P4 direct die water cooling

[Puzzdre]

dax: blu-tack is made by Bostik, and uhu-tac by UHU (Germany). I got (and used) both of them, so far I cannot say the difference in quality except the color , blu-tack being bluuuu, and uhu-tac yellow...

[hara]

uhu-tac is softer

[TomC]

Since you're making a new block anyway, may I suggest an improvenent?

Glue a neoprene gasket to the block and clamp that to the CPU.
It is 100% watertight and allows easy removal for modifications or cleaning. It takes the strain off of the CPU pins too.

I might glue a "mounting base" to the PCB and then clamp the block assembly to that.
I use Marine Goop for the adhesive. An alternative would be an automotive trim adhesive. This should be perfect for the job but I have not tested that.




My block:

[utabintarbo]

Originally posted by TomC
...

Glue a neoprene gasket to the block and clamp that to the CPU.
...

How would it be "clamped" to the CPU without interfering with the socket? This has always bugged me.

Bob

[hara]

Volenti, are you sure direct die is better than a waterblock? The core is too small to be effectively cooled by water. What happens if you unplug your Block and place carthar's block instead (assuming you could mount it)?

You've already removed the IHS

[sunblade]

Here in the US, I've seen liquid gasket type products that might be similar to blue-tac in automotive? It comes in both adhesive and plain gasket varieties.

Also, does anybody know how durable the P4 core is? I wonder if the core would start to erode at all. I'm probably being too paranoid, tho

About Intel's IHS, does anybody think it'd work using the IHS as a baseplate? I mean, remove the IHS, then attach a direct-die block to the heat spreader and reapply the heat spreader with thermal epoxy?

[Arcturius]

Originally posted by sunblade
About Intel's IHS, does anybody think it'd work using the IHS as a baseplate? I mean, remove the IHS, then attach a direct-die block to the heat spreader and reapply the heat spreader with thermal epoxy?

I probably don't understand your question correctly, but the way I interpret it, you want to remove the IHS, use it as the base of a poly-topped block, then reattach it?

I would say you're just adding another poor thermal interface at that point with the thermal epoxy, since you are replacing Intel's thermal paste/epoxy with a different kind that is quite possibly of lower quality, and I doubt you could mount the IHS as well as they did.


It does make sense to me to remove the IHS, though. I had a great boost in my ability to cool my old K6-2 when I de-capped it; I doubt the P4 is much different in that respect (removing two thermal interfaces which are _not_ flat, b/c of the serials, etc...).

[TomC]

Originally posted by utabintarbo
How would it be "clamped" to the CPU without interfering with the socket? This has always bugged me.

Bob

I use the stock mounting cage with the 2 cam levers to mount my P4 blocks. The socket doesn't interfere. You could use the 4 holes in the motherboard and fashion a plate type hold-down. If you're dealing with socket mounting "ears" and an older style socket it gets a little harder.

I really like the P4 mounting system.

[Volenti]

Originally posted by hara
Volenti, are you sure direct die is better than a waterblock? The core is too small to be effectively cooled by water. What happens if you unplug your Block and place carthar's block instead (assuming you could mount it)?

You've already removed the IHS

It's better than any water block I've ever used, and I've made many, including a pretty close copy of Cathar's block.

Note though, it has a very small margin for error, and proper (multiple) jet design and implimentation is critical to get better-than-copper-water-block results.

anyway, here's a big update.

I've made a new-n-improved CPU direct die block, as well as blocks for the northbridge and Radeon 9700pro GPU.

I've gone back and improved the multiple jet approach that worked so well before, also incorporating a little "rice" in the design allowing the cores to be very visible.

But enough of my rambling, since you've only came here to see the pictures :P

on with the show;

The 3 volunteers;


The multiple layers of the blocks block laid out under their respective cores;


The 3 blocks after assembly and a polish;


The addition of the copper hose connecters;


Close up view of the GPU block, the surface mount components on the chip surface were sealed with epoxy, the "dreaded" blu-tack gasket can also be seen through the perspex;


Shots of the blocks mounted in position;




And finally the 1/2'' to 4x 1/4'' splitters I made up to run the whole mess.


Since I hosed the previous install of XP (only just got back up and running) I haven't had a chance to run any tests, I'll get some performance results up soon.

[hara]

Volenti, you rock!

This is revolutionary, You should start a company. Direct die was dead and you resurrected it.

I still think you should test the same setup with a good block to really see if your setup is the absolute best.

Let us know your results. It would be cool if you could stick everything in a case. Also, don't use pure methanol of the gpu core cause we all know what would happen if something overheated

[hara]

How about directly cooling the hd's with water?

Yeah, useless, like direct die cooling a nb

I really admire your work


One thing that i would like to ask, is direct die phase change cooling possible?

[hara]

I don't understand how the cpu block works. How exactly does the water exit?

[Volenti]

Originally posted by hara
Volenti, you rock!

This is revolutionary, You should start a company. Direct die was dead and you resurrected it.

I still think you should test the same setup with a good block to really see if your setup is the absolute best.

Let us know your results. It would be cool if you could stick everything in a case. Also, don't use pure methanol of the gpu core cause we all know what would happen if something overheated

my early results show this block to be slightly better than my previous multijet direct die block, by a mere 1/2 a degree or so (c)

since this setup is using evaporative cooling, I'm using straight tap water with a dash of dish washing detergent.( I use a microjet irrigation in-line filter to prevent the jet's from being clogged up)

I don't understand how the cpu block works. How exactly does the water exit?

The water enters via 2 diagonally opposite tubes that only run half way into the block, they feed into the central chamber that contains the jet holes, the water then travels through the jets, over the core, then back up the other 2 tubes that extend almost all the way to the bottom of the block. The outlet tubes also travel through the centeral chamber, however they are sealed at either end to prevent the water short cutting.

[Volenti]

Originally posted by hara
How about directly cooling the hd's with water?

Yeah, useless, like direct die cooling a nb

I really admire your work


One thing that i would like to ask, is direct die phase change cooling possible?

I could do a harddrive easily enough, but it would only be good for the wank factor, like the northbridge

as for the phase change, I doubt it, no point anyway, they're already killing the p4's with phase change + high V-core (northwood sudden death syndrome)

[hara]

High vcore kills cpu but i doubt really low temps kill them. What's better is that with a low enough temp you can stably oc p4s without running a high vcore.

[hara]

I heard you used a filter. I'm currently working on a project to Wc several PCs with a single bong. Can you give me more info on these filters, how they work and where to search for them?

Thanks

[Volenti]

Originally posted by hara
I heard you used a filter. I'm currently working on a project to Wc several PCs with a single bong. Can you give me more info on these filters, how they work and where to search for them?

Thanks

Any hardware/gardening store that sells those micro irrigration setups for the garden will also stock in-line filters (to stop the fine spray heads from clogging up) They look like a black plastic pipe bomb with hose fittings on either end, they have a removable internal sive that can be cleaned and re-used,and their cheap.

[nicozeg]

I've thoght about using that filters, but they are designed to use with higher pressure water. Have you tested the flowrate change with and without the filter?

[Volenti]

Originally posted by nicozeg
I've thoght about using that filters, but they are designed to use with higher pressure water. Have you tested the flowrate change with and without the filter?

I haven't actually measured it, but the ones I'm using seem to have the same impact that a single 90 degree elbow does.

You can always use 2 in paralell with Y adaptors.

[MadDogMe]

Are you using the triple 'head modded' pump setup for this Volenti?, cause you're gonna need abit of pressure/head for those three puppies are'nt you? ...

Kudos on another inspiring job...

[hara]

Tell us:

Ambient temp
Water temp
core temp
pumps

[nicozeg]

Are those hoses of the fuel line Kind? I’ve used that some time, are soft and cheap, but tend to develop a white layer in contact with water. It grows over time and deposit on the rest of the system. That could be a long term problem with your jets.

[Volenti]

Originally posted by MadDogMe
Are you using the triple 'head modded' pump setup for this Volenti?, cause you're gonna need abit of pressure/head for those three puppies are'nt you? ...

Kudos on another inspiring job...

I will be using the 3 stage pump mod, after I finish polishing it up, and get the valves sorted, currently just using a boring old "standard" 2.5m head pond pump. All 3 blocks combined, since they are plumbed in paralell, don't stress the crap out of the pump like the single CPU clock did before, however they are still far more restrictive than "normal" water blocks, and a high pressure pump will still be prefered.

Tell us:

Ambient temp
Water temp
core temp
pumps

since my ambient temps vary so much during a daily cycle all I can give is a snap shot at this point in time (the reason why I normally just give cpu above water temps)

ambient 29 c
water temp 23.5 c (evaporative cooling remember)
core temp 32.5 c (usual disclaimer about differences between different mobo's ect)
currently just a standard 2800L/H 2.5m head pond pump (to be replaced by the 3 stage pump mod)

Are those hoses of the fuel line Kind? I’ve used that some time, are soft and cheap, but tend to develop a white layer in contact with water. It grows over time and deposit on the rest of the system. That could be a long term problem with your jets.

Allthough I did buy the hose as fuel hose, it looks and feels like normal silicone hose and have been using it on and off for over a year with out any problems, I will keep an eye out for it though.

[nicozeg]

Nevermind, maybe mine were low quality stuff, definitely not silicone.

Man, your work is so stimulating!

Wish I have more time, sure I’ll be pump modding and direct dieing.

[Puzzdre]

I must admit that those see-through blocks look really adorable!
Do you think that direct die on an AMD XP would be too difficult because those L bridges? They could be insulated, but what would be good and TRUSTY insulation?

Keep up the great work!