Cooling Help me people ...

O.k. .... today I installed my system into my new modded case plus I switched the Waterblock from a copy of the swiftech MCW462 to a modified MC 462 A with 1/2" tubing all the way. Now what I don't get, is that I still have the same temps, although the rad has a better position with more air being sucked through the fins and the Waterblock should be more efficient. I've tightend the screws as far as possible, so what could be causing the rather unpleasing result ? Any ideas ? Thanks ...

Cheers

[DE]FreD_S

 

what pump are u using? the faster the flow the faster it can transfer the heat way from ur block

i can't comment on the swiftech cos i've never used one but i asume i'm using a inferior design .. old Z3 .. and i now get 3.9c temprature diffrence between water temp and cpu temp ...... all 1/2 fittings .. using a ehiem 1250 tho

what pump?

 

Your readings are not always the true fact. Its more accurate to use a probe directly on the block.

Top Tips:

1. Draw cool air through you rad and place the fan about 30mm away with some sort of shroud. This will prevent any dead spots and will be more efficient.

2. Water-Wetter will act as an inhibitor and coolant and will also help the cooling process.

3. The 1250 Ehiem pump is design for 1/2" systems.

Don't be to surprised with the temps there is a little science to it and above all its quiet!

Keep tinkering, it's all part of the fun.

 

Thanks guys ... I'll take a look at that water wetter stuff ... anywayz I tightend the screws a bit more today ... I suppose having 1.2 Kg hanging on your CPU you need to pull em tighter than normal - That got me 4°C ... I'm down from 44 to 40°C on an XP 1800+ now ... let's see what water wetter can do hehe ...

Cheers

[DE]FreD_S

 

That is quite high for a xp1800. My watercooled xp1800 are 8 degree C over ambient at full load.

DD maze 2, DD Chipset and DD GF3 block in a 1/2" system.
DD heatercore rad and Eheim 1048 pump.
Radiator cooled by a 120mm YS tech fan running at 7V.

 

with regards to the temps, the only accurate way is to attach a behing the core prob or start soldering to the back pins on the mobo. Though you are only getting an indicator, it will be enough. The adviced method is to gently tighten the screws until you gain no further decrease in temp. Too much and core goes crack too little and u get wopping temps.

That is down to u i'm afraid.

Water wetter will not give you improvements in temps, maybe a degree, but you do need to use either it or purple ice. I have purple ice which i prefer.

The eheim is designed for 3/4" hosing on ponds and fountains. You therefore don't want to decrease it to less than 1/2".

What is then ambiant temp where you are, if the room is 30c then the temp is fine, if its 24 like in my house, then it is a little high.

 

Well ambient is about 30°C in my room ... my room's directly under the roof, that's problem :-( I now have a cpu temp of 48°C according to MBM 5. I'll connect the MCW462 replicant to the system tomorrow to compare the performance of both blocks ... I have this funny feeling the replicant is gonna score more here ... let's see then

Cheers

[DE]FreD_S

 

well, 10c above ambiant is not totally sad to be honest. You get lots of, i run 4c, but very little evidence ever appears when questioned.

If I remember you mahe that masive 1 kg block. It should be getting better results that that though.

What else are u using

 

Here's the line-up : "that massive block" ;-) , a mora GA, an aquatube and an eheim 1048. I've got a 120 Enermax fan running on lowest speed on the rad, sucking air through it. I hope I'll get that damned USB Cable this week so I can post pics ...

Cheers

[DE]FreD_S

 

yup anything+15c over ambient is when u know u have a problem......try uping the fan to medium...might c better results.....also i saw that you are sucking air through the radiator....does that mean ur sucking case air through the radiator???

 

a good figure to work on is the C/W rating of your block/system.

That it delta (cpu temp - ambiant) / cpu watts = C/W.

The lower the figure the better. Yours is about 0.15 (the lower the figure the better.)

You do need a better means of measuring than the mobo diod.

But it means that if you overclock your cpu to 2gigs it will therefore stick out 98.20watts x 0.15 = 15.34c above ambiant your system should run at. so in your room at 30c it will run at 45.5c.

Your system is producing reasonable figures. A C/W figure is good as it give a figure that you can compare with other systems with colder ambiant and cpus.

Me got carried away i think.

 

I would guess the problems will be with the rest of your setup. The mora and aquatube both have 10mm bores and you are running 1/2" tubing. I think that is correct.

 

feel like im back in math class

 

Hmmm, sounds reasonable ... that's also a reason why I think the other block will do better, as it has 10mm barbs on it ... I could also put 10mm barbs on the MC462mod ... I still have some lying around with 1/2" threading ( google word for German "Gewinde" ) I hope it's the correct word. I'll try that first - unmounting a 1.1 Kg waterblock isn't exactly easy the way I have my system built together, hehe

Cheers

[DE]FreD_S

 

Not to seem wierd or anyhting but 40°C with watercooling, I don't have watercooling and I've got a my 1900XP down to 44°C and I'm using a crappy £8 heatsink but I have a 2 80mm ex fans and 1 120mm blowhole.

Is watercooling worth it?

 

If you do it right - yes, definetly ! It's much quieter in any case ...

Cheers

[DE]FreD_S

 

its a shame because i though your block is great.

I would go looking for car heater core (vauxhall cavalears are great) a tube res from ocpc and upgrade to 1/2 through out. But the mixing of tube sizes will hammer your system. Also the eheims dump about 10watts of heat into the water, which could well use a good heater core to remove.

 

Also how are you measuring the temps. You cannot compare water reading with air on a modo reading. With my water system i could turn an 80mm fan on the socket and get 2c below ambiant readings.

As to is it worth it. Without doubt.

 

If water-cooling was that simple everyone would be doing it. I'm sure in time to come it will be the standard as air is now.

Until those happy days we can only investigate document and experiment until everyone is convinced.

The only way to know the real scores on the doors IMO is a good probe! I do like the Maplins one reviewed here.

http://www.bit-tech.net/review/95/

 

The flat probes are the best. It also depends where you stick the probe. If you place it beside the core touching it you will get a better result than mobo reading, but for real results you need a behind the core mounted one. That requires cutting through the socket ziff so is only for the brave, however I have done it and it was not that bad.