Cooling for back of motherboard?

[nathanddrews]

Has anyone ever tried this and seen good results?

 

[Valantar]

Cooling what? How? Do you mean cooling onboard components (like the chipset) from the back? Given that the CPU sits on top of a socket, I'd gather there's little to gain there, at least. And outside of the CPU and chipset, there are few heat-generating components on a standard motherboard. Also, there's usually not room for any type of cooling back there, although I suppose you could stack a whole bunch of thermal pads on top of each other to transfer heat to the motherboard tray. Wouldn't be very efficient, though.

 

[Burpo]

Does this belong in CPU & overclocking?

 

[Valantar]

Does this belong in CPU & overclocking?

Excellent question. I'd say the Motherboards subforum sounds like a better fit, no?

 

[nathanddrews]

Cooling what? How? Do you mean cooling onboard components (like the chipset) from the back? Given that the CPU sits on top of a socket, I'd gather there's little to gain there, at least. And outside of the CPU and chipset, there are few heat-generating components on a standard motherboard. Also, there's usually not room for any type of cooling back there, although I suppose you could stack a whole bunch of thermal pads on top of each other to transfer heat to the motherboard tray. Wouldn't be very efficient, though.

Based upon what I've read about some weird, existing products and experiences, gains to CPU cooling are within the margin of error, but cooling for other motherboard components can see greater reductions. It's obvious that many people think it's a waste of effort - and for the most part I think that's correct. While not CPUs, some GPU water-cooled backplates offer very good gains. Maybe I'm just thinking out loud, but it seems like there are some opportunities for better cooling that are still only niche.

Does this belong in CPU & overclocking?

Excellent question. I'd say the Motherboards subforum sounds like a better fit, no?

It's sort of a hybrid question, so either location works for me, but it seems like the people that most often attempt it are doing overclocking. Several methods have been attempted over the years with minimal effect. Now that we're in an era of such incredible transistor density, more creative cooling methods are required. With the exception of Ryzen, delidding seems to be super effective in many scenarios, but requires a higher degree of fortitude and money (if it fails) than fans or heat sinks.

 

[Valantar]

It's sort of a hybrid question, so either location works for me, but it seems like the people that most often attempt it are doing overclocking. Several methods have been attempted over the years with minimal effect. Now that we're in an era of such incredible transistor density, more creative cooling methods are required. With the exception of Ryzen, delidding seems to be super effective in many scenarios, but requires a higher degree of fortitude and money (if it fails) than fans or heat sinks.

Even so, there's nothing on the back of the motherboard that actually gets hot directly, and the CPU die/dice sit on top of the substrate which sits on top of its pins which sit on top of the socket which sit on top of the fiberglass of the motherboard. That's quite a few layers of insulation. While the pins transfer heat into the copper layers of the motherboard, that's still rather negligible compared to the amount of insulation in between. Even if the PCB might heat up to 50-60C over time, cooling that off will have negligible effect on the CPU.

 

[nathanddrews]

Even so, there's nothing on the back of the motherboard that actually gets hot directly, and the CPU die/dice sit on top of the substrate which sits on top of its pins which sit on top of the socket which sit on top of the fiberglass of the motherboard. That's quite a few layers of insulation. While the pins transfer heat into the copper layers of the motherboard, that's still rather negligible compared to the amount of insulation in between. Even if the PCB might heat up to 50-60C over time, cooling that off will have negligible effect on the CPU.

Then again, maybe not. Using a pretty ghetto setup, this guy shaved 14C off the socket temps and 7C off core temps by adding a HSF to the backplate of his OC 3930K.
http://stanislavs.org/lga-2011-cpu-socket-backplate-cooling-modification/

 

[Valantar]

Then again, maybe not. Using a pretty ghetto setup, this guy shaved 14C off the socket temps and 7C off core temps by adding a HSF to the backplate of his OC 3930K.
http://stanislavs.org/lga-2011-cpu-socket-backplate-cooling-modification/

That the socket temp went down is a no-brainer, as it's much closer to the back of the motherboard than the CPU. Then again, socket temp doesn't affect anything at all. The 6C CPU temp reduction isn't bad, though. Higher than I would have expected. Then again, that seemed to be a hot-running system to begin with. Also, most backplates are not that flat and heatsink attachment friendly. I think the VRM cooling that he's doing at the bottom of the post has more merit, as those are mounted directly onto the board.

 

[nathanddrews]

...The 6C CPU temp reduction isn't bad, though. Higher than I would have expected...

The change is significant enough to make me want to try it.

 

[Valantar]

The change is significant enough to make me want to try it.

Well, if your case can fit something like that, your motherboard and/or backplate is flat enough to fit some heatsinks, and you have the time/parts/money, why not?

 

[nathanddrews]

Well, if your case can fit something like that, your motherboard and/or backplate is flat enough to fit some heatsinks, and you have the time/parts/money, why not?

Exactly, why not? LOL

 

[coercitiv]

The change is significant enough to make me want to try it.

Go ahead then, looks like fun. Just don't be surprised if you can't replicate the results.

It's one thing to be curious about something, in the sense of challenging common knowledge (based on science) through both creative thinking and testing, and a completely different beast to do the same based on someone else's testing, with only select evidence and results available for inspection.

Personally I find it very hard to believe one can remove 7C worth of heat from an 130W+ TDP processor using just a small fan at low rpm, but I certainly encourage you to try for yourself and prove the skeptics wrong!

 

[nathanddrews]

Personally I find it very hard to believe one can remove 7C worth of heat from an 130W+ TDP processor using just a small fan at low rpm, but I certainly encourage you to try for yourself and prove the skeptics wrong!

Therein lies the problem - if his results won't change preconceived notions, mine won't either.

 

[coercitiv]

Therein lies the problem - if his results won't change preconceived notions, mine won't either.

On the contrary, if you were to come up with your own results, presenting both findings and testing methodology, we would be having a far more productive conversation.

 

[Valantar]

On the contrary, if you were to come up with your own results, presenting both findings and testing methodology, we would be having a far more productive conversation.

Exactly. Two samples are far more representative than one, and would lend far more credence to this.

 

[monkeydelmagico]

Just be careful not to short out your mobo. Alot of exposed solder on the backside. I've done it with an improperly spaced CPU cooler backplate.

 

[nathanddrews]

Just be careful not to short out your mobo. Alot of exposed solder on the backside. I've done it with an improperly spaced CPU cooler backplate.

Yeah, I'm not touching any solder points! Thanks for the reminder though.