my new watercooled setup

chylld · Post by **chylld** » Sat Feb 21, 2004 3:07 pm

case stays upright

rotating the radiator 90/180/270 degrees will require drilling new holes and i don't have the time to go to such measures. it was a shortsight i should have fixed when i was planning out the system but for now i'll just live with it.

i've now narrowed down the cause of the highish temps to 4 good reasons:

1) temperature reading is the cpu socket temp which receives next to now airflow with a wc setup, hence it reads a higher temp (this is explained well on the swiftech site)

2) crappy cpu waterblock and mounting.

3) too small radiator - will upgrade to a heatercore later.

4) order of tubing: should be pump -> radiator -> cpu, not pump -> cpu. The pump is putting a significant amount of heat into the loop.

edit: not really 'narrowing it down' to these 4 reasons as such - more like, these are the 4 reasons why it isn't performing well.

peteamer · Post by **peteamer** » Sat Feb 21, 2004 3:28 pm

Umm.. You dont have to redrill holes etc... just let it (the rad) 'hang' to get rid of the airlock.

Your other 4 reasons whilst holding water,

I think are.... only to be followed,
once your sure you have no airlocks in your rad ,..like the other posts have shown concern over the same thing.

Besides... if you prove that.. you can say... "Pete... Go back to sacrificing goats"

yours humbly(?)

Pete Amer

chylld · Post by **chylld** » Sun Feb 22, 2004 2:09 am

how do i just let it 'hang' there? it's pretty heavy u know

peteamer · Post by **peteamer** » Sun Feb 22, 2004 2:17 am

Ummm.. er...

Hand hold it for a few mins?

Or my favourite... bits of string

or something to prop it up temporarily.

Stack of AOL cd's, everyone must have enough of those

Good luck.

Hope it does what we/you want.

Banish that evil air

Pete

Gooserider · Post by **Gooserider** » Sun Feb 22, 2004 6:07 pm

Chylld:
case stays upright

I think the idea was not to tilt it long term, but just for a few minutes to get rid of the suspected airlock in the rad. You could probably do pretty well just by picking it up and tilting. I would probably lean the case back onto it's backside. If you have the res filled all the way then you should be OK with doing a momentary tilt. If you have as big an air bubble as I expect, the res may end up full of air, and you will need to repeat the fill and tilt drill a couple of times, but that will GREATLY improve your cooling. If you have the size bubble I suspect you may be loosing 50% or more of your possible flow.

rotating the radiator 90/180/270 degrees will require drilling new holes and i don't have the time to go to such measures. it was a shortsight i should have fixed when i was planning out the system but for now i'll just live with it.

True - somehow it always seems to take longer to fix it than it does to have done it right in the first place... :p However I don't know that it would take THAT much work to just do the rad flip. I agree that other fixes would be harder, but I don't see why that one would be a big problem.

i've now narrowed down the cause of the highish temps to 4 good reasons:

1) temperature reading is the cpu socket temp which receives next to now airflow with a wc setup, hence it reads a higher temp (this is explained well on the swiftech site)

Possible, no real way to tell without some form of better temp readout.

2) crappy cpu waterblock and mounting.

Possibly the mounting - though the block itself is supposedly pretty decent. Is there any way you can bend anything on the mount to make it press a bit harder?

3) too small radiator - will upgrade to a heatercore later.

The small rad isn't helping, but I suspect the vapor lock is more of a problem. Just as a quick thought, if you turn off just the fan, does the rad get warm? If not, there is definitely a flow prob with the rad, it should get just as warm as any other part in the system without airflow to make it cold.

4) order of tubing: should be pump -> radiator -> cpu, not pump -> cpu. The pump is putting a significant amount of heat into the loop.

DEFINITELY NOT!!! The entire 'component order' thing is a myth generated by poor thinking and the older badly designed low flow systems. In any system with a decent flow rate (i.e >1/2 GPM (1-2 LPM) there will be less than 1*C difference between the highest and lowest temps in the loop, and there will be no significant difference in CPU temp due to component sequence.!

I repeat the oft made assertion that your biggest single problem is the air bubble that nearly everyone that has posted on this thread thinks is present based on your pix and the rad mount that you used.

Gooserider[/quote]

chylld · Post by **chylld** » Sun Feb 22, 2004 6:20 pm

Gooserider wrote:
Chylld:
case stays upright
I think the idea was not to tilt it long term, but just for a few minutes to get rid of the suspected airlock in the rad. You could probably do pretty well just by picking it up and tilting. I would probably lean the case back onto it's backside. If you have the res filled all the way then you should be OK with doing a momentary tilt. If you have as big an air bubble as I expect, the res may end up full of air, and you will need to repeat the fill and tilt drill a couple of times, but that will GREATLY improve your cooling. If you have the size bubble I suspect you may be loosing 50% or more of your possible flow.

Sounds like a plan. i'll do it right after i post this.

Gooserider wrote:
i've now narrowed down the cause of the highish temps to 4 good reasons:

1) temperature reading is the cpu socket temp which receives next to now airflow with a wc setup, hence it reads a higher temp (this is explained well on the swiftech site)
Possible, no real way to tell without some form of better temp readout.
2) crappy cpu waterblock and mounting.
Possibly the mounting - though the block itself is supposedly pretty decent. Is there any way you can bend anything on the mount to make it press a bit harder?

If it presses any harder, the computer shuts down. the cpu core is cracked already, thanks to this waterblock. call me unco, but other people have had mcw5002 problems as well.

Gooserider wrote:
3) too small radiator - will upgrade to a heatercore later.
The small rad isn't helping, but I suspect the vapor lock is more of a problem. Just as a quick thought, if you turn off just the fan, does the rad get warm? If not, there is definitely a flow prob with the rad, it should get just as warm as any other part in the system without airflow to make it cold.

If i turn off the fan, the cpu temp goes up several degrees very quickly indeed. i figure that this means the rad is doing something...

Gooserider wrote:
4) order of tubing: should be pump -> radiator -> cpu, not pump -> cpu. The pump is putting a significant amount of heat into the loop.
DEFINITELY NOT!!! The entire 'component order' thing is a myth generated by poor thinking and the older badly designed low flow systems. In any system with a decent flow rate (i.e >1/2 GPM (1-2 LPM) there will be less than 1*C difference between the highest and lowest temps in the loop, and there will be no significant difference in CPU temp due to component sequence.!

i've heard people say this many times but i still do not understand why it is so.

actually i just thought about it for a few seconds and now i understand.

chylld · Post by **chylld** » Sun Feb 22, 2004 7:05 pm

okay i just finished tilting.

tilted it almost 180 degrees backwards (i.e. case was almost entirely upside down) and it made no difference with the pump on or off. water level in the res is the same, as are the temps.

the top of the radiator gets quite warm with the rest of the radiator though

m0002a · Post by **m0002a** » Sun Feb 22, 2004 7:57 pm

Wow! Even IBM mainframes are now air-cooled.

chylld · Post by **chylld** » Sun Feb 22, 2004 8:35 pm

m0002a wrote:Wow! Even IBM mainframes are now air-cooled.

??

m0002a · Post by **m0002a** » Sun Feb 22, 2004 8:59 pm

What What. IBM mainframes were once water cooled, but are now air-cooled (although the room has to be air-conditioned). I am wondering if water cooling is a bit extreme for a PC.

chylld · Post by **chylld** » Sun Feb 22, 2004 9:25 pm

right, i don't know much about mainframes so i didn't get what you were saying at first. (i assumed you meant to type 'water-cooled').

water cooling is extreme and it is by no means necessary. those of us that do it do it for fun, and often we have good reason to watercool as well (higher stability when overclocked, and silence, at the same time.)

having said that, processor heat output is constantly on the rise - watercooling is simply a way to keep ahead of ever-toughening cooling requirements - and in style too, i might add.

peteamer · Post by **peteamer** » Wed Feb 25, 2004 11:26 am

Any news or progress chylld?

Regards

Pete

chylld · Post by **chylld** » Wed Feb 25, 2004 2:26 pm

well it's been raining cats and dogs here in sydney for the past few days so the ambient is down a few degrees - as such cpu temp has now dropped to 49-51C.

i think i need to get my hands on a temperature probe to get accurate cpu temp readings.

but yeh recently i've been real busy and actually need to use the computer - no time to fiddle with it.

peteamer · Post by **peteamer** » Wed Feb 25, 2004 2:40 pm

Thats a fair sized temp. drop.

I know my air-cooled main rig varies between
CPU:48C-53C
Seagate V 80GB: 38C-42C
System: 31C-36C

For room temps between: 16C-22/23C.

Any idea how much your room temp. is varying?

Pete

chylld · Post by **chylld** » Wed Feb 25, 2004 3:03 pm

room temp is about 20-22c now.

Seal · Post by **Seal** » Thu Feb 26, 2004 12:03 pm

chylld i remember on my first setup cpu temps read about 15 deg c higher than they actually were because there are no longer any fans around the cpu area and the still block of air where the cpu probe was made the temps read higher than they were untill i placed a fan over that area... just something for you to try really

chylld · Post by **chylld** » Thu Feb 26, 2004 1:48 pm

yeh.. short of physically placing a temp probe next to the cpu die, is there any other way to accurately measure cpu temps?