Is This right? SwiftechMCW462-UHT Setup?

[charleybot]

I am in total shock! Before this morning I had a liquid cooled setup for about 2 months which consisted of a BE Cooling WB, MaxiJet 1000, 265gph Powerhead Pump/Tank Combo, a 7 3/4" x 7" heater core (1990 Ford Bronco) with a 120mm 34CFM fan, all on 3/8" tubing. You can see my setup in my sig. Ambient temp is 24-26C. This setup would maintain 21C at idle, and go no higer than 27C at full load running prime95 for about 2 hours. I wasn't happy with this, so I decided to give the Swiftech MCW462-UHT a shot. The new setup consists of the following:

Swiftech MCW462-UHT 226W TEC Assembly
Hydor L30 320GPH Pump
Mean Well S-320 Switching PS
Danger Den Reservoir

All of this is on 1/2" tubing.

I have to add that Swiftech does an AUSOME JOB with their WB'S, and this was a no-brainer to install. :rolleyes:

Here's what I get for results: -10C at idle :eek: and 2.5C full load running prime95 for 2 hours.

I was using a DigiDoc 5 to monitor the CPU temp, however since it does not read negative temps(It will get to 35C and then roll over to .01C and correctly work from there while running prime95), I switched to a CompuNurse and also hooked up another temp sensor I had laying around to make sure that the temps were correct. My first thought was that I put something in wrong!!

I'm not complaining, but is this right? I did not think that this setup would get temps like this. MY system is running with no problems, no crashes or lockups. Just wanted to see if anyone else has had results like this.

I could only imagine using a phase-change setup to cool the hotside of the TEC down more...

[I.M.O.G.]

wow... impressive

[nealric]

How much did ya spend on that cooling setup. (Looks awesome)

[Warlord2]

looks right to me! make sure you dont see any condensation, at those temps you should really insulate it good. oh yeah you might wanna upgrade to a bigger fan, 34cfm is horrible for a 120mm fan, get a 80cfm and see if your temps drop, of even put a house fan by it to see if it makes a difference.

good job!

[gwcz]

you put some of that non conductive silocone (n/s on spelling) in the plastic thing so you don't get corrosion right?

[charleybot]

Originally posted by gwcz
you put some of that non conductive silocone (n/s on spelling) in the plastic thing so you don't get corrosion right?

Hi,

Actually I have used Dielectric grease to fill that socket up, as well as put dielectric grease on the socket as well(The pinholes where the CPU sits)as well as put neoprene on the backside of the motherboard. Like I was saying in my first post, Swiftech does an ausome job, especially about proper insulation.

[charleybot]

Originally posted by Warlord2
looks right to me! make sure you dont see any condensation, at those temps you should really insulate it good. oh yeah you might wanna upgrade to a bigger fan, 34cfm is horrible for a 120mm fan, get a 80cfm and see if your temps drop, of even put a house fan by it to see if it makes a difference.

good job!

Thanks!

I'm going to try a house fan today and see if it helps any. I know that the 120mm fan setup I have now, I have also made a custom shroud for it. (Nothing like cardboard and ductape :D )

I think I have a 130CFM 120MM fan around here, have to look for it, if I remember right though, the thing was pretty loud.

[charleybot]

Originally posted by nealric
How much did ya spend on that cooling setup. (Looks awesome)

Well, for this setup it was just under $300.00, but WELL worth it. I was thinking about the Vapochill or the prommy boxes, but I just couldn't go that high, and I think that the prices on those things have to come down some first..

[Wicked Klown]

Originally posted by charleybot
Here's what I get for results: -10C at idle :eek: and 2.5C full load running prime95 for 2 hours.


Sorry to say but there is no way your getting temps like that. I have the MCW462-UHT and MeanWell PSU on my 1900+ at stock we put off the same amount of heat. Best I have ever done is 6C idle and thats with the computer right in front of an open window. My cooling set-up is a MCW462-UHT, MeanWell PSU, 86 Chevette heater core with shroud and 120mm 131cfm YS-Tech, and a 1250 Eheim.

[I.M.O.G.]

hmm it would seem we have a dilemma. the only thing i can toss in is that the bronco is a big a.s.s. car and the chevette is small, is it possible for the heatercores to be that much more efficient? i wouldnt think so especially with the difference in air being pushed across them. i have another question for both of you with the MCW462-UHT anyways... when you have that running under full load (or even without a full load) does your heatercore feel like its raising the room temperature near your computer? like does it make the entire room its in uncomfortably warm, or would it do so in a small computer room?

oh and i had another thought on his processor, dont they say the tbredA's run really cool? still a big difference, and sounds a little hard to believe, but im trying :)

[Warlord2]

well my only guess would be that Klownin79 is using on-die temps reading and charleybot is using a compunurse and doc, the on-die reads inside the cpu while hes reading the temps outside the cpu. I donno if that would make that much of a difference but who knows. Both are great temps, mounting pressure and thermal contact also plays a role in this, also some pelts are better thin others.

[Wicked Klown]

The air coming from mine is 30C idle and 31C to 32C load. As for me yes I'm using on-die but even with my Compnurse it said 24C. The 1900+ puts off 68 watts of heat where the 2600+ puts off 76 watts of heat. Reading on die will always give a lower temp. With his 2600+ overclocked he is putting off 96.38 watts of heat as for me right now I'm running at 150X12=1800 at 1.85 and puttin off 85.49 watts of heat. With that my current temp is 12C idle and 15C load.

[Warlord2]

maybe you guys can compare temps of your coldplates or something see if one is colder/hotter thin the other and by how much.

[Wicked Klown]

Originally posted by Warlord2
maybe you guys can compare temps of your coldplates or something see if one is colder/hotter thin the other and by how much.



I know with mine without touching the CPU I can get about 1/8" to a 1/4" of frost on it. Would give temps but the compnurse wouldn't go that far down.

[charleybot]

Originally posted by I.M.O.G.
hmm it would seem we have a dilemma. the only thing i can toss in is that the bronco is a big a.s.s. car and the chevette is small, is it possible for the heatercores to be that much more efficient? i wouldnt think so especially with the difference in air being pushed across them. i have another question for both of you with the MCW462-UHT anyways... when you have that running under full load (or even without a full load) does your heatercore feel like its raising the room temperature near your computer? like does it make the entire room its in uncomfortably warm, or would it do so in a small computer room?

oh and i had another thought on his processor, dont they say the tbredA's run really cool? still a big difference, and sounds a little hard to believe, but im trying :)

Hi,

When I have my setup under full load you can feel that the radiator is slightly warmer than room temp, however it doesn't make the room any warmer.

I spent the day yesterday re-doing my setup, even hooked up another compunurse temp probe on the CPU just to make sure, I am not going by the MB CPU temp sensor because they are not accurate once you start adding things like this.

My results are the same, running -10C at idle(over a hour), and at full load (Prime95 for 2 hours) my cpu temp is 2.5C.

[Wicked Klown]

Not to offend you in anyway but it's impossible for you to have -10 idle and 2.5C load. For the simple fact your putting off 96 watts off heat. Only way I can see you getting temps like that is either with a fridge or a broke compnurse.

[charleybot]

Originally posted by Klownin79
Not to offend you in anyway but it's impossible for you to have -10 idle and 2.5C load. For the simple fact your putting off 96 watts off heat. Only way I can see you getting temps like that is either with a fridge or a broke compnurse.

Hi,

No offense taken, i'm still doubting it myself however out of 5 different temp sensors I find it hard to believe that they are all broken. Something I didn't see in your setup, are you using 1/2" tubing? On your PS for the TEC, what voltage are you running it at?(there's an adjustment on the back where you can raise/lower the voltage).

I am at this point not going to be taking this apart again, it's been done correctly 3 times, and I'm not to keen on taking it apart again. I was thinking about hooking up a fridge setup, but even if these temps are off 1 or 2 C, I'm satisfied with it(which is a first for me considering I'm always looking for something better).

[Wicked Klown]

I'm using 1/2" tubing all around, my voltage is set half way between 12 and 13 volts. Aslo does your board support the on die sensor? I know at one time when using MBM5 I was getting temps of -123C but I knew there was no way I was getting those kinda temps.

[charleybot]

Originally posted by Klownin79
I'm using 1/2" tubing all around, my voltage is set half way between 12 and 13 volts. Aslo does your board support the on die sensor? I know at one time when using MBM5 I was getting temps of -123C but I knew there was no way I was getting those kinda temps.

The MB I have (A7N8X) does support the On-Die sensor.

[charleybot]

Just found out that MBM5 does not support the Nforce2 chipset as of yet.....

[Kato]

I just setup the same config as you guys, and now I realized that the below 0c we saw is the coldplate temp, not the cpu temp. No matter how close you put your compunurse probe near the die, it will still picks up the colder temp from the coldplate. From Asus probe, it reads the ondie temp, and mine is 18C, running 1.85V for vcore and F@H for 10 hrs.

[Wicked Klown]

I know with my compnurse it read 24C. For awhile that was right for my external temp but I'm watching the on die temps now.

[charleybot]

Originally posted by Kato
I just setup the same config as you guys, and now I realized that the below 0c we saw is the coldplate temp, not the cpu temp. No matter how close you put your compunurse probe near the die, it will still picks up the colder temp from the coldplate. From Asus probe, it reads the ondie temp, and mine is 18C, running 1.85V for vcore and F@H for 10 hrs.

Hi,

The problem with ASUS PC Probe is that it is not accurate, you are getting incorrect temp readings from it. If you installed a plain WB on the cpu, PCProbe would still be inacurate.

When I attached the temp probe to the CPU, it's put as close as possible to the DIE as possible, and is taped onto the cpu with special tape(the yellowish looking stuff). Now if I have it as close as possible to the DIE, and it's on the side of the DIE then how would it read incorrectly?
If you are reading the coldplace temp, then the sensor is not installed correctly. I've been through this numerous times, and my temps are as acurate as possible. Even though my MB supports ONDIE temps, there is no way to read them currently. I'm waiting for MBM to be updated because right now it does not work with the Nforce2 chipset that my MB has.

[Wicked Klown]

See the thing with the compnurse and other is that it reads the external temps. The thing is that the external temps will be 10C higher then the on die temps. While I type my external temps are 20C and the on die temp is 10C. I'm putting off 76.63 watts of heat which is less then you. One of the main reason my temps are that low is cause the computer is bout a foot or so from an open window if not my temps can get up to 15C to 19C idle. The lowest I have had so far was thew other night and that was 6C and the temps in side the house were 21.5C and case temps were 16C. So there would be no way your getting temps that low when putting off 96.38 watts of heat. Something has to be wrong and I hope you find out what it is.

[arkan]

Originally posted by Klownin79




I know with mine without touching the CPU I can get about 1/8" to a 1/4" of frost on it. Would give temps but the compnurse wouldn't go that far down.


if your getting frost anywhere then you dont have it insultaed right and thats causing youre higher temps.

[Wicked Klown]

Originally posted by arkan



if your getting frost anywhere then you dont have it insultaed right and thats causing youre higher temps.



Read that again please. In other words with the CPU in the box and just the pelt running like five feet between the two. Also the temps I'm getting aren't high and if you think they are you need to read more.

[qwerty57]

well if you cool the block with 5c chilled h20 or colder you can run like some Ive seen
2600+@2780mhz @ 1.95v
-3c load
-25c idle

so don't say its impossible. for ambeint 20c h20 yeah thats alittle much to believe, but lower the h20 temps enough and you could get some serious temps.

running my gpu mcw40-t modded with hotplate and 172w tec at -18c h20 temp's I ran almost -40c no load... take away the h20 to ambient cooled and it ran -10c yeah 30c from just h20 temps. :)

[I.M.O.G.]

12-20-2002 01:08 AM

[UselesKnoledge]

Both of you are not running the exact same system set up the exact same way, so it shouldn't even be close. Also, even if it was exact, it still wouldn't be the same temp because every component still has minor differences even if it is the same model #.