GPU Shims

[ratdude747]

I need to do a re-pad on my Dell Latitude D630... last time I cleaned it out the pads didn't do too well... it will be do for another cooling clean soon and might was well put fresh pads in while I have the heat sink off.

My question is whether those copper GPU shims they make have any improvement or whether they are a waste of money.

[PCBONEZ]

Re: GPU Shims

https://www.badcaps.net/forum/showth...ighlight=Shims

[ratdude747]

Re: GPU Shims

ok. noted.

[thesloc]

Re: GPU Shims

Yes, they work.

I had an HP that would inevitably overheat. The problem was fixed with a shim. It's about keeping a tight thermal connection. The shim assures you that there's little to no gaps between the heatsink and chip.

[mattbrad2]

Re: GPU Shims

Agreed - with using copper shims. I'm a little surprised with some of the members here saying not to use them. I see the argument, but real world results speak louder than theoreticals. I've seen huge GPU temperature differences in using a shim vs the stock thermal pad.

[ratdude747]

Re: GPU Shims

too late, "ordered" thermal pads already (somebody I know is shipping me some).

the pads eliminate gaps too... they have a "squish" to them which is form fitting...

[mattbrad2]

Re: GPU Shims

Thats what thermal paste is for. You won't get anywhere near the lower temps you would using copper but I suppose first hand experience is always the best way to learn.

[ratdude747]

Re: GPU Shims

whatever...

but, anyway, i did soem research, and in the case of the D620/30, it isn't that great of an upgrade due to the modified fan curves.... if the fan was at full blast all the time or the cooling system was ineffective (such as with the HP), yeah, it would make a difference. however, the copper isn't a big upgrade due to the agressive fan curves (lower temps= slower fan speed). all you might do is lower your power usage marginally, but thats it.

also, iff you look at how the HS is mounted, you will note that the parts that get a pad are outside of the screw clamp perminter, which means those parts are only clamped to one side in respect to the chipset and GPU chips.

For a rigid cooling stem (like a shim or CPU HS) to work, you need clamping force on at least 2 sides, with even distribution of clamping forces to ensure even and flush contact; otherwise, you could be contacting only part of the chip, which kills cooling efficiency. anybody who has had an Intel LGA push-pin fail on a running system will know about this issue.

hence, a pad is needed, as it has some "squish", which will always contact the chip evenly due to the form fitting nature of the pad.

thermal compound is not meant to do such, as it only is designed to better utilize the surface area of the chip, not to accommodate minute inconsistencies in HS clamping force or clamping angle (caused by the unbalanced clamp forces). In fact, one piece of evidence of an LGA push-pin failure is when the thermal compound is noted to have leaked off the chip/hs and onto the the mobo area below the LGA socket (had this happen before).

thus, unless you get the shim size and surface angles EXACTLY right on a individual unit basis, you could ruin cooling performance.

[thesloc]

Re: GPU Shims

Well yes, you need to clamp it down evenly...

I think the last Dell D630 needed in fact a copper shim. But I think the thermal pad might work too.

But you can't go wrong with a combination of the both either.

[thesloc]

Re: GPU Shims

Also, how can a GPU shim be a waste of money?

Dude they're like .99 cents on eBay at most, SHIPPED.

If you don't think it's worthy of experimentation then you're flat out broke or just cheap.

[ratdude747]

Re: GPU Shims

umm... they were more like $7 if you bought them from the US... I don't buy from china.

as for not wanting to experiment, this laptop is a daily driver for me... for my college studies it is mission critical.

at this point arguing is a , so I am getting off the soapbox.

[thesloc]

Re: GPU Shims

Nah, they're .99 cents from an U.S. seller.

Copper is copper.

[ratdude747]

Re: GPU Shims

Nah, they're .99 cents from an U.S. seller.

Copper is copper.

and copper is expensive.

to each his own, this dicussion is officially a

[thesloc]

Re: GPU Shims

""ok""

[ratdude747]

Re: GPU Shims

""ok""

good.

[severach]

Re: GPU Shims

The reason against shims is that they are all the wrong size. Too thin and it won't cool. Too thick and it stresses the chip. Too wide and gravity will work it out. Most sellers don't bother showing the size or don't provide enough decimal places. Buy from someone who knows sig figs and how to use a caliper.

The paste shims are even worse. Paste doesn't conduct heat well. 1mm of paste conducts heat much worse. Too many reapplications and it breaks or squishes. There's no way to win with this stuff.

[thesloc]

Re: GPU Shims

Size doesn't matter as much as thickness does. The contact (ferrite beads??) on the top of GPUs on the one hand, could benefit by being covered with kepton tape to avoid direct contact with copper. I'm guessing that's a no-no.

My shims are usually 1mm thick. It's just that nVidia graphics suck on laptops. ALWAYS bound to breakdown.

[ratdude747]

Re: GPU Shims

not all nvidia laptop chips are bad, just certain ones...

that said, i have heard of "bad" nvidias being long lived... I think it is a combo of good cooling maintence (cleaning out the heatsink fins, etc.), using up to date bios (in the case of dells with the modified fan curves), and perhaps a bit of luck.

my laptop was made in june 2008, and it has been strong so far (knock on virtual wood). Even though it is a black sheep among certain forum members, I like it.

[thesloc]

Re: GPU Shims

most laptop gpus manufactured after 2007 will suffer from inevitable breakdown due to alloy use (lead-free).

lead-free (aka that "RoHS" shit) will cripple the electronics industry especially now since the demand for speed and process is even higher = higher heat. and since cooling systems especially in laptops aren't catching up, we have a problem.

have you ever done a reball or reflow of a gpu, ratguy?

[ratdude747]

Re: GPU Shims

I have never done a reball but i have used a hot air station before and I know about that.

that issue is fixable... AFAIK once you reball it and make sure you get the reflow done the right way, the problem is solved.

what my laptop has is the NVIDIA underfill epoxy bug... not fixable, nor do I think cramming a shim will fix. Due to the way the fan curves work on my laptop, adding the shim makes it worse (the fan takes longer to kick in and as a result, the chip stays hotter for longer).

BTW, did i mention this thread was a ?