocase moded fanless PSU is now in order.... Neverless this is another thred.

Now about the mosfet - the 15N03L type in TO-263 package (D2pak) is pretty old suxxka. It is a Infineon OptiMOS Power MOSFET, 30V, RDSon = 12.9mOhm, 42A. http://pdf1.alldatasheet.com/datasheet-pdf/view/79698/INFINEON/IPP15N03L.html
...and seems someone still get 'em: http://www.aeri.com/search/15N03L
But maybe it better to use the higher quality one?

After some discussion I got recommended the IRF3709ZS mosfet - http://www.irf.com/product-info/datasheets/data/irf3709z.pdf
Because it is also 30V, but 87A and mainly a very small resistance - 6.3 miliohms. Same gate charge as 15N03L Infineon - 17nF. It could be bought from Digi-key, but I get from BigPope the two orginal Infineons instead - it was out of stock at the time (at it looks like it is still out of stock, after all the time, lol), so...
But maybe someone get lucky and can obtain it. It should be a nice improvent for the board and it should generate notably less heat (almost half) in operation

Sucess!!!
A BIG thanks to Big Pope - in package from him come two 15N03L mosfets, thanks a lot! I replaced the upper mosfet and the mobo still did not kick in - it allways shut off immediatelly. So I checked the bottom mosfet better and it seems to be conductive real well (0.4 ohms, while the wires used to measure has 0.2 ohms itself) between it's collector and emitor, so I replaced it (thanks a lot Big Pope for the second one!) and the mobo then kicked in w/o a glitch


PS. the mainboard has a reserved place for another two mosfets. They will run in parallel to the existing ones (so it does not improve the Vcore design from one phase to two phases - even sharing one coil - it just halve the resistance and hence halve the resistance loss - but also double the switching loss - two mosfets instead of one...) and it is worth a considering if is not a good idea to make the mobo more robust to add them. It will probably not help overclocking as there is no Vcore adjustments in bios, sadly, but still worth considering. I would vote for the IRF3709ZS mosfets rather anyway

 

[Ripper3]

I really really like the project you did, making a fanless C3 Nehemiah server. Nicely done.
Shame about the problems you've been having though. But nice to see you recovered.

I'm not an expert with electronics, but if adding the extra MOSFET improves stability you might as well give it a try I say.

 

[tkpenalty]

Hmm Trodas, is it possible to buy solid aluminium capacitors? I want to recap my whole mobo with them since they are more compact.

 

[trodas]

Ripper3 - thanks! Well, I had to do it. I did not have any spare room, so I have to sleep with it, so... it has to be fanless and entierly unaudible. And I mean entierly - we live in very quiet place, next to woods...
And adding the two aditional mosfets that will run in parallel to these existing ones will make the CPU Vcore more robust (no question about it) but it also increase the switching loss, that is like 4W per mosfet... While it halve the about 2W current loss per mosfet so in the end I get like 3W more per one of the two added mosfets. Six more wats of heat I can't stand. I would much rather get two of the IRF3709ZS mosfets - but no idea yet where to buy them and frankly untill something happen - I will not exchange the mosfets, lol


tkpenalty - I don't know about solid ALUMINIUM capacitors, but the solid polymer caps can be bought (sometimes I getting few of them from Big Pope for special experiments) and can be used. Be warned about the China-made ones, tough.
http://hardware.slashdot.org/hardware/07/01/09/016259.shtml

I have seen them last less than a year. All you have to do is check where the capacitor is from. If it is from China (which is likely), then it has a high probability of failing very quickly. This is due to their stealing the formula from a Japanese company who became aware of the attempted theft and fed the women a recipe from the early 60's (and well known to hold up for only a year).

For most cases a good electrolyte cap will last a lifetime. OST and much less the G-Luxons and Hec caps are NOT good caps...