Streacom DB4 Water Build (now flat heatpipe)

[msystems]

@msystems> so you think StevenG will get better results with a shim inbetween the GPU die and the flat heatpipe? Even when the latter is supposed to have a better thermal conductivity than pure copper?

Yes. 2 reasons:



1. GPU die is smaller in width than the flat heatpipe. The flat heat pipe is composed of tiny heat pipe lanes and unless they all contact the heat source I don't think they reach Max effectiveness.


2. Not really a performance issue but safety/ durability. Mounting pressure will deform the flat heatpipe over the die, causing permanent damage to the heatpipe and possibly the die. They deform super easily.

 

[StevenG]

Yes. 2 reasons:



1. GPU die is smaller in width than the flat heatpipe. The flat heat pipe is composed of tiny heat pipe lanes and unless they all contact the heat source I don't think they reach Max effectiveness.


2. Not really a performance issue but safety/ durability. Mounting pressure will deform the flat heatpipe over the die, causing permanent damage to the heatpipe and possibly the die. They deform super easily.

I will get some longer bolts on order and give it a try :-)

 

[StevenG]

Ordered a shim and some bolts, what about the CPU, shim aswell?

 

[msystems]

You could probably get away without a shim on the CPU. it has that copper IHS to protect the die and spread the heat to a wider area which should be almost the width of that heatpipe. The wider area also should mean it won't deform as bad.

 

[BaK]

Sorry for the late reply, had to do some searches

Yes. 2 reasons:
1. GPU die is smaller in width than the flat heatpipe. The flat heat pipe is composed of tiny heat pipe lanes and unless they all contact the heat source I don't think they reach Max effectiveness.

2. Not really a performance issue but safety/ durability. Mounting pressure will deform the flat heatpipe over the die, causing permanent damage to the heatpipe and possibly the die. They deform super easily.

While I'm more than ok with number 2, I don't think reason 1 is valid.

How I understand the quotes below, there is no lanes inside the flat heatpipe / vapor chamber but only posts, and the heat is spread in both X and Y axes on its base.

Here [on the right] a vapor chamber replaced the copper base plate, spreading heat more evenly across the heat source and transferring it more effectively to the heat pipes.

One piece vapor chambers make direct contact with the heat source, have a multi-directional heat flow, and can support clamping forces of up to 90 PSI.
Source: http://celsiainc.com/blog-heat-pipes-and-vapor-chambers-whats-the-difference/

Inside a vapor chamber is a series of small posts that keep the structure from collapsing under exterior atmospheric pressure. (The whole thing is in a hermetically-sealed vacuum). The posts also serve the purpose of helping the liquid flow to where it is needed.

Source: https://www.maketecheasier.com/what-is-vapor-chamber-cooling/

The photo above shows the internal structure of a typical copper/water vapor chamber. Arrays of internal posts provide structural support to accommodate the pressure difference across the vapor chamber envelope. A thin layer of wick lines the inner wall of the vapor chamber.

The vapor chamber is bonded to the base of the heat sink. It accepts concentrated heat inputs and transfers the heat uniformly over the entire heat sink base.

Source: https://www.1-act.com/products/vapor-chamber-assemblies/

Dunno then if adding a copper shim sort of prevent the vapor chamber to spread the heat as fast as it could without one.
Or maybe this downside is only present at the start of the process, until the shim gets hot all over.
Intuitively I would use a shim for reason 2 but a thin one to get it hot as fast as possible. Dunno if that makes sense...
The effect of the shim is probably minimal and there will not be any difference in real life. We will see if @StevenG will notice something

 

[BaK]

Tricky to get the right angle now, i will get some better ones next time the side is off but that Arctic silver 5 is like glue !!!!

Thanks for the pics!
This Streacom stock CPU block looks really versatile, Intel, AMD and even NVidia compatible!

Too bad I don't see it listed in their available accessories.

 

[msystems]

You may be correct regarding the vapor chamber. This was one of my assumptions from the way the surface finish looked (with lines running lengthwise) and also the semicircular grooves/indentations, found at the very end of each flat heatpipe. I sort of figured that meant that that there were metal lanes in there. But its maybe an imperfection left from the machine that pressing them, leaving those lines there, and from the machine that cut & seals the end leaving a grooved pattern. I'm tempted to cut one open and find out.

Either way, the thickness of the shim used should be kept to a minimum.

 

[StevenG]

Shims have arrived so i will test your theory at the expense of £2 haha. Still waiting for my parts from streacom so until those turn up not much point in changing anything.

 

[StevenG]

Thanks for the pics!
This Streacom stock CPU block looks really versatile, Intel, AMD and even NVidia compatible!

Too bad I don't see it listed in their available accessories.

Thy should list it then they coul sell them to people wanting to do the GPU aswell

 

[StevenG]

Tiny update, CPU mount and additional brackets are in transit coming from Streacom in china, so not sure when they will arrive. Copper shims and longer bolts are here waiting to be installed id rather do it all at once.

 

[StevenG]

CPU mount turned up :-)



[/URL]

 

[BaK]

You may be correct regarding the vapor chamber. This was one of my assumptions from the way the surface finish looked (with lines running lengthwise) and also the semicircular grooves/indentations, found at the very end of each flat heatpipe. I sort of figured that meant that that there were metal lanes in there. But its maybe an imperfection left from the machine that pressing them, leaving those lines there, and from the machine that cut & seals the end leaving a grooved pattern. I'm tempted to cut one open and find out.

Thx for the explanations.
I would be very interested to see how they really are inside!
But take care, from what I read they are sometimes filled up with something else than water that you may not want to inhale/touch...

 

[StevenG]

Thx for the explanations.
I would be very interested to see how they really are inside!
But take care, from what I read they are sometimes filled up with something else than water that you may not want to inhale/touch...

Yea I wouldn't cut the Amec ones as they are filled with Acetone.

Sorry for the late reply, had to do some searches

While I'm more than ok with number 2, I don't think reason 1 is valid.

How I understand the quotes below, there is no lanes inside the flat heatpipe / vapor chamber but only posts, and the heat is spread in both X and Y axes on its base.



Dunno then if adding a copper shim sort of prevent the vapor chamber to spread the heat as fast as it could without one.
Or maybe this downside is only present at the start of the process, until the shim gets hot all over.
Intuitively I would use a shim for reason 2 but a thin one to get it hot as fast as possible. Dunno if that makes sense...
The effect of the shim is probably minimal and there will not be any difference in real life. We will see if @StevenG will notice something

We will soon find out, when my brackets turn up I will get back to it! Got a couple of different thickness' but I don't want to be taking it apart too much and the thinner one (1.2mm) isn't flat so. It inclined to use it.

 

[StevenG]

Ok guys sorry it’s been so long I’ve been flat out at work but I’ve added the shim and cannot quite belive the result ! It’s knocked 30 degrees C off my load temperatures, I ran furnark for a few minutes and got it up to 56, the 3d mark hitting a top temperature of 61 degrees load! Very happy. I seem to have a power issue but I’m sure I can sort that somehow atleast the temps are a lot better:



Streacom delivery :



Shims 1.6mm x 2 and 1.2mm (didn’t use as it wasn’t flat)






Checking contact

Results

 

[StevenG]

Would like to say thanks to Msystems for the suggestion it’s worked better than I had imagined ! I think the flat heat pipes certainly work well with the DB4! Thanks

 

[msystems]

Wow, that turned out great. I think you have one of the best DB4 Gpu mods I have seen on here. Seems like a very good result and is also compact due to the pipes being flat. Probably less cost than a large copper block.

 

[StevenG]

Wow, that turned out great. I think you have one of the best DB4 Gpu mods I have seen on here. Seems like a very good result and is also compact due to the pipes being flat. Probably less cost than a large copper block.

The only downside is I’ve got to take it all apart to install the CPU mount nuts ! Gutted ! I may have some time later to do it all and get everything cleaned up! I definitely agree with you though it’s a better use than the copper heat pipes and it’s a 120w GPU aswell! I’m sure I could get something more powerfull to work too! A single long pipe would be even more effective than my current setup I belive.

 

[StevenG]

CPU done this wasn’t easy at all and very messy with all the TIM results are not great I need to check the seating position of the shim but after all the work today it’s going to have to wait, initial idle was 27 degrees sat in Windows, Prime 95 on blend about 78 degrees and on small FFTs I saw 99 degrees and stopped the test. The sides really don’t even get warm so I’m wondering if the heat pipe can’t take the bends or the seating is incorrect. Fine for normal use though.

 

[aquelito]

Wow stunning bending job !! Did you do it by hand ?
Those coolpipes are easy to bend but do not forgive any mistakes : almost impossible to bend again without breaking them.

Regarding your CPU temps, I see two possible issues :

1. From my experience, superposing two coolpipes does not work. Only the pipe in direct contact with the CPU will work.
2. From Amec recommandations, the coolpipes work much better vertically : the radiator stays above the heat source so that the condensated liquid can go back to the heat source quicker, thanks to gravity.

I think one pipe is enough for the CPU ; what you could do is try to "link" two rads with one pipe

Last week I tried to cool an i7 7700 with my HFX mCubed, made around two rads measuring 220 x 75 x 40.
I'm using a 50 x 50 x 5 mm piece of polished copper between the CPU and the 250 mm pipe.
Around 50 mm (20 mm on the left, 30 mm on the right) of the pipe is on contact with both rads.

Idle temp stabilized around 35° (ambiant temp around 20°).
After 20 min of Prime95, the CPU reached a temprature of 75°C, open case.

Both rads were very hot to touch and did saturate very quickly ; I did not try to test it any longer but I guess I would have reached some pretty toasty temps.

Conclusion : rad surface matters a lot, as well as respecting a vertical setup.
Surface contact between the pipe and the rad did not matter that much, as only 1/5 of the pipe length was enough to saturate the radiators.

 

[msystems]

Wow those CPU heat pipes are crazy. It is true that they are more effective vertically, but I think there is a problem with the thermal transfer. I hope it is not the bend radius. Hard to say.


I am suspicious of the contact with the CPU. The heat pipe looks like it is not completely flat due to the tiny ridges. Is it?

And also, any chance the CPU retaining socket is interfering with the contact?