Possible D-Tec fuZion 1 modification.

[TheThreeDegrees]

Right gonna stick my neck out here and say I may have found a way to improve further on the FuZion 1 block.

Has anyone tried this:

http://i21.photobucket.com/albums/b300/fabtrader/21052008395.jpg

Basically I have removed the pins from around the die impact area.
My reasoning for this is that once the water impacts the top of the pins it is actually diverted away from the point of impact which should IMO be the base plate of the block.

These tests are not conclusive and by no means scientific but here are some VERY BRIEF before and after shots of OCCT running on the proc.

BEFORE CPU 1:
http://i21.photobucket.com/albums/b300/fabtrader/2008-05-21-11h52-CPU1.png

AFTER CPU 1:
http://i21.photobucket.com/albums/b300/fabtrader/2008-05-21-13h31-CPU1.png

Before CPU 2:
http://i21.photobucket.com/albums/b300/fabtrader/2008-05-21-11h52-CPU2.png

After CPU 2:
http://i21.photobucket.com/albums/b300/fabtrader/2008-05-21-13h31-CPU2.png

Anyone else want to give it a go??!!!

[Giannis86]

this looks very interesting..i'd love to see some more definite results..

it is kinda late now but however i think that you should have relied on a better testing method. for example stress testing with Orthos or Prime and logging the temps with realtemp

anyway thumbs up for testing this

[nikhsub1]

There is no doubt that that mod may be of significance... how did you remove the center pins? I have a couple of dtek bases laying around, perhaps I can give it a go.

[Kayin]

Very interesting, I'm going to be watching this...

[Zehnsucht]

Too bad you only left it running for 5 minutes. It's hard to make any conclusions after only that short time.

I agree with RRR, it's very creative and deserves a big thumbsup, but that little data to compare before and after doesn't seem very thought-through after such a permanent mod.

[jas420221]

Looks promising!!!

But as you and other stated, needs more testing to come up with any conclusion. Too little data to extrapolate anything conclusive at this point in time.

/subscribed!

[Xilikon]

Sorry to be a bit critical but the improvement could be caused by a better mount. This is why if you want to prove a improvement, you need to do the testing the proper way like Scott do ;)

[WoZZeR999]

I'm going to have to join Xilikon on this one. I also noticed that the before and after are running at 2 different FSB's, so who knows what else changed in between tests.

[Xilikon]

I'm going to have to join Xilikon on this one. I also noticed that the before and after are running at 2 different FSB's, so who knows what else changed in between tests.

good eye, there's a slight overclock in the before pic which can increase by 1-2 degrees. I bet that if you don't do that, there would be no difference at all beside error margin.

[WoZZeR999]

After looking again at the charts, with cpu1 having a 1c difference, and cpu2 having a 0c difference, I have to call BS. I have had my Q6600 undervolted/clocked to 266x6 @ .9v (calibration for real temp), and I saw at least 2c (from what I remember).

If I am wrong, and someone else can show that this will make a processor produce no extra heat at load, then I will fully admit that I was wrong. I just don't see how removing the minor flow direction change could give that much difference in temps.

[TheThreeDegrees]

Just spotted that slight OC myself.Its the damn Uguru that resets itself when i have to take out the power supply and do some plumbing:p:

anyways no BS here:up:
Its just me playing around with my system experimenting with the blocks etc.

It is a pity that I didnt run Orthos longer but I am hoping if someone else wants to give it a try that would go some way to being conclusive.the pins are easily removed by bending them a little.

This is xtreme systems forum isnt it?;)Cmon lads push the boundaries a little and do something crazy once in a while.
You never know what you might come up with!!!!!!!!!!!!!!!!!!!!!!!!!!

[iandh]

There is no doubt that that mod may be of significance... how did you remove the center pins? I have a couple of dtek bases laying around, perhaps I can give it a go.

I'd pretty much put money on it. I don't know if you remember the last block I did, but it had a cupped center at the inlet, so that the water would hit the base directly instead of being deflected by a high pressure area where the water hits the center pins/slots.

Preferably the pins would be snipped very precisely, so that there are still small round nubs existing at the center of the base. This would be preferable to completely removing them IMO.

The block already has this design feature in its shorter center pins (along with another reason), but I had always thought that another step or two shorter pins as you progressed towards the center would be of benefit, I just never had one on hand to play with since I always have made my own blocks.

edit: I don't hold much stock in the testing done above, especially with the chip slightly underclocked on the lower temp "after-mod" shots. I could definitely see a hundred mhz causing a slight temp difference. I think with this done correctly and with nubs still existing right over the core, we could stand to see maybe a 1C total improvement.

[jas420221]

This is xtreme systems forum isnt it?;)Cmon lads push the boundaries a little and do something crazy once in a while.
You never know what you might come up with!!!!!!!!!!!!!!!!!!!!!!!!!!Well since you base numbers are already established and the only variable is the actual mod you made, and you are the only one that has it done, you should probably try to test it a little more properly and thoroughly before others go off half :banana::banana::banana::banana:ed for something that may/may not work. Im sure if there is some chance (more than what you have done so far) people will help out and attempt it. :up:

[Giannis86]

just oc the cpu to those levels and try again...

ofc there is the inconstency of mounts also to be considered along with other variables but at least there will be a "more fair" comparison

[TheThreeDegrees]

Ok just to dispell the myth that the slight overclock caused such big differences I have re-run Orthos again with the same Ambient as earlier today and these are the results after 6 minutes.

http://i21.photobucket.com/albums/b300/fabtrader/2008-05-21-20h25-CPU1.png

http://i21.photobucket.com/albums/b300/fabtrader/2008-05-21-20h25-CPU2.png

[WoZZeR999]

Do you have a stuck sensor? Have you had core 0 less than 41c or core 1 less than 42c? The only reason I ask is because it may help show the true delta a bit more. Not that you can fix this, other than letting your cpu temp raise to around 45 by turning the fans off and running.

[Zehnsucht]

The first minute of the OCCT test is IDLE. 5 minutes is hardly enough time to let the water warm up. For my Noctua AIR cooler it takes almost four minutes to reach max temp (see attachment). I'd image it takes longer for water to stabilize.

I really don't understand why you can't let it run for one full hour?

[WoZZeR999]

The max temp isn't what is important, it's the delta from low to high. Waiting for max temp really only shows what your radiator/water is capable of doing. But a good water block will have a lower from idle to load.

[FrogBite]

The blocks have pins to improve heat dissipation, removing the pins and getting better results makes no sense. Use your common sense.

By your conclusion a block should have no pins at all...

[Gunlance]

The blocks have pins to improve heat dissipation, removing the pins and getting better results makes no sense. Use your common sense.

By your conclusion a block should have no pins at all...

Harsh words from someone who hasn't given an idea yet. At least the OP is TRYING things.

I can think of at least 1 water blockw ith no pins and thats the mcw 30.

[WoZZeR999]

The blocks have pins to improve heat dissipation, removing the pins and getting better results makes no sense. Use your common sense.

By your conclusion a block should have no pins at all...

That isn't quite what the OP was saying. His opinion was that the pins were directing water away from the core. With water hitting the core, it should remove the heat from the center, with the slower moving touching areas with more surface area.

[Garrett]

lol nice find... I guess...
anyone willing to try it with a FuZion v2? :D

[TheThreeDegrees]

I put my hands up and say YES I botched the testing:(

[jas420221]

I put my hands up and say YES I botched the testing:(Less botching and more (complete) testing!! Thats all we ask! :)

[aspire.comptech]

Harsh words from someone who hasn't given an idea yet. At least the OP is TRYING things.

I can think of at least 1 water blockw ith no pins and thats the mcw 30.

The mcw30 is a chipset block.

Chipsets don't put out near the amount of heat a cpu does.

Therefore it really doesn't need the pins...

[Snyxxx]

I put my hands up and say YES I botched the testing:(

I admire your experimentation.:up:

[Sparky]

Dude I'd be tempted to do this if my whole PC loop wasn't at stake lol.

[MonkeyHood]

Interesting. I know this would be extremely hard to do, but maybe instead of removing the pins, you sharpen them? I realize the OP can't do that now, but if someone has a spare D-Tek base lying around, they could try making them pointy. More surface area than the OP's version, but less deflection.

It's still unknown though whether deflection has a insignificant impact (pun not intended) or not. :)

[HotGore]

If only I had the money to risk I would do it.

[TheThreeDegrees]

I'm back with an even crazier plan:D

Implants anyone?!:rofl:

The proc on the right is my first victim!

http://i21.photobucket.com/albums/b300/fabtrader/27052008414.jpg

Installed nicely and sealed up.

http://i21.photobucket.com/albums/b300/fabtrader/27052008417.jpg

http://i21.photobucket.com/albums/b300/fabtrader/27052008418.jpg

And the finished implant water leak tested and sealed very nicely.

http://i21.photobucket.com/albums/b300/fabtrader/27052008422.jpg


I am having difficulty seating/mounting it correctly in the mobo socket.
Does anyone have any suggestions on how to mount it safely in the socket with the correct amount of pressure applied?
Made a right mess of that E8500 up there on the right when I tried to seat it in the mobo.CPU is fooked and almost killed my mobo also.(this is before I removed the IHS):(
Experimenting with Celerons from now on:p:

Man I gotta give up this water cooling business,its costing me a fortune!:rolleyes:

[manyga]

You are crazy....hahaha

[Giannis86]

WOO THIS IS CRAZY!
iv seen direct ihs watercooling before but it didnt yield good results. (less surface area) ..i sure hope you have a better luck though

[requ13m]

i'm liking the look of this man, its nice to see boardsys on xs, especially doing what your doing! Good luck

P.S: i noticed your ad on adverts, if you want to split that up i'll gladly take the rad for a reasonable price ;)

[coolfris]

Recently sombody on procooling forum did something simulair with a custom made top. (here (http://forums.procooling.com/vbb/showthread.php?t=14751))
The results were not great, but maybe with the Fuzion top it will be different.

But no matter how the results will be, I really admire youre style of trying these things :up:

[cegras]

If you're crazy enough, drill small indents into the area above the die to improve impingement and turbulence. DO IT!

[evil-98]

wow.. couldnt you just went IHS-less instead of killing the dtek lol if you think about it, its the same and the fuzion has a higher surface area for heat transfer..

[nikhsub1]

wow.. couldnt you just went IHS-less instead of killing the dtek lol if you think about it, its the same and the fuzion has a higher surface area for heat transfer..
Yeah I'm wondering the same thing...

[Cptn Vortex]

That's awesome.........You should just aim that jet of water DIRECTLY onto the Die....put some silicon around the die, and hope it doesnt leak out around the board....

[TheThreeDegrees]

Thanks for the encouragment lads:up:


wow.. couldnt you just went IHS-less instead of killing the dtek lol if you think about it, its the same and the fuzion has a higher surface area for heat transfer..

Too late to try this now but when I buy another D-tec block for comparison I'll give it a go.


If you're crazy enough, drill small indents into the area above the die to improve impingement and turbulence. DO IT!

Tegras I take it you mean drill them on the IHS die area?Will give it a go.
I want to try it with the IHS and also without the IHS.
Do you think it would be safe to mount the chip in the FuZion block without the IHS.I mean its ok for water to directly impact the die area without the IHS?Think this is what Cptn Vortex meant.


i'm liking the look of this man, its nice to see boardsys on xs, especially doing what your doing! Good luck

Hi Requ13m,good to see you here also.
I got banned from boards for trying to help some guy out on a system he was buying.He had posted a shopping list that just didnt make sense and I gave my tupence worth and so fourth..so along comes a heavyweight and I ended up telling him where to shove his warning.Childish and stupid really.
But I miss Pog mo Thoin and L31mr0d.We had some great times learning and pushing our systems till they nearly caught fire:D.Anti comes over here from time to time also.
Still keep in contact with Chosen though.He's mad for those SSD's!

Yeah last weekend I was pissed off after killing the E8500 and almost my mobo so wanted to pack it all in and hence placed the advert.

Problem I have is that I only have 1 system in my place and this is it.So if I screw up i'm in deep :banana::banana::banana::banana:

[requ13m]

Hi Requ13m,good to see you here also.
I got banned from boards for trying to help some guy out on a system he was buying.He had posted a shopping list that just didnt make sense and I gave my tupence worth and so fourth..so along comes a heavyweight and I ended up telling him where to shove his warning.Childish and stupid really.
But I miss Pog mo Thoin and L31mr0d.We had some great times learning and pushing our systems till they nearly caught fire:D.Anti comes over here from time to time also.
Still keep in contact with Chosen though.He's mad for those SSD's!

Yeah last weekend I was pissed off after killing the E8500 and almost my mobo so wanted to pack it all in and hence placed the advert.

Problem I have is that I only have 1 system in my place and this is it.So if I screw up i'm in deep :banana::banana::banana::banana:

Yeah i try to aviod helping people on building and upgrading, half of them just want someone to come to their house to build them a pc. Chosen is just mad for computer hardware, the amount of money that guy spends is nuts. And don't worry about screwing :banana::banana::banana::banana: up just take my attitude "they're only things and they can be replaced":up: keep it up man

[Boogerlad]

excellent job. Can you tell us how you removed the ihs? did you use crazyxp2000's method?

[TheThreeDegrees]

excellent job. Can you tell us how you removed the ihs? did you use crazyxp2000's method?

Didnt catch his method but would like to know how he does it.

I used the blade method by rotating the blade under the IHS going in about 3/4mm deep.
Then I sat it on a hot plate on my cooker to heat/melt the solder a little and with a little pressure using the blade the IHS just popped off.

[Boogerlad]

here is crazyxp2000's method.


http://img.photobucket.com/albums/v320/CrazyXP1700/Computer%20Crap/Guide/DSCN0925.jpg

go around the IHS with the razor blade... some cpu's have resistor banks (like my 805) so only go in around .5cm... you just want to get the RVT silicone off

http://img.photobucket.com/albums/v320/CrazyXP1700/Computer%20Crap/Guide/DSCN0926.jpg

Put a razorblade in all 4 corners of the IHS...
now this is going to be used to do 2 things
1. it's going to suspend the PCB so that it will drop right out once the solder reaches it's melting point..
2. the razors are actually going to be applying all the pressure that you need to seporate the IHS from the PCB... so you shouldn't have to twist or apply any pressure to it... once it's warmed up... all the presure is just gonna shoot that CPU down... and the razors should be holding up the IHS still

http://img.photobucket.com/albums/v320/CrazyXP1700/Computer%20Crap/Guide/DSCN0928.jpg

place the razors holding the CPU over a vice, or something so the cpu can drop freely... now you'll want to put some foam or something like that below because it will literally POP off!

Core 2 duo's may be a little tricky... basically you've got a couple of these you've got to pop... i think warming up faster would be the best option... that way one core doesn't pop before the other... circular motions around the IHS would be the most ideal, the heat should transfer throughout... but still a fration of a second later could send one core popping off while the other is still firmly attached... (i havn't done one yet!)

http://img.photobucket.com/albums/v320/CrazyXP1700/Computer%20Crap/Guide/DSCN0929.jpg

I used a big ass propane torch, a fire maker, 4 razor blades, and a CPU to risk!

[aspire.comptech]

here is crazyxp2000's method.

I dont get what he means when he says a core 2 duo might be trickier, they still only have one physical core area.

It's the quads where you have two separate dies.

[Xilikon]

I dont get what he means when he says a core 2 duo might be trickier, they still only have one physical core area.

It's the quads where you have two separate dies.

Guess he goofed up with the description, it's a C2Q which is trickier (2 dies to take care).

[HotGore]

Nice! You are crazy! I doubt it will be better, even worse but it is worth a try.

[Duniek]

my modification of d-tek
http://img526.imageshack.us/img526/6064/82136610ny1.th.jpg (http://img526.imageshack.us/my.php?image=82136610ny1.jpg) http://img517.imageshack.us/img517/5724/70614244ua8.th.jpg (http://img517.imageshack.us/my.php?image=70614244ua8.jpg) http://img233.imageshack.us/img233/4844/42745234pw6.th.jpg (http://img233.imageshack.us/my.php?image=42745234pw6.jpg) :D

[aspire.comptech]

my modification of d-tek
http://img526.imageshack.us/img526/6064/82136610ny1.th.jpg (http://img526.imageshack.us/my.php?image=82136610ny1.jpg) http://img517.imageshack.us/img517/5724/70614244ua8.th.jpg (http://img517.imageshack.us/my.php?image=70614244ua8.jpg) http://img233.imageshack.us/img233/4844/42745234pw6.th.jpg (http://img233.imageshack.us/my.php?image=42745234pw6.jpg) :D

Smexy... :up:

[smee]

my modification of d-tek

Dude!! thats SICK!!! :shocked::shocked::shocked:

How does it perform compared to original top?

[Pedalmonkey]

nice and acrylic top. Maybe someone could uncap a CPU and mount the D-Tek directly to the cores. also if you machined the contact with the cores to be as close to the contour of the cores as possible. However i dont see a whole lot of gain in this. after decaping a CPU not much can be done to make the contact much more direct. Now is up to the pin design to create maximum heat transfer with maximum flow.

[H2omg]

Wow. Gnarly. I like it. Do more stuff please : )

[Duniek]

Wow. Gnarly. I like it. Do more stuff please : )

I show my all my stuff on polish forums

click right here (http://forum.purepc.pl/temat/Chlodzenie-Woda-i-Inne-Metody/33/Moje-Topy-Do-Blokow-Wodnych-I-Nie-Tylko/220767/)

[brinox]

isnt direct-die liquid cooling so inefficient at the pump speeds we are usually running at?

IIRC the water moving through/over the die needs to be moving rather quickly to be effective at all... on top of that, i would think there would be no gain unless sub-ambient temps are available.

[Duniek]

top for pro mount set like in V2

http://img48.imageshack.us/img48/243/21nt1.th.jpg (http://img48.imageshack.us/my.php?image=21nt1.jpg)

[TheThreeDegrees]

Duniek your work is good but its still only modifying what we already have in terms of blocks.

Personally I want to see more "direct die" cooling attempts.Move away from the tried and tested and explore the "Unknown"

Although have you tried running compressed air straight down one of those tops onto a die area?This been done?


Here I go.
About to begin "Project Implant":-)

I have just implanted a Proc into the base of the FuZion.The processor is not lapped and the IHS is installed as default.
Gonna run Orthos on it for an hour at stock settings.
Then after throw some voltages at it and see how it performs then.

After,
I will either drill some holes around the die area and do the same series of tests OR get right to the point and implant the proc into the FuZion without its IHS.


Let me know if I am forgetting something or if you wish for me to change my approach.

p.s I will close the window just to keep ambient steady:up:
good Luck and I blame all you crazy lot for making me do this:rofl:


http://i21.photobucket.com/albums/b300/fabtrader/30052008426.jpg

[TheThreeDegrees]

1.I'm just using distilled water in the loop.
2.I am not using any fans on the PA.
3.The purpose of this experiment is to determine whether direct die cooling is a viable alternative to using "inconsistant copper block mounting"options.


Here are some early results.
In about 7 days time I will have the completed results as I have to wait for the arrival of a "standard"Fuzion block.
Therefore I do not want to remove the IHS on this particualar cup untill Ive tested it on the standard fuZion.That way nobody says "ah well its a different revision cup"etc etc..

again let me know if I can improve on the actual test stage.


I had core temp running at the same time and it was consistantly displaying 15degrees less than what the charts are showing for cpu1/2.


So here is standard settings with an ambient of 23.5 degrees celsius.


(Proc implanted in the base of FuZion)No nozzle.

http://i21.photobucket.com/albums/b300/fabtrader/2008-05-31-00h11-CPU1.png

http://i21.photobucket.com/albums/b300/fabtrader/2008-05-31-00h11-CPU2.png

http://i21.photobucket.com/albums/b300/fabtrader/2008-05-31-00h11-CPU3.png

http://i21.photobucket.com/albums/b300/fabtrader/2008-05-31-00h11-VCore.png


And these with a slight OC.


http://i21.photobucket.com/albums/b300/fabtrader/2008-05-31-00h44-CPU1.png


http://i21.photobucket.com/albums/b300/fabtrader/2008-05-31-00h44-CPU2.png


http://i21.photobucket.com/albums/b300/fabtrader/2008-05-31-00h44-CPU3.png


http://i21.photobucket.com/albums/b300/fabtrader/2008-05-31-00h44-VCore.png

So now have to wait for the arrival of the standard block.
Then I will install the proc IHS LESS into the modified FuZion and see how it runs.

[Martinm210]

Wow...I'm not worthy. I was feeling pretty confident about lapping my processor and you're direct IHS cooling an E8500...that's awesome!!

Next up, I want to see somone solder a block base to the IHS...:D

Great work!!

[TheThreeDegrees]

Thanks Martin.You are my idol:up:

anyways I couldnt wait around for a new FuZion so I went ahead and...well I'll let the photos say the rest.

http://i21.photobucket.com/albums/b300/fabtrader/31052008430.jpg

http://i21.photobucket.com/albums/b300/fabtrader/31052008435.jpg

http://i21.photobucket.com/albums/b300/fabtrader/31052008432.jpg

http://i21.photobucket.com/albums/b300/fabtrader/31052008436.jpg



And the results:

With the IHS installed without the hole:
http://i21.photobucket.com/albums/b300/fabtrader/2008-05-31-00h11-CPU3.png

Water going direct to the core using a 5.5 nozzle:
http://i21.photobucket.com/albums/b300/fabtrader/2008-06-01-08h40-CPU3-1.png

direct to the core using a 3.5mm nozzle:
http://i21.photobucket.com/albums/b300/fabtrader/2008-06-01-09h56-CPU3.png


and to summarise:

Well looks as if going direct to the cores with water is not the way forward.
I even tried a stronger pump but no joy in matching the results I got from directly implanting the proc with its IHS intact into the FuZion.

As always Martin you are on the ball.Soldering a block base to the IHS could be fruitfull.

Bye for Now.

Fab.