Check this homemade WB out

[Sideroxylon]

Aircooled WB

This guy is claiming a 9c temp improvement over a Heatkiller LT. I'm not sure what to make of this. I just can't see that happening.


EDIT (by shazza): a bit late adding this edit, but since this is the first post in the thread, I've taken the liberty of adding to Sideroxylon's first post - Before you add a comment to the thread, please read the quoted source. The point of nateman_doo's waterblock - and a point he emphasizes in the quoted post at EVGA is to make a hybrid block that can work on air and on water. Yes, he does say that his block beats the Heatkiller by 9 degrees - BUT a very important distinction is that he's saying it beats it in HIS setup. He's not saying it will beat the Heatkiller by 9 degrees in all setups. A point of emphasis on this block is the fact that it can reportedly act as a satisfactory heatsink for either air or water cooling.

Just want to make sure everyone is clear on that before they post their comments.

[marxviper]

I wonder what would be the temp be when same size of block is used for water-cooling instead of air-cooling. I bet the temperature would be much cooler than this same block being air cooled

[Helloworld_98]

If this is true then maybe some heatsink company should make a water block which uses small heatpipes

[jumper2high]

69*C on a watercooled i7 on stock clocks?

I get similar temps on my i5-750 @ 4.0 GHz, air-cooled with a single 1300RPM fan.

[Vapor]

GPU temp is 18C lower on the homemade test (despite going to SLI), "System" (IOH) temperature is 19C lower on the homemade block test, and VREG temps are 33C lower (partially due to the presence of airflow, surely). These three point to a massive different in ambient temperatures in favor of his own block.

I know nateman_doo posts here, maybe he can clear some things up since this test looks pretty poorly setup.

[jumper2high]

GPU temp is 18C lower on the homemade test (despite going to SLI), "System" (IOH) temperature is 19C lower on the homemade block test, and VREG temps are 33C lower (partially due to the presence of airflow, surely). These three point to a massive different in ambient temperatures in favor of his own block.

I know nateman_doo posts here, maybe he can clear some things up since this test looks pretty poorly setup.

Ah, that might explain the somewhat high CPU temps under the Heatkiller

[Zeus]

Questionable to say the least...

He's using a lapped Heatkiller block too.

[Sideroxylon]

He states that during the test of his block that his gpu was in the loop vs the Heatkiller being all alone on the loop. Which makes the temp difference even more incredible.

He refuses to post any pics of his internals. Not sure why.

I think this is an interesting design but the results are extremely questionable.

[NaeKuh]

Im calling shins to this.

Reason is the medium used to transfer.

im sorry he's doing nothing but adding a copper sink with fins on it and throwing in a fan.


Well, his test would come true if the copper sink was chilled b4 it was placed.
The copper sink holds mass, so it would take longer for the sink to fully saturate.

[Six]

Maybe he will say: "Sorry forgot to tell you that i used for my block a chiller because room temperature was so high in the evening but who cares"

[OggeOJ]

Im calling shins to this.

Reason is the medium used to transfer.

im sorry he's doing nothing but adding a copper sink with fins on it and throwing in a fan.


Well, his test would come true if the copper sink was chilled b4 it was placed.
The copper sink holds mass, so it would take longer for the sink to fully saturate.

True, its a bigger mass for the CPU to heat up, so for even results, it should be on for a big bit longer.

[Genesis:.]

seems very sketchy. though imagine if it actually worked..(i know, no way in hell lol..)

[tool_462]

GPU temp is 18C lower on the homemade test (despite going to SLI), "System" (IOH) temperature is 19C lower on the homemade block test, and VREG temps are 33C lower (partially due to the presence of airflow, surely). These three point to a massive different in ambient temperatures in favor of his own block.

I know nateman_doo posts here, maybe he can clear some things up since this test looks pretty poorly setup.

Same thing I noticed. Much, much lower ambient.

[wez]

Well, he just have a 2x120 rad cooling an i7, so surly there must be benefits of adding a solid copper HS w. fan to the mix?

[vodka3dg]

wow, looks promissing... what i do not understand is why he is having so high temps on an i7 that runs stock speeds and stock voltages...

even for a bad batch, those temps are really high!

[jumper2high]

wow, looks promissing... what i do not understand is why he is having so high temps on an i7 that runs stock speeds and stock voltages...

even for a bad batch, those temps are really high!

You think it may have something to do with his room temperature being 95F (35*C) for the Heatkiller test, and his block being absolutely inferior (considering the likelyhood of him having appropriate tools to make 0.25mm micro channels like the best performing water blocks tend to have)

[ohms]

I really don't know why so many people here at XS are so quick to knock people for trying new inventive things.

[Vapor]

I really don't know why so many people here at XS are so quick to knock people for trying new inventive things.

Like turning on the AC?

[Alexandr0s]

Lol, I have to agree. Inventivity is great, but we'd like to see some solid evidence that it's actually an improvement. This isn't what I'd consider solid evidence.

[nateman_doo]

Man... lots of skepticism going on in here. I guess I can't please everyone, & but judging by how people criticize even far more reputable testing people I should expect nothing less then what I read here. Such is life.

Both tests were done exactly the same, nothing was switched but the block. I did one test right after the other, and didn't even touch the AC.

The air conditioner was cycling between 70-74°F for the duration of each test. The 9° temp difference was in core 1 and it peaked towards the end of the test. Both tests peaked at identical times during each test. There is no smoke and mirrors. The test was identical. No CHILLERS were involved (thank you Six for the deliberate flame)

The test was poorly setup INTENTIONALLY. I wanted worst case scenarios so that if anyone wanted one, they would have much better results. The more important fact is the test was IDENTICAL. Same ambient temp, same amount of water, same thermal paste, same program, same everything. I turned on the test, and shut the door to the room, and walked away. Checked back at the end and posted the results.

The only explanation that I can give for the huge swing of temps on the board is the fan pulling in air in that a regular heat killer block does not.
Fancy that, a fan pulls in cooler air. Wow, guess that's why the block was designed.

As for the internal surface area (jumper2high), The HK block is simply a copper plate-more like a shim-with micro fins on it. My block has channels that go around the perimeter of the block which if you figure the liner distance of "copper tube" it is considerably more surface area then a shim with fins on it. It is over 1/4" tall, by 1/2" wide by probably no less then 8" of linear distance through the block. Plus due to my rudimentary milling capabilities (being addressed in a few days) the chopping milling also adds far more finer "fins" which adds to additional surface area.

So if any other questions regarding the block you would like answered I would be happy to address them. If you think I don't know how to turn on an Air Conditioner, you can go trash some other thread. The block is well over 2lbs of copper with tons of internal surface area, tons of external surface area, and a fan.

How about offering better suggestions for testing instead of trashing my competence?

[nateman_doo]

He states that during the test of his block that his gpu was in the loop vs the Heatkiller being all alone on the loop. Which makes the temp difference even more incredible.

He refuses to post any pics of his internals. Not sure why.

I think this is an interesting design but the results are extremely questionable.

The pictures of both tests are posted there, and you see air cooled video cards. What GPU in the loop are you referring too? Perhaps a typo in my post which I should address?

[PiLsY]

The fact that you lapped the heatkiller destroys the test as a comparitive piece. Everyone knows lapping the HK kills its performance.

Even so, if you manage to match the performance of an unmodified HK without sacrificing flowrate you've done incredibly well. Doesn't look like there's much room in the base for water to travel - i'm expecting this block to be really restrictive. Similar internally to the xspc razor series?

Anyway, well done, nice block, you've actually done what most people just talk about.

[nateman_doo]

Thanks! There is plenty of room in the block, it is 1/2" thick before you even to the fins. The channel looks close to an "A" which goes around the perimeter of the block It has just a bit larger diameter then the same 1/2" hose that feeds it, so I don't think it is restrictive at all, but I have no way to calculate the resistance. I figure if has more cross sectional surface area then the bitspower fitting, then its not restrictive at all.

side note: On the HK block, how does lapping the HK block and the Chip skew the results of the block? if both are lapped to 2000 (i think? or 1500) then both blocks are making the desired contact. I could see lapping only one, and not the other, but if both are lapped on the same glass surface then both are making same 100% contact right?

[Halo_003]

Well guys, I believe him. I may be biased a bit since I'm on the same for a load more than here, and his tests may be biased a bit by the lapped heatkiller. I don't honestly know how much difference lapping it makes, I have an Apogee GTZ, so I can't say. I've watched the development of this block from the beginning and seen the results (OCs) on it compared to the HK results and these have been better. I'd always wondered if it helped at all to have a fan on a block but for whatever reason I never considered having a fan and fins on the block. It looks like it helps and it is legit. Even if the lapped HK doesn't do that well. Besides, everyone knows that the HK 3.0 CU version is the best. So maybe another test should be done vs. a CU version, or at least an LT, not an LC.

With regard to chillers, come on. Look at the size of this thing and look at the size of the HK. Yeah, this would hold the cold longer. I'll give you that. I think for a few minutes it would work damn well, but not more than 6 hours straight.. (IMO)

Another thing you could do to simplify it Nate, and make it more believable to the naysayers, run it again, and show the temps, and the prime 95 window. 'Cause that's one thing that was missing.

The fact that you lapped the heatkiller destroys the test as a comparitive piece. Everyone knows lapping the HK kills its performance.

Even so, if you manage to match the performance of an unmodified HK without sacrificing flowrate you've done incredibly well. Doesn't look like there's much room in the base for water to travel - i'm expecting this block to be really restrictive. Similar internally to the xspc razor series?

Anyway, well done, nice block, you've actually done what most people just talk about.

True. the flow rate is an interesting point actually. I'm not sure on how well it would work.

Even so, it is at least competitive to the HK line of blocks. Good job Nate.

[wesley]

side note: On the HK block, how does lapping the HK block and the Chip skew the results of the block? if both are lapped to 2000 (i think? or 1500) then both blocks are making the desired contact. I could see lapping only one, and not the other, but if both are lapped on the same glass surface then both are making same 100% contact right?

AFAIK, the HK block was designed not to be lapped.

Wes