when considering the performance of a water block...

Albigger

Member
Is it sufficient to only take into account the water temperature (assuming at full load, water temp is constant)?

Since87

Member
Water temperature is greatly dependent on the radiator and the airflow through it.

So, water temperature alone is not a good measure of a blocks performance.

Liss

Member
wut bout the water right wen it comes out of the outlet.. ???

OP
A

Albigger

Member
I can see how the ambient air and cfms through the rad. affect the radiator's ability to cool the water, but, for a given water temperature, how would this effect the block's ability to cool the cpu.

sure the block's ability to cool depends on the temperature of water going through it, but, making relatively simple assumptions (that the cpu dissipates most heat into the block and not into the air, and that the block dissipates most heat into the water and not into the air, so long as ambient temp is relatively close to the block/water temp) I don't see how ambient temps affect it.

obviously, if there is colder water going to the block, the cpu temps will be lower. but as far as a block's ability to cool based on the temperature of water going through it - this is not doable?

OP
A

Albigger

Member
Liss said:
wut bout the water right wen it comes out of the outlet.. ???

ok, measuring inlet and outlet temperatures is fine. I just wanted to know if it could be done (maybe having to insulate block from air) to find a relationship between block and water temp. only

frodoski

Member
Given a constant ambient air temperature, your system will achieve an equalibrium temperature. This temperature should remain constant.

take the example of my system. The ambient temperature at the face of the radiator is 19.6 degrees C (Celcius), the ambient air temp of the room is 21.2 degrees C, the water temp is 20.4 degrees C. These temps have been constant. The CPU temp as reported by Mother Board Monitor has been a constant 36 degrees C, while running [email protected] full time while I do my regular work/goofing off/stuff.

Does this help?

Since87

Member
If the you have a water chiller keeping the water at a constant temperature, (say 25.000C) then knowing the output temperature in combination with the flowrate of the water would be meaningful information.

In a typical block->pump->rad->block loop, the water temp is combination of; the blocks ability to move heat from the CPU to the water, and the rad/fan's ability to move heat from the water to the air.

OP
A

Albigger

Member
ok so your cpu temp is 15.6 degrees higher than your water temp. is there a way to judge how "good" his water block is by this info??

I'm mainly asking because my rad. is at the top of my case, and the system temp reported by my mobo, can fluctuate quite a bit, and I wouldn't consider it a good controlled environment for testing.

however if my water temp stays fairly constant, i'm wondering if I can just measure this and it would be a better way to tell how the block is performing? But if this doesn't work then I won't bother...

OP
A

Albigger

Member
Since87 said:
If the you have a water chiller keeping the water at a constant temperature, (say 25.000C) then knowing the output temperature in combination with the flowrate of the water would be meaningful information.

In a typical block->pump->rad->block loop, the water temp is combination of; the blocks ability to move heat from the CPU to the water, and the rad/fan's ability to move heat from the water to the air.

so if I measure the water temp and it is fairly constant then if i knew the output temp. and flowrate that would be meaningful?

or still no because the other components of the system would be adding some 'error' effect??

how about temporarily hooking up a hose to the sink and running water straight through just to get some measurements like you suggest in the top part of your post?

Since87

Member
frodoski said:
The ambient temperature at the face of the radiator is 19.6 degrees C (Celcius), the ambient air temp of the room is 21.2 degrees C,

How is this possible?

OP
A

Albigger

Member
Since87 said:

How is this possible?

i was wondering also, but didn't want to get off topic.

Since87

Member
Albigger said:

so if I measure the water temp and it is fairly constant then if i knew the output temp. and flowrate that would be meaningful?

or still no because the other components of the system would be adding some 'error' effect??

how about temporarily hooking up a hose to the sink and running water straight through just to get some measurements like you suggest in the top part of your post?

Have you read the flowrate sticky? That, combined with everything BillA has written, probably covers every significant issue involved.

OP
A

Albigger

Member
Since87 said:

Have you read the flowrate sticky? That, combined with everything BillA has written, probably covers every significant issue involved.

ok sorry, i was just wondering if it could be done or not. i'll try to look up some more research on this later.

I HAVE read the entire flowrate sticky, but i think I need to read again, as that was a couple months ago and i probably forgot most of it...

frodoski

Member
The water temp was measure at the bleed tee, prior to the radiator. I have not inserted a tee after the radiator to monitor the water exiting the radiator.

My set up is Pump->Block-> Radiator->Pump.

To be honest I expected better temps with the water cooling. The most visible benifit is a smaller spread between full load and idle (1 or 2 degrees C).

OP
A

Albigger

Member
well, your temps, 36C full load (16 C above water temp, or more, after it goes through the rad) do seem kinda high for a 1.55 volt celeron....

frodoski

Member
Albigger said:
well, your temps, 36C full load (16 C above water temp, or more, after it goes through the rad) do seem kinda high for a 1.55 volt celeron....

Since87

Member
Albigger said:

ok sorry, i was just wondering if it could be done or not. i'll try to look up some more research on this later.

I HAVE read the entire flowrate sticky, but i think I need to read again, as that was a couple months ago and i probably forgot most of it...

LOL, don't worry about it. I'm far from expert on this stuff, my first and second posts are about as far as I can go without getting in over my head. You might try asking the same question over at procooling. Many more experts on this subject hanging out there.

frodoski

Member
Brother, I make no claim to expert status!! I merely am reporting my current experience and results.

gone_fishin

BandSaw King
Ok, take the water temp and the cpu temp when at full load(use an app such as K7 or cpu burn-in, they are much better than seti). Subtract the water temp from the cpu temp (on die is much more acceptable than socket thermisters) to get the delta (difference). Now you must know the wattage (how much heat you are moving) in order for this information to come together. This is all assuming that the rad air intake temperature remains constant so that the system can reach equalibrium. Divide the wattage into the delta and you get a C/W number. This number is useless to other people's systems (because of variations from system to system) but if you were to take another block on your same system and do the same procedure, you can get a rough comparison (both blocks should be equally lapped at the base). This is only after mounting each of the two blocks numerous times and recording at each mounting. The reason you need to do multiple mountings is that the tim joint will never be the same between any two given mounts. If you have multiple mountings and take the best reading from each set of data for each block, you are at least attempting to minimalize the variation in the tim joint (which can swing your readings +-3 to 5C).

Recording the ambient rad air intake temp along with it will serve to validate the readings (most people who do this often can spot BS a mile away if given the full set of data.)

When you are done with your comparison, it will only be good for the specific flowrate induced by the pump you are using because blocks performace changes along a curve with flowrate. Each blocks performance curve is likely to be different and you will not know this curve unless you test with different flowrates.

Last edited:
OP
A

Albigger

Member
gone_fishin said:
Ok, take the water temp and the cpu temp when at full load(use an app such as K7 or cpu burn-in, they are much better than seti). Subtract the water temp from the cpu temp (on die is much more acceptable than socket thermisters) to get the delta (difference). Now you must know the wattage (how much heat you are moving) in order for this information to come together. This is all assuming that the rad air intake temperature remains constant so that the system can reach equalibrium. Divide the wattage into the delta and you get a C/W number. This number is useless to other people's systems (because of variations from system to system) but if you were to take another block on your same system and do the same procedure, you can get a rough comparison (both blocks should be equally lapped at the base). This is only after mounting each of the two blocks numerous times and recording at each mounting. The reason you need to do multiple mountings is that the tim joint will never be the same between any two given mounts. If you have multiple mountings and take the best reading from each set of data for each block, you are at least attempting to minimalize the variation in the tim joint (which can swing your readings +-3 to 5C).

Recording the ambient rad air intake temp along with it will serve to validate the readings (most people who do this often can spot BS a mile away if given the full set of data.)

When you are done with your comparison, it will only be good for the specific flowrate induced by the pump you are using because blocks performace changes along a curve with flowrate. Each blocks performance curve is likely to be different and you will not know this curve unless you test with different flowrates.

...awesome. thanks.

conclusion: reliable and meaningful testing is a PAIN!

Replies
0
Views
136
Replies
7
Views
447
Replies
3
Views
286
Replies
28
Views
916
Replies
21
Views
908