-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
Coollaboratory Liquid PRO and copper-bottom aluminum heatsinks?
- Thread starter SPL Tech
- Start date
- Thread Starter
- #3
Can you clarify this? I dont understand what you mean here. I have some liquid metal left over and so I figure there is no reason to spend $15 on more TIM that will do a worse job if I can use liquid metal. For what it is worth, I am looking at using this on my GPUs.
- Joined
- Sep 15, 2007
- Location
- Small town Emlenton, PA
Your fine, use the CL Pro. The bottom of the HS is copper, there will be no galvanic issue.
- Joined
- Dec 14, 2010
You're talking about two different elements here, Tin versus Zinc. That liquid metal TIM reacts differently to different types of elements.
- Thread Starter
- #10
Well I replaced the TIM with the liquid metal today. I lapped the GPUs to remove the tin and I left the stock cooling pads on the VRMs alone (in other words, I dident replace them). I booted my computer up and it seems like the temps have dropped a bit, although not nearly what I was expecting.
Anyway, I am having problems with one of my cards. About 1-2 min in a 3D application my monitor will loose video input. In other words, the cards shut off. At first it seemed like the GPU was overheating because the problem only comes up in a GPU-intensive application. So I ran Heaven benchmark and watched the temps. The cards crashed at 67C, so it is not a GPU core temp issue. I disabled SLI and then took the functional card all the way up to 93C without a crash, so the issue is only with my secondary SLI card.
So the big question, what could it be? Does Evga or nVidia have some monitoring program to document card crashes? It seems like maybe one of the secondary objects that came with a cooling pad might be overheating. However, all the pads are still in place. I removed some of them during the laping process, but I put them all back and they appeared to be just as functional as they were before I touched them. They all stuck to the heatsink without issue. Last, I only covered the inner 90% of the CPU IHS with TIM so as to ensure the liquid metal would not spill over the sides. I screwed the GPU together then disassembled it again before final assembly to ensure the TIM was not spilling over the sides--it wasent.
Anyway, I am having problems with one of my cards. About 1-2 min in a 3D application my monitor will loose video input. In other words, the cards shut off. At first it seemed like the GPU was overheating because the problem only comes up in a GPU-intensive application. So I ran Heaven benchmark and watched the temps. The cards crashed at 67C, so it is not a GPU core temp issue. I disabled SLI and then took the functional card all the way up to 93C without a crash, so the issue is only with my secondary SLI card.
So the big question, what could it be? Does Evga or nVidia have some monitoring program to document card crashes? It seems like maybe one of the secondary objects that came with a cooling pad might be overheating. However, all the pads are still in place. I removed some of them during the laping process, but I put them all back and they appeared to be just as functional as they were before I touched them. They all stuck to the heatsink without issue. Last, I only covered the inner 90% of the CPU IHS with TIM so as to ensure the liquid metal would not spill over the sides. I screwed the GPU together then disassembled it again before final assembly to ensure the TIM was not spilling over the sides--it wasent.
- Joined
- Sep 15, 2007
- Location
- Small town Emlenton, PA
Swap cards and see if the secondary one as primary does it when it's not in SLI
- Joined
- Oct 10, 2012
- Location
- Secret lair
my guess is that the TIM did not set correctly so you have an uneven "reflow"
(or you have one microdot spillage somewhere)
Set the card up as single primary card and push it.
re-apply LM Tim and push again as single primary.
(or you have one microdot spillage somewhere)
Set the card up as single primary card and push it.
re-apply LM Tim and push again as single primary.
- Thread Starter
- #13
Well, I took the card back apart, removed some of the TIM and now it seems to work fine in SLI. However, one of the DVI ports does not seem to work. I get "unreadable signal" on my monitor when I use the main port. I guess it does not matter as I am just using that card for SLI anyway, but strange none the less. I also blew the card off with compressed air again, so who knows what actually fixed the problem.
However, it does not seem like my modification really did anything. My load temps only dropped 1-2C from the looks of it and I have an idea why. I think the reason why EVGA cakes on so much TIM is because there is a gap between the IHS and heatsink due to the VRMs. Really, to close that gap I think they need to use 0.25mm pads on the VRMs instead of 0.5mm. Lapping the GPU made it even worse because I sanded off some of the height of the IHS. I tightened the screws on the card a fair bit but that doesn't really seem to close the gap much.
I am suspecting the gap exists because when I applied the recommended amount of LM TIM and assembled the card and then dissembled it, there was nearly no LM TIM sticking on the HSF, it was all on the IHS. So I globed on 3x the recommended amount (about 3CC) and then the LM TIM started to contact the IHS. But even so I think I have a pretty limited amount of actual TIM contact with the IHS. I say this because when I pulled the card back apart, the thinnest sections of LM TIM on the IHS appeared dry and "cured" where as the thicker sections were still liquid. That implies the thinner sections of the LM TIM are not touching the HSF. I applied more TIM, but that actually increased the temps so removed some.
In short, the LM TIM is WAY better than the stock stuff, so if it was fully contacting the entire HSF, I am sure the temps would have dropped far more than they did. The fact that the temps did not drop as much as they should have implies the gap is too large for the LM TIM to fill.
I tried three variants: recommended TIM amount, 2-3x recommended and 3-4x. At 1x the temps were way too hot. There was almost no HSF to TIM contact. At 3-4X the temps were reasonable, and 2-3x I achieved the best temps, and that is the variant I am using now.
However, it does not seem like my modification really did anything. My load temps only dropped 1-2C from the looks of it and I have an idea why. I think the reason why EVGA cakes on so much TIM is because there is a gap between the IHS and heatsink due to the VRMs. Really, to close that gap I think they need to use 0.25mm pads on the VRMs instead of 0.5mm. Lapping the GPU made it even worse because I sanded off some of the height of the IHS. I tightened the screws on the card a fair bit but that doesn't really seem to close the gap much.
I am suspecting the gap exists because when I applied the recommended amount of LM TIM and assembled the card and then dissembled it, there was nearly no LM TIM sticking on the HSF, it was all on the IHS. So I globed on 3x the recommended amount (about 3CC) and then the LM TIM started to contact the IHS. But even so I think I have a pretty limited amount of actual TIM contact with the IHS. I say this because when I pulled the card back apart, the thinnest sections of LM TIM on the IHS appeared dry and "cured" where as the thicker sections were still liquid. That implies the thinner sections of the LM TIM are not touching the HSF. I applied more TIM, but that actually increased the temps so removed some.
In short, the LM TIM is WAY better than the stock stuff, so if it was fully contacting the entire HSF, I am sure the temps would have dropped far more than they did. The fact that the temps did not drop as much as they should have implies the gap is too large for the LM TIM to fill.
I tried three variants: recommended TIM amount, 2-3x recommended and 3-4x. At 1x the temps were way too hot. There was almost no HSF to TIM contact. At 3-4X the temps were reasonable, and 2-3x I achieved the best temps, and that is the variant I am using now.
- Joined
- Oct 10, 2012
- Location
- Secret lair
since you found the issue, if you replace the pads with thinner pads, then you are settled 
Or... some thin copper or silver sheet to put between the ihs & the block...
Or... some thin copper or silver sheet to put between the ihs & the block...
- Joined
- Oct 10, 2012
- Location
- Secret lair
no, most pads are terrible performers... not good enough for the main GPU... you really need a shim or thin copper sheet or foil, depending on how much thickness is needed.
Even an IHS recycled from somewhere might an option. Worst case: cut a section out of a (metal) cookie tin. Or use a razor blade (which is steel).
The guys in the Peltier section have experience with those kind of thin shims, sheets et all although they call them hot plates & cold plates
Have a look overthere
Check for dimensions / thickness:
http://www.frozencpu.com/products/14971/ex-blc-1051/EK-VGA_Supreme_HF_HD7970_Cu_Adapter.html
http://www.frozencpu.com/products/20764/thr-217/Pure_Copper_Thermal_Shim_-_20mm_x_20mm_x_05mm.html
http://www.frozencpu.com/products/20761/thr-215/Pure_Copper_Thermal_Shim_-_15mm_x_15mm_x_05mm.html
Note that dumping the pads completely an using a non-conductive tim like ic-diamond is an option... but you'll have to do several mounts to find out if there is contact everywhere.
Even an IHS recycled from somewhere might an option. Worst case: cut a section out of a (metal) cookie tin. Or use a razor blade (which is steel).
The guys in the Peltier section have experience with those kind of thin shims, sheets et all although they call them hot plates & cold plates
Have a look overthere
Check for dimensions / thickness:
http://www.frozencpu.com/products/14971/ex-blc-1051/EK-VGA_Supreme_HF_HD7970_Cu_Adapter.html
http://www.frozencpu.com/products/20764/thr-217/Pure_Copper_Thermal_Shim_-_20mm_x_20mm_x_05mm.html
http://www.frozencpu.com/products/20761/thr-215/Pure_Copper_Thermal_Shim_-_15mm_x_15mm_x_05mm.html
Note that dumping the pads completely an using a non-conductive tim like ic-diamond is an option... but you'll have to do several mounts to find out if there is contact everywhere.
Last edited:
- Joined
- Dec 14, 2010
- Thread Starter
- #18
no, most pads are terrible performers... not good enough for the main GPU... you really need a shim or thin copper sheet or foil, depending on how much thickness is needed.
Even an IHS recycled from somewhere might an option. Worst case: cut a section out of a (metal) cookie tin. Or use a razor blade (which is steel).
The guys in the Peltier section have experience with those kind of thin shims, sheets et all although they call them hot plates & cold plates
Have a look overthere
Check for dimensions / thickness:
http://www.frozencpu.com/products/14971/ex-blc-1051/EK-VGA_Supreme_HF_HD7970_Cu_Adapter.html
http://www.frozencpu.com/products/20764/thr-217/Pure_Copper_Thermal_Shim_-_20mm_x_20mm_x_05mm.html
http://www.frozencpu.com/products/20761/thr-215/Pure_Copper_Thermal_Shim_-_15mm_x_15mm_x_05mm.html
Note that dumping the pads completely an using a non-conductive tim like ic-diamond is an option... but you'll have to do several mounts to find out if there is contact everywhere.
What about the liquid metal Coollabatory pad? The problem is now that I have LM TIM on it, it is going to be a serious PITA if I want to change out to something other than a pad or more LM.
- Joined
- Oct 10, 2012
- Location
- Secret lair
the LM pads are thin, very thin... and very conductive and they need reflow application, which makes it unsuitable for anything else than CPUs. Even on the main chip of a GPU application is risky business... need to make it melt, not overspill without "burning up" the card... not for the faint at heart
http://www.coollaboratory.com/en/products/liquid-metalpad/
http://www.coollaboratory.com/en/products/liquid-metalpad/
