Intel D945GCLF Bloated Samxon GC due to overheating

[RJARRRPCGP]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

those boards started dieing at the 1 year mark in some industrial pcs i service.
and there is a chiller on the cabinet.

And it's not even a Core i5! Wow! They shouldn't even have issues of caps being stressed, much less a cap failure within just 1 year.

[kc8adu]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

those boards started dieing at the 1 year mark in some industrial pcs i service.
and there is a chiller on the cabinet.
the few that had rubycons from the factory are still ok.
i preemtively fixed all the others and half were bulging at 1 year.
were still running but were not stable.
this run was 400+ units and they had to be pulled and mailed to me.
lets just say this client will never spec an intel board again for these builds.
the first 6 he bought in the developement phase all had rubycon.i greenlighted them figuring they should be ok for the 5 year warrenty on the product.
the samxons showed up in later shipments.intel blew them off when we noticed failures.
we reserve the right to substitute components of equal quality.
nice!

[teigaff]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

How does the caps in the powersupply look and maybe you should look at improving the airflow aswell?

I received this board from a friend of friend from him without power supply. When the board will be alive again, I'll look for a power supply.

So I can take a 220uF 16v 6mm ZLH Series as replacement for my 6,3V, 220uF. Right?

[ben7]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

I didn't see other components near them. That sounds like a poor design.

Samxon capacitors ... so-so quality

[Per Hansson]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

I choose the caps I did based on what was available. (It's always fine to increase voltage).
As you can see in my first post the 1000uF 10v caps only see ~1.5v so reducing voltage in this case is fine aswell...

Interesting that your 3300uF 6.3v Samxon GD cap in the corner has failed too, I have not replaced that on my board...

Yea probably if that capacitor has failed too you should replace all caps, because I think then your system has been running too hot...
How does the caps in the powersupply look and maybe you should look at improving the airflow aswell?

[teigaff]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

I've the same board and it won't power up. The two big capacitors 10V, 1000uF are bulged. And one in the top left corner, it's a 6,3V, 3300uF, 10mm x 25mm.
Because I don't know exactly which caps else are defective, I want to replace all. For the most I found replacements. They are totally 17 pieces:
2 x 10V, 1000uF, 10mm, 15mm
1 x 6.3V, 3300uF, 10mm, 25mm
1 x 6.3V, 820uF, 8mm, 24mm
5 x 10V, 470uF, 8mm, 12mm
1 x 25V, 33uF, 5mm, 11mm
2 x 6.3V, 220uF, 6mm, 12mm
5 x 25V, 100uF, 6mm, 12mm

Why you chosen: 6.3V, 1000uF and 16V, 220uF? That's crossed. Is that possible?
I didn't find the: 6,3V, 220uF on the online store. Which should I choose?

Thanks for the explanation.

[lti]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

I didn't see other components near them. That sounds like a poor design.

[Per Hansson]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

As I wrote in the first post the system has seen 24/7 usage for about 3 years now...
With capacitors rated for 105°C with a lifetime of 2000h
The temperature on the PCB where the capacitors are is 65°C (Note the two mosfets besides them...)

The general rule of thumb says that for every reduction of 10°C in temp the capacitors life doubles, so that turns out to be 32000 hours / 24 = 1333 days / 365 = 3.6 years
That's close enough since it is now 20°C inside, but yesterday it was 30°C in here

[lti]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

Those caps don't look like they were close enough to the heatsink or exposed to heat for a long enough time to have failed due to heat.
How old is this board?

[Per Hansson]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

Th3_uN1Qu3; Yea I know, sad really
I actually contemplated to buy a new ITX mobo based around the Intel NM10 chipset
However it would be around $70 and I would need to do allot of research to know that my dual PCI riser with the 64bit Intel NIC would work...

Much easier to just change some caps
Actually the reason I noticed the failure was that the WLAN card stopped working, bunch of warnings about resets in the /var/log/messages logfile...

wifi0: hardware error; resetting
wifi0: stuck beacon; resetting (bmiss count 11)

370forlife; I've considered it but it's a desktop system so it would kind of defeat the desktop part (can't put anything ontop of it)
In reality I've just been to lazy to haul the chassi off to work and just LASER cut it

[Th3_uN1Qu3]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

Now Intel generally gets allot of credit for their designs, but I dunno on this one.
First the choice of this ancient chipset using 12 times as much power as the CPU.

In case you didn't know, all the netbooks are the same too. I think the mobile 945 draws something like 6W, but that's still 3x the consumption of the Atom CPU. Talk about crappy design.

[370forlife]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

To make it quieter why not just put one large 120mm fan over everything and run it slow?

[Per Hansson]

Bought myself a complete ITX system some years ago on eBay, sadly the Socket A Jetway mobo in it died so I replaced it with the cheapest low power ITX board I could find at the time, the Intel D945GCLF mainboard... (Almost 3 years ago now)

It is a Intel Atom 230 single core 1.6Ghz "beast" which I run IPCOP on.
You might know that the original fan is really crappy and also not thermally controlled.
I replaced it with a big copper heatsink hoping I could cool the fantastic Intel 945GC chipset passively.
Well, about 30mins later the heatsink was over 100°C hot and the system crashed, so I put a fan ontop of it anyway
Now I want this system to be silent since it is in my bedroom and runs 24/7 so it spins really slowly, still keeps the heatsink quite cool, temp sensor reads ca 40°C

The CPU on the other hand is passively cooled with a tiny heatsink, it only becomes 50°C
No wonder when it's rated at just 2W TDP, unlike the chipset which is rated for 22W TDP!

Now Intel generally gets allot of credit for their designs, but I dunno on this one.
First the choice of this ancient chipset using 12 times as much power as the CPU.
Then ontop of that they put 4x electrolytic caps close to it, but the CPU which needs just 2W gets a nice Polymer capacitor? Meh...

I contacted Big Pope and ordered some Samxon solid polymer caps from him
Find below a list of the original caps, they where a custom order by Intel, the sizes do not exist in the Samxon catalog...

2x Samxon GC 1000uF 10v 10x12.5mm
2x Samxon GD 220uF 6.3v 6x12mm

Replacement capacitors:

2x Samxon X-CON ULR 1000uF 6.3V 8x11.5mm
2x Samxon X-CON ULR 220uF 16V D6.3x11mm

For information here are voltage readings under the capacitors, incase you have a similar board and would like to change any of the caps to another value. Note that the 220uF caps fit well however I had to bend the legs on the 1000uF because the leg spacing was too narrow...
(Look at the photos to understand which cap is which)

Front left 1000uF: 1.11v
Front right 1000uF: 1.57v
Rear left 220uF: 1.57v
Rear right 220uF: 2.54v

The lone original polymer by the CPU sees 1.16v
The little DC/DC PSU which can be seen in the upper left corner was already recapped by me with Samxons about 2 years ago because the orignal crap caps failed