Intel D945GCLF Bloated Samxon GC due to overheating

[Wester547]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

I should add that the 820uF/6.3V MCZ pulled from the near the DIMM slot looks fake - the bottom bung is the one with concentric circles, which I've never seen on an authentic Rubycon on ammo tape or bag before; and the ESR measures about 40 mohm, which seems to be a bit high, but that's after 6 years of intermittent use. OTOH, the 3300uF/6.3V Samxon GD seems to be within spec at 10 mohm.

0.04 ohms isn't bad considering MCZ's sensitivity to heat and relatively high H2O content in the electrolytic mixture. Also, the rubber seals do degrade with age, and that too causes drying up. 0.01 ohms is pretty good, but another important question is how high the capacitance measures, because if a capacitor is going leaky or is partially shorted, it could measure in spec for ESR but would yield a capacitance reading much higher than the 20% tolerance allows for (early HMs and HNs failed exactly in this manner).

Did Intel China knowingly use fake Rubycons, or did they also get conned by a flaky vendor? Or alternatively, did Rubycon outsource production of (some) MCZs to a volume OEM producer in China?

This one is harder to say. The Rubycon MCZ probably looks completely genuine if you put aside the cheapo bullseye bung? This is another reason that I'm not convinced Intel/Foxconn/Flextronics (sometimes Intel used ASUS or Pegatron as their OEM as well) sourced their components directly from the factory - they likely rely upon previously trustworthy distributors who could eventually foul up with counterfeits slipping into the supply chain amidst the genuine capacitors.

Rubycon is the most expensive brand for a reason. They have no factories in China and they important everything from Japan, so their electrolytics are pretty much built to the highest order. Either we are looking at potential counterfeits or Rubycon were specified not to use their thicker bung design for some reason.

[linuxguru]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

This one is harder to say. The Rubycon MCZ probably looks completely genuine if you put aside the cheapo bullseye bung?

Yup, so good that I didn't even bother looking at it closely until a day ago.

Rubycon is the most expensive brand for a reason. They have no factories in China and they import everything from Japan, so their electrolytics are pretty much built to the highest order. Either we are looking at potential counterfeits or Rubycon were specified not to use their thicker bung design for some reason.

Looks like counterfeits in the supply chain, and the OEM chose to look the other way - maybe with some malfeasance/collusion among incoming inspection/qualification teams at the OEM. Maybe they measured a few, and it met specs, so they went ahead and used it despite the dodgy provenance.

[linuxguru]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

FWIW, I just got my hands on an earlier Mini-ITX board from Intel, the D201GLYL, which uses a Yonah Celeron-M 215 in a BGA479 package (similar to the ones used in some early Celeron-M notebooks). It's 32-bit, single-core, no HT and runs at 1.33 GHz. On the plus side, it has an out-of-order execution engine derived from the Tualatin PIII, reasonably large 512 KB L2 cache, and the P4 FSB clocked at 533 MHz.

The fan is installed on the Celeron-M CPU, while the SiS Northbridge is passively cooled. All the caps are Japanese - 2x 820uF/6.3 Rubycon MCZ for CPU Vcore, and 2x 3300 uF/6.3 Nichicon HM next to the DIMM slot (all stock). The board dates to 2007, and there's nothing failed or bulging, including the MCZs (they don't have hot air from the CPU HSF blowing on them, by virtue of being on the side of the HSF without the vanes). The board POSTs and runs Memtest86+ without hiccups at the rated DDR2 speed of 533 MHz. In some ways, it is better-built for embedded appliances than the Atom-based D945GCLF, but there are some downsides, including poor Video support, no SATA and single on-board NIC.

All in all, not bad at all for an 8-year old board for which I paid ~$5.

Edit: The 2x 820uF/6.3 MCZ are gold-on-black, 8 x 25 mm (unlike the white-on-black, bull's-eye bung 820uF/6.3 MCZ that I pulled from the D945GCLF). I haven't pulled the 2x MCZ from this board yet, so I can't tell if they have a bull's-eye or the traditional thick Rubycon bung. The lettering is also of sharper and higher finish.

[mockingbird]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

Edit: The 2x 820uF/6.3 MCZ are gold-on-black, 8 x 25 mm (unlike the white-on-black, bull's-eye bung 820uF/6.3 MCZ that I pulled from the D945GCLF). I haven't pulled the 2x MCZ from this board yet, so I can't tell if they have a bull's-eye or the traditional thick Rubycon bung. The lettering is also of sharper and higher finish.

So these are 2007-era MCZ caps? Curious, what is the factory code on it?

Also, can you post pics of the MCZ caps off the intel board with the concentric bunges?

I'm beginning to think that one should either use polymers to replace MCZ-class caps, or only Panasonic FL (impossible to obtain new AFAIK) and Nichicon HN/HZ. Late Rubycon UESR caps don't seem to be consistently reliable and Samxon GC/GA quality is dependent on the batch.

[linuxguru]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

So these are 2007-era MCZ caps? Curious, what is the factory code on it?

Yup, the code is T0716 on the gold/black ones. The later white/black bull's-eye one is T0915.

Also, can you post pics of the MCZ caps off the intel board with the concentric bunges?

Will do so presently, but briefly - there are some obvious, and some subtle differences:

1) Colour scheme on sleeve - gold/black vs. white/black.
2) Concentric ring ones have PET sleeve (legend PET below 105 c)
3) Position of MCZ text differs - begins below 3v and ends below 20 on 2007 MCZ, while it begins below v and ends just past 0 on 2009 MCZ. (This may not be relevant, since there has been a sleeve change to PET, at which time the legend may have been altered by accident).
4) Can dimensions/material appear to be identical - slightly shiny aluminium with narrow vent scoring, as compared to the more matt aluminium with wider vent scoring on most of the Axxxx-code Rubycons that I have (again, may not be conclusive - some Txxxx-code Rubycons also have the matt-finish aluminium cans).

I would venture the guess that the 2009 bull's-eye MCZs are authentic, but special-order OEM caps for Intel (China) with some cost-reduction (which coincided with the use of Samxon on the D945GCLF) due to competitive pressures. Note that the MCZ hasn't actually bulged or failed in 6 years, even for the OP (Per Hansson).

I'm beginning to think that one should either use polymers to replace MCZ-class caps, or only Panasonic FL (impossible to obtain new AFAIK) and Nichicon HN/HZ. Late Rubycon UESR caps don't seem to be consistently reliable and Samxon GC/GA quality is dependent on the batch.

Other more reliable (but not UESR) options for 8mm are FJ and ZLH, of which the later may be unobtainium now. Perhaps KZH, but not sure if it's available in 8mm.

[Per Hansson]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

I never replaced the original Samxon GD 3300uF cap or the Rubycon MCZ 820uF cap next to it by the memory and it's still working fine like this over 7 years later.
System is still in use 24/7 as my firewall with only a tiny silent fan on the chipset heatsink.

I should add that the 820uF/6.3V MCZ pulled from the near the DIMM slot looks fake - the bottom bung is the one with concentric circles, which I've never seen on an authentic Rubycon on ammo tape or bag before; and the ESR measures about 40 mohm, which seems to be a bit high, but that's after 6 years of intermittent use. OTOH, the 3300uF/6.3V Samxon GD seems to be within spec at 10 mohm.

Did Intel China knowingly use fake Rubycons, or did they also get conned by a flaky vendor? Or alternatively, did Rubycon outsource production of (some) MCZs to a volume OEM producer in China?

Well, curiosity killed the cat!
I just could not have this board sitting on my desk infront of me each day without checking how the remaining caps looked
And it's just like linuxguru says, the 6.3v 820uF Rubycon MCZ at the edge of the board by the memory has a cheap bung with concentric circles on mine too!
And the 6.3v 3300uF Samxon GD cap next to it looks weird too, the bung looks collapsed inwards, never seen anything like it before!
My guess is that these are fakes that entered the supply chain, probably Intel was shooting for a very low price point so allot of corners had to be cut.

I replaced the Rubycon with a Samxon ULR 2.5v 820uF polymer. (It just sees VDIMM so 1.9v so it's fine).
The Samxon cap sees 5v though and I did not have a suitable polymer in stock so I took one of my ancient Rubycon MBZ's and reformed it and then soldered it in the Samxon's place.

I also replaced the 5x 10v 470uF Nichicon VR 85°C caps with Panasonic NHG 16v 470uF 105°C caps, the one by the southbridge was starting to go bad due to the heat.
I also replaced the 2x 25v 22uF Samxon GS 85°C caps with UCC KMG 50v 22uF 105°C caps, the one by the onboard NIC had been well baked and was significantly out of spec!

Finally for good measure I replaced the 5x 25v 100uF Samxon GS 85°C caps by the audio controller with UCC KMG 16v 100uF caps, they all measured perfect though.
Since I don't use the onboard audio and the caps are not in a hotzone they survived, I measured 2.5v on four of the caps and 5v on one.
Obviously the voltage will go up with increasing volume but since I don't care about the onboard audio it will be fine like this anyway

[Wester547]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

My guess is that these are fakes that entered the supply chain, probably Intel was shooting for a very low price point so allot of corners had to be cut.

If they are counterfeit, I otherwise give them a standing ovation. Those Rubycon cans look incredibly real. I'm beginning to wonder if certain manufacturers were too cheap to buy Rubycons with a thicker bung so they specified a cheaper, flat bung (though why Rubycon would use that el cheapo concentric bung is well beyond me - the residue on it is proof enough that it causes more problems than not). That Rubycon is a custom order, which tend to be more expensive, so that could be part of it. Rubycons tend to be comparatively expensive anyway. It also looks like the bottom of the sleeve sustained some damage.

And the 6.3v 3300uF Samxon GD cap next to it looks weird too, the bung looks collapsed inwards, never seen anything like it before!

Seems to me that they just didn't use as thick a stand-off for the rubber so the bung has a flatter appearance.

[Per Hansson]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

I'm beginning to wonder if certain manufacturers were too cheap to buy Rubycons with a thicker bung so they specified a cheaper, flat bung (though why Rubycon would use that el cheapo concentric bung is well beyond me - the residue on it is proof enough that it causes more problems than not).

Just seems strange that Rubycon would agree to that, it's their reputation on the line after all...

That Rubycon is a custom order, which tend to be more expensive, so that could be part of it. Rubycons tend to be comparatively expensive anyway.

Which begs the question: why put a Rubycon on a board if you are cost constrained?
Why not one of the cheaper first tier brands or even second tier brands...

It also looks like the bottom of the sleeve sustained some damage.

That was me during desoldering, as linuxguru mentioned it was a quite difficult cap to desolder...

Seems to me that they just didn't use as thick a stand-off for the rubber so the bung has a flatter appearance.

Yea maybe, I attached another photo with the old caps standing next to two new unused caps.
There it's more clear to see how much the bung is collapsed inwards on the Samxon.
It's also striking how similar the print is on the old Samxon and Rubycon caps!

[Wester547]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

Just seems strange that Rubycon would agree to that, it's their reputation on the line after all...

Yup, it is strange. I've never seen a genuine looking Chemi-con with a concentric bung before, only Panasonic, Rubycon, Sanyo (Suncon/Sun Elec[tronics]), and Nichicon.

Which begs the question: why put a Rubycon on a board if you are cost constrained?
Why not one of the cheaper first tier brands or even second tier brands...

Probably because of the bad rep the Chinese and Taiwanese brands were given in the duration of the "capacitor plague". At least in critical areas like the CPU VRM, I've noticed the tendency for OEMs to mostly use Japanese brands over the last decade.

That was me during desoldering, as linuxguru mentioned it was a quite difficult cap to desolder...

Yup, multilayer boards with RoHS solder and thick copper planes aren't at all easy to desolder components from.

Yea maybe, I attached another photo with the old caps standing next to two new unused caps.
There it's more clear to see how much the bung is collapsed inwards on the Samxon.

Either they used a different rubber seal or it really is a bit "depressed" inward.

It's also striking how similar the print is on the old Samxon and Rubycon caps!

The print on the Rubycon MCZ and Samxon GD isn't as clear as the print on the Rubycon MBZ, that's for sure. IIRC, Rubycon MBZs have always used that font for the series name itself and same goes for MCZ. Another reason why I didn't surmise that they're counterfeit is because to the best of my knowledge, Foxconn (or/and Flextronics) was the OEM for Intel back then. And as far as I know, Foxconn source their capacitors directly from the factory, unless they are sometimes getting their capacitors from distributors like Arrow...

[stj]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

the fake rubycon has a slip of the maker,

rubycon uses a coma "," for sub voltages - 6,3v
the others all use a stop "." - 6.3v

also the relative height of the "v"

[ChaosLegionnaire]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

errr if u use a cheaper bung, wont that decrease the overall effective lifespan of the cap? it seems their cost cutting is more aimed towards planned obsolescence and simply getting the product to outlast the warranty by one day more than anything else.

not to mention their way of cost cutting (more expensive custom order with a cheap bung) is simply nonsensical/illogical and makes no sense at all. rubycons are the most expensive cap brand around. they are more like a luxury item for bragging rights. there are other jap cap brands around offering similarly specc-ed caps for a cheaper price.

the 6.3v 3300uF Samxon GD cap next to it looks weird too, the bung looks collapsed inwards, never seen anything like it before!

the bung on the samxon does not look depressed inwards or dented to me. it appears to be a different bung being used that is flat at the spacing in between the two leads. not sure why samxon would have multiple different bungs tho. probably spares in case one type ran out as they had to spam caps as quickly as possible at the factory?

the fake rubycon has a slip of the maker,

rubycon uses a coma "," for sub voltages - 6,3v
the others all use a stop "." - 6.3v

also the relative height of the "v"

lol thats nice and a good catch! didnt realise the difference there. u made me dig out all of my stuff to look at all my rubycon caps. are u sure thats a definite sign of a fake tho? all of my mbzs use a dot for the 6.3v no idea why the one hansson has uses a comma? manufacturing variance or mistake on the part of the manufacturer?

[Per Hansson]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

Really good catch STJ!
I'm just more and more convinced that the whole lot of caps on this board are counterfeit.
Too many odd things with the caps are wrong...

Just did a Google image search which brought up some old posts with similar looking caps:

https://www.badcaps.net/forum/showthread.php?t=14198

https://www.badcaps.net/forum/showth...412#post101412

The print on the Rubycon MCZ and Samxon GD isn't as clear as the print on the Rubycon MBZ, that's for sure

Compare the print of the known genuine Samxon GD with the one found on the Intel board.
The one on the Intel board has more shitty quality of the print, and the edge of the stripe is jagged.
Now compare the known genuine Rubycon with the other one, coincidence I think not!

[Agent24]

Re: Intel D945GCLF Bloated Samxon GC due to overheating

Could they just be badly printed labels, 'seconds' as it were?

Maybe Dell got a discount on them.