I had a little idea about increasing the surface area of the ihs. How about lapping the ihs, then applying lots of tiny little blobs of copper paint. The conductive stuff that you can get to repair car window demisters, which used to be needed to unlock some cpu's and even small amounts of vmodding. I'd say have a grid of dots, maybe like ~half the ihs surface area. Next it would be great to scrape the ihs between each of these dots, so as to roughen the ihs maximising surface area. After this, clean and apply another layer of blobs ontop of the original blobs, so that they are now twice as tall. This can be repeated until a nice height is obtained, say i dunno, a few mm would be awesome, but a couple of mm is probably good enough? that's probably anywhere from 5 - 10 layers? The stuff dries pretty quick so it's not that much work really, the copper paint has a high % of copper in it so they will conduct the heat really well, and you have alot more surface area for the water to extract heat from. You'd probably wanna do this on a prescott Unless you don't scrape the ihs, just use blobs of copper, and it can always be lapped off at the end of the experimenting, leaving you a flat ihs. I dunno about how much pressure is needed for the o-ring? There's a couple of links here that will have good info for ya.
Yes I agree I should have aimed for the centre of the block so here is revision 2, it also makes it a bit more simple to produce. There is a cavity behind the jets to allow coolant to even out. I might try and simplify it more by changing the exit port to a side mounted right angle connector. As far a the o-ring is concerned, looking at the specs if I do a 2mm wide 1mm deep groove I need a 1.5mm o-ring and the presure becomes moot as it is compressed till the block meets the cpu.
I don't really get the point of this project, as the IHS is never going to be as good as a dedicated waterblock base, and if you want more direct heat removal with less layers acting as a medium for heat transfer then remove the IHS and apply a waterblock directly to the core.
but with the waterblock you have the copper base and the IHS between the heat source and the water, this has just the IHS and so less unhelpful, so to speak, materials that the heat has to pass through to be conducted away.
But as I said removing the IHS is far easier and a superior solution to removing the waterblocks base
I like it... depending on how thick the IHS is, maybe you could mill in channels or something? I think that would work better than just sanding it, if it worked.
The IHS is probably pretty thin but you could rough it up with an engraver by shading the surface. Better than sanding? Also the exit port on the rerevised waterblock looks a bit restrictive, but then again that may be a good thing in that it will give the water more time to absorb the heat, yes? I expect your going to polish the inside of that block to help improve flow silly me of course you are! NZmm
well, the IHS is not flat from the factory.. and without pressure on the IHS itself it is not in direct contact with the cores underneath. The IHS has it's own thermal paste between it and the cores too afaik.
I know the IHS is thick enough to cut a channel in to place a small thermistor for HSF qualification. I *think* it's in the thermal & mechanical specifications document for the CPU.
Okydoky how about this modification, make the jet block deeper and grooved, so it applies presure to the IHS and breaks up flows even more. If I was clever with the milling I could also angle the ends of the grooves to channel the flow toward the outlet.
I like it. Edit: OOOps! Poisonous got in there first. I meant the new block, although that avatar is nice.
I have a sieg X2 mini mill from axminster in the UK, its prety good for rough stuff, I would like to convert it to CNC though.
i can see why a water block is more effective, metal + paste + metal, the paste acts as a high conductor to something that is able to cool better.
The latest design looks good. I agree with applying pressure to the IHS as well. Another thing you could do - apply a blob of copper paint to each 'pin' that's thouching the IHS, so it also acts like an extra heat transfer block!
I don't see why people thing a generic block will be more effective than this, am I missing something. Surely, whatever is closest to the heatsource is warmest. With a waterblock you have more material for the high temperature to pass through to then be dissipated whereas with this is can be dissipated straight away.
I agree with DaveVader - surely it's better to gain the efficiency that you would lose on a conventional copper block + TIM?
Generic not so much, but i like to think of a good thermal paste as something like a metal bond between the two metals.
If this doesn't work better than a regular block, it will be because the benefit of increased surface area between a block and the water outweighs the thermal contact resistance between the ISH and the block.