Swiftech MCW7000-T for Corei7 announced

[yup]

You need something like this .. This is what i used to put a 62mm TEC under a standard water block

This is exactly what i'm gonna do, but the hot plate's measure must be 11cm x 6cm, to have enough room for 2x OCZ blocks. Can you give me the depth of your cold plate?

@Chruschef: Of course i lap these block until they're flat.

[Ultrasonic2]

This is exactly what i'm gonna do, but the hot plate's measure must be 11cm x 6cm, to have enough room for 2x OCZ blocks. Can you give me the depth of your cold plate?

@Chruschef: Of course i lap these block until they're flat.

11 cm ? 6cm ?

Each block is just over 11mm thick ... and the step is 5mm .. so total hight above components when measured from the motherborad* is 13mm


* assuming the top of the CPU is 8mm from the motherboard

im making a better block than what's pictured right this second if you'd like one

[Ultrasonic2]

This is exactly what i'm gonna do, but the hot plate's measure must be 11cm x 6cm, to have enough room for 2x OCZ blocks. Can you give me the depth of your cold plate?

@Chruschef: Of course i lap these block until they're flat.

here's some measurements of my new blocks

Hot side

[yup]

11 cm ? 6cm ?

Each block is just over 11mm thick ... and the step is 5mm .. so total hight above components when measured from the motherborad* is 13mm


* assuming the top of the CPU is 8mm from the motherboard

im making a better block than what's pictured right this second if you'd like one

Each OCZ block measures 5.5cm, so the hot plate's lenght is about 12 cm, width is ~6 cm. Thanks for your pictures

[Ultrasonic2]

Each OCZ block measures 5.5cm, so the hot plate's lenght is about 12 cm, width is ~6 cm. Thanks for your pictures

What are you trying to cool ?

cos 6cm is to small to fit a 62mm TEC and 12cm is very long for one TEC but far to small for 2 TEC's?

[Ultrasonic2]

i just read back. ok you want it that large to fit to water blocks on ... i could make the hot plate 12cm long no problems .. it still need to be wider than 60mm though.. why / where does it need to be smaller


i came up with this in a minute ..would this work

[gabe]

As you can tell, we have run into a snag with our TEC vendor. The initial TEC I chose wasn't giving me satisfaction, so I spec'd another unit that can run at close to max efficiency on a 12v PSU. Problem is I am still waiting for the sample module...don't laugh! since this isn't a top priority, I haven't bugged my vendor on a daily basis as I usually do, and he slacked off. I'm back on it now, and hopefully should get the module soon so I can resume testing.

Sorry for the delay, and not posting updates. It's been quite busy around here lately!

[MulletBoy]

yay updates finally!!!

[zipdogso]

The initial TEC I chose wasn't giving me satisfaction, so I spec'd another unit that can run at close to max efficiency on a 12v PSU.

Oh right....so anything over 250w is pie in the sky, judging from this thread there is going to be some disappointed peps !!!!

[genec57]

It will have to be well over 250w. I have a 226 pelt hooked up to a 320-15 Meanwell. Just for grins I put in on my i7 920 overclocked to 4.0ghz under a really good water cooling system. As I was told to expect it promptly was overwhelmed by the heat of the 920. I had my alarms set at 75c and they went off at about 2 minutes and temps were still climbing rapidly. Without TEC the same water cooling holds temps at 4.0 to about 70c fully stressed.
If Gabe can come up with something at ~400 watts I will go for it. Much less and I don't think it will be worth it.

[zipdogso]

It will have to be well over 250w. I have a 226 pelt hooked up to a 320-15 Meanwell. Just for grins I put in on my i7 920 overclocked to 4.0ghz under a really good water cooling system. As I was told to expect it promptly was overwhelmed by the heat of the 920. I had my alarms set at 75c and they went off at about 2 minutes and temps were still climbing rapidly. Without TEC the same water cooling holds temps at 4.0 to about 70c fully stressed.
If Gabe can come up with something at ~400 watts I will go for it. Much less and I don't think it will be worth it.

The problem is Gabe said he was looking for efficiency at 12v...

Well half the world thinks the efficiency point of a TEC is 75% max power which might indicate he is planning to use a 15.6v TEC. A 15.6v TEC is a 127 couple unit and there are some hefty ones about especially as it was hinted it was a custom unit. To be bigger than 275w at 75% power it will have to be at least 350w + Qcmax so it will be at the limits of 62mm 127 couple design I believe. You certainly wont get the 400w your looking for....well not in my estimation anyway - that's why I thought his original posting was very hard to believe. I think it was a brave thing to do to make an announcement like that when at that point he didn't even have a TEC to test let alone consider how he was going to power it- -I just hope that hasty decision does not impinge on his business too much !

He could be using a TEC with a higher couple number especially as it was hinted it was custom but the undervolting efficiency point is the COP efficiency point of around 30% max power but to get 275w at this point is just ridiculous. Not least the average Joe bloggs wil have difficulty getting the 20ºC delta needed to reach that efficiency anyway.

So he could use a TEC with a large couple number and carefully choose the undervolting to 12v but he will be a way off the so called "max efficiency" he was referring to. Besides I wouldn't to run a 400w TEC at a COP of 0.5 - 0.75 (which is what it will be in this scenario.) for any length of time not with the massive cooling system that be required with it.

So what was Gabe talking about ?

It really will be a magic TEC.....

To be honest I think single TEC direct die cooling has nearly had it's day due to todays high TDP's a multi-TEC cooler is a much better proposition - Coolit's offerings are about at their peak due to their use of molex connectors for power so all that remains really is self-builds - with too few people around who know enough about TEC's to implement it effectively.

[Ultrasonic2]

i don't see why you'd want to run the TEC at 12volts anyway? i mean your gonna have to buy a new psu of this thing any way what difference does it make if it's a 12volt one or 24?

[yup]

Hey guys, what thermal is the best for TEC? AS5 or AS ceramique?

[zipdogso]

i don't see why you'd want to run the TEC at 12volts anyway? i mean your gonna have to buy a new psu of this thing any way what difference does it make if it's a 12volt one or 24?

Your quite right...but I am thinking his is concerned at the cost of a PSU to run his unit. He will sell more if he can keep the PSU cheap but as I already said unless his TEC can pack some oomph there will be a mighty lot of disappointed peps here. He may ultimately NEED to use a 24v supply.

[genec57]

It has been awfully quiet for a while now. Has anyone heard anything from Gabe. I would love to see an i7 strength TEC.

[antiacid]

I just hope that hasty decision does not impinge on his business too much !

Well, besides the 15-20 folks who've read about this project here (and I hope everyone understands that it is in development), I doubt that there's much harm done. If anything, he was testing the water by announcing it and seeing what kind of response it would get :P

Besides, he stated that it wasn't a top priority anyway for his business hehe


Now the real question is: Why not plan ahead for 32nm since core i7 cooler market is already well saturated? It seems like the best move would be to plan ahead for next gen and really look sharp when you can do a simultaneous launch.

[Voodoo Rufus]

I'm watching this thread now, as well. If a decent PSU and power requirements solution could be worked out, this could be interesting.

[TopFuel1471]

Gabe keeps rolling out the goodies. First the MCR420, now this. It's like christmas for overclockers.

[NaeKuh]

i don't see why you'd want to run the TEC at 12volts anyway? i mean your gonna have to buy a new psu of this thing any way what difference does it make if it's a 12volt one or 24?

i think people still have there heads in the clouds on hoping they can plug a 437W TEC on a 12V molex cable from your PSU.

Im waiting for a fire to happen, and ultra u know exactly what i mean. :P

How oxymoron is that? Your liquid cooled computer caught fire.

[flak-spammer]

I wouldn't be surprised if it actually happened, I've warned some large number of people about the PSU issue and I'm sure I havn't even skimmed the surface. I suppose if people wish to be stupid it is their right to do so.

[duckbillclinton]

idle in windows @ -10°C
Full load on 8 treads @ +28°C and this 24/7.

Yeah the load numbers sound a little low, I ran a artic web block with a 437W 24V TEC with a q6600 G0 stepping and was in the 20s at load at stock speed and 45C at 3.9GHz.

ddc/xspc top > black ice gtx360 6 high panflos push/pull (25mm shroud both sides)> ddc/xspc top > Artic Web

I am converting to a 527W 16V and Q9650 probably by the end of next week. I should be able to tame Q9650, but a i7 is a little different. I doubt the swiftech block stands a chance.

1. You should probably check the output capacity of your 24V DC supply. For a 437W TEC, you will need at least a quality 400W DC PSU or better a 450W one. Also, use some meter thingy to ensure your TEC do indeed is draining 400+ wattage. (There might be also a small chance that you got an over marked TEC.)

2. Please also check the temperature on your TEC's hot side. From what you wrote, you may not have sufficient cooling on the hot side since you indeed, need to dissipate as much as 437W of heat.

The 437W TEC you used I am assuming it specs as 24V 18.2A, then the cooling capacity is 437 x 0.6 = 262.2W.

In order to produce 262.2W of heat, your Q6600 will absolutely need to use more than 262.2W electricity to produce that, since no chips would have 100% efficiency on heat conversion. (If so, we would probably use all those fail-to-meet spec chips from Intel/ AMD for very efficient heaters. It is just like if you have 1 lb of orange, you can never produce more than 1 lb of juice, please don't use the add water thing as a joke:P)

So how much electricity can your Q6600 use at 3.9 OC? Well, if referring to Intel's technical paper on core 2 quad processors (doc number 315592-005, table 2, 4) and ATX specs, if running your CPU at 1.6V (ouch), you will use at most around 325W electricity (which will never ever ever happen!) Consider 80% of that turned to heat, then your Q6600 ended up producing 260W of heat (TDP). What about the other 20%? Well, that's the computing power you gain, right? In reality, your Q6600 produce far less than 260W TDP.

From the look of your TEC's spec it should be able to handle your processor easily and should have achieve far lower temperature when system is loaded. With your TEC's hot side being cooled around 60C, you should be able to get near zero when full loaded.

Currently, assuming if you are truly able to dissipate 437W of heat successfully from the hot side, then according to your CPU's reading of 45C, with 60C delta added to TEC's hot side, that's 105C... It's just insane, something does not add up here. So please check 1 and 2 mentioned above.


On the other hand, I have checked ChillerZero's work log on the site he listed on his post (25+ pages, now he's into phase change cooling, lol) All his numbers and results looked very legit to me. He was close to find the holly grail, but he missed a number of important factors. Also, his post didn't provide enough data for analysis, especially room/ ambient temperature, TEC hot side temperature, TEC's actual specification (need label printed on the plate), his CPU's default voltage (when mobo posted w/o OC). But then, it will be too much for me to write, lol

DBC

[flak-spammer]

Actually the 437w he's talking about is a 437w Qmax (50C) or 360w Qmax (27C) so we must be slightly careful about how we tackle this the PSU has to be closer to 700w to actually give the full 360w Qmax and even then you won't get the best performance of any TEC.

[duckbillclinton]

Actually the 437w he's talking about is a 437w Qmax (50C) or 360w Qmax (27C) so we must be slightly careful about how we tackle this the PSU has to be closer to 700w to actually give the full 360w Qmax and even then you won't get the best performance of any TEC.

You are not exactly right, actually a 600W PSU can do 360 Qmax just fine with efficiency of 0.6. Now days, a 600W PSU can handle 600W or overload a bit I believe. Also, there's really no way you could find a TEC rated 16V with 527W Qmax (mountable on any CPU) from the post I quoted. So I believe the 437W he (Bond Number) referred to was power usage but not Qmax.

On the other hand though, it is absolutely possible to get a 24V DC PSU rated at 600W or even 1000W (just look for communication/ surveillance equipments). The only problem here is money... We could however though, pair up two 320W DC PSU together for 600W , lol.

From my previous post, the point I was trying to make was, with ChillerZero's 24V 320W DC supply, at most Qmax he could pull is 192, and with that 192W, he was able to cool his i7-920 to near 30c full load w/ OC close to 4 ghz, so Gabe's proposal is DEFINITELY doable.

Lots people on this forum claiming i7 pulling 300W TDP OCed at 4 ghz is just way off the proportion. That's why I ran through Intel's tech paper and ATX spec for the actual answers (by looking at Vmax and Imax allowed).

In fact, an i7 OC to 4.7 ghz with 1.5V will generate at most 298W TDP, and 5.0 ghz with 1.6V will get 340W. So how many would actually hit 4.7 ghz or use 1.5V above?

DBC

[Ultrasonic2]

Im still surprised that no one credible has tried a 62mm TEC on a i7 yet ?

In theory ChillerZero's results are impossible. But thats based on the CPU load calculators being correct. Which they may not be. So for me im still waiting for someone else to give it a try. I've got a 62mm TEC and water block pitty i dont have a i7 id be happy to sell it too

[flak-spammer]

duckbillclinton, I suggest you read a bit more on TEC's before making claims you are unable to back up. A 360w Qmax TEC requires nearly 600w of power to run at 360 Qmax. Assuming the voltage is lower than Umax then you no longer get the full 360w Qmax. As muffy stated these results just don't make any sense.