new homade cooler

This is a report on what often is known as a ghetto rig. It is an external cooler utilizing alot of 120v stuff and a lot of distlled water only. The pump and two radiators are housed in a coffee table and the inlined cores are behind it. From the front it looks like a big black tower on a mahogany cylinder table. Real simple looking.


Specs like this:

Innovatek Rev 4
3.2 E @ 3.75
FSB 940
GeIL PC 4000
industrial inline 5 GPM 120v pump, like Iwaki 20 or 30
Criticool Powerplant II
approx. 1.5 gallons (six liters) distilled water only

Four radiators.....two craptastic 3/8" copper tube oil cooler type
[ these were like $20 apiece off eBay]
cooled by a 120v table fan +
two heater cores A. Danger Den B. Critcool
[ both of these have ADDA 120v 120mm fans]

I ran it four hours or so today. The water was well circulated. On playing half life 2 and Unreal 04 online extended times over an hour the CPU temperature was less than 40C over ambient, or 60C. With the craptastic experimental oil cooler rads i ran 65C, but I inlined the cores and added a quart of H2O i dropped 5C under real load use. Needless to say on start up the idle is like 47C. CPU temp falls back to 51C within a short time when it unloads.

88888888888888888888888888888888

MOther BOard MOnitor 5

I set Winbond Sensor 2 to read the Winbond 2 2N3904 Diode and now I read
28C idle and 36C LOADED

I was set on Sensor 2 for the other temperatures I reported.

ANYway the spread from idle to load is still roughly the same, a snitch under 10C.



888888888888888888888888888888888

I guess that 14C cold idle to hot load delta T is not too terrible.

You got any pics of this set-up? I really want to see what it looks like.

You said that your CPU temp is like 60c? That is way too high for watercooling. Even given that temp MB sensors aren't that great, you should be nowhere near that high.

yeh that temp looks way to high, i woulld expect that on good air cooling :s

^^ agreed

even one legged, 3 toothed crackhead ghetto water cooling should be allot lower than that.

Guessing you are possibibly readin goff your onboard sensor with no fans in your system?

Or a reseat on your water block is in order.

same

I was busting 70C on air and beeping badly with no OC.

This Rev 4 is basically almost 10C higher than the other Rev 3.
The Rev4 really seems to be a poor 478 socket waterblock.
I am reading MBM5.

This is one of those C stepping processors as well.

I used the white Innovatek thermal grease too.

look

Ad Rock said:

You got any pics of this set-up? I really want to see what it looks like.

Click to expand...

It looks like a big black box sitting on a mahogany cylinder end table. I got the end table at a university auction for $2.00. It is like prison furniture in as much as it is so heavy and awkward the drunk kids could not easily pick it up and throw it.

I should post a pic at that.

I sure is not a UV light transparent panel with tubes too look at.

8888888888888888888888888888888888

As for the poor results: Hey, I am stable a couple weeks and holding a excellent OC. The fastest 775 socket is 3.8GHz non EE. I am in the ballpark maybe minus 10% performance with the garden variety 3.2E CO stepping. What is wrong with that?

Seeing as with that much cooling power you should be max 10oC over ambient...

Your temps don't sound to good, you should check your block to see if it's properly seated. If it is you should throw it away and buy a cheap block from someone in the classifieds. Even with that chip oc'ed you should not be above 45º-55ºc under load.

temps

mysterfix said:

Your temps don't sound to good, you should check your block to see if it's properly seated. If it is you should throw it away and buy a cheap block from someone in the classifieds. Even with that chip oc'ed you should not be above 45º-55ºc under load.

Click to expand...

This is a C0 Prescott here. I am defragging and surfing and running 54C.
I ran the combined SANDRA Benchies and the peak is 60C, onw 61C for a flash even. It did like 54c to 59c basically while loaded with that. I went 235x16 just a bit ago and the benchmarks for files realy jumped. Why would that be? A sweet spot? The rest of the SANDRAs were linear increases. This is a 478 socket.

Albatron promises the 1200 bus and the GeIL promises the 250DDR as the stock basis. Without low latency DDR2 I should do OK. This memory is 2.5 CAS at the DDR500.

My temps never go above 40ºc at 2.75ghz and 1.8v and I also have a videocard waterblock installed running on a single heatercore. The videocard is also volt modded so your temps seem way too high even for a prescott.

How much heat is your pump giving off? You said its an industrial pump which sounds like it may not be designed to run cool. Its an idea to say the lest. Also do you know how much head pressure the pump has? If the pump has weak pressure that could be your problem as well.

I am happy. I won't do nothin'.

Things are really fine.

The water in the system is like 7 degrees Fahrenhiet over ambient. It is the block design.
I ran it for hours and you are focusing on the peak. the Rev 4 is 10C behind the Rev 3 for Socket 478......too bad for me regarding being My Supercool.

I am not convinced the sensor is all that great either. I can put my hand on the and around the base of the block and it is not even sensibly warm.

"Turn the alarm off, Kenneth!!"

If was really like 60C I could tell some heat.

Hey all you. What are your chipsets temps?
What are your case temps?
What are your ambients?

I gained 5C just because the room went to 75F and I clocked it to 3.75Ghz.
I figure they both contribute.

One time I kinked the hose and hit 70C, then I really worried.

Forget the pump. My pump house is ventilated. Nothing to do there. The volute and rotor are advance poly. This ain't an plastic sheathed aquarium pump out of water.....do not worry.

Stuff is like it is.

I'm not worrying about the longevity of the pump, I'm asking if it gives off a lot of heat because some of that heat will go into the water which will affect your temps. If you had a pump that did 1000 Gal/Min but put 100W of heat into the water you'd have a vert poor performing system unless you had massive radiators & an otherwise overkill system. Often times less is more . If its running fine then thats cool. You may gain performance however by switching to a cooler pump if your pump is making lots of heat.

MOTO--in the twenty four hour preinstall run, yes, I am careful, I could feel hardly any heat and could leave my hand on it comfortably. Not so with the Japanese Iwaki I used in the first. I could feel a great deal of heat with the Iwaki. I use that Iwaki as in my plant experience the motor end of pums these days is high. We have some plant pump that will make your hand red in a heartbeat and YOU WILL pull yur hand off the pump.

You willnot pull your hand off either of these pumps, nor get a red hand. That is the methodology I use and figure it is OK.

MOTO, you will find very little sensible heat transfer through a plastic head. In kit number 2 that we jumped off on the motor end was cool. I would spec it out but the price et al meant it was a real blowout deal, a super deal for me. Some guy was cleaning out the parts bin or whatever. Yes, this pump is about 1 amp or wattage like you say and has a max 15psi. well below tube tolerances, the tubes are not ballooning out or anything freaky, even it you dead head it. And speaking of dead heading a centrifugal pump that will not hurt a pump either.

I never had an Ehiem. Are Ehiem volute and rotor centifugal or are they paddle wheel rotors?

I think that if you read up on much of this you'll see that heat transfered into the water v.s. flow v.s. pressure is not such a light deal. Its probably the most overlooked factor in water cooling. This is not to say that bad performance will be found by using too much pump, but you may not be at your optimum performance. The difference between too much pump & just enough pump is probably half a degree to a full degree which is can mean the difference between stability & instability in a heavily overclocked system. Too little pump of course will hurt your performance a good bit more. I'm guessing that you were using an Iwaki 20RZ or 30RZ. From the sounds of it your pump isn't putting off a ton of heat which is good.

Here are some things that help show what I'm talking about here (registration at OCAU requred):

http://forums.overclockers.com.au/showthread.php?t=322434

http://forums.overclockers.com.au/showthread.php?t=242651

From the findings you'll see that the 30RZ ends up performing slightly worse than an MCP600. Cathar's testbed is setup so that it can take a lot of heat. In systems with lesser cooling capabilities such as mine with a single heater core the difference between the two pumps in my system would be bigger for sure especially since my block doesn't really benefit as much from higher pressure than a relatively restrictive design like the cascade would.

Like I said though, it its running just fine (meaning performing well) then its cool & not a big deal. I was mostly curious seeing as it was an industrial pump. My experience with them is that they are very hot but using some common conjectures about pumps I've seen online & hearing your comparison it isn't all too bad. A 1 Amp motor its using 110W of power when running off of 110VAC & something like 1/3 of the energy becomes heat which can enter the water (~36W) which will place it under the Iwaki 30RZ. Pumps don't empty all of their heat into the water when in in-line operation but a good deal still is.

If the head pressure is similar to the 30RZ then you have a much better pump I would think.

MOTO-----"I was mostly curious seeing as it was an industrial pump. My experience with them is that they are very hot but using some common conjectures about pumps I've seen online & hearing your comparison it isn't all too bad"

These are both magdrive so the motor is seperated by the mag drive that is basically not heat producing. You might think they are direct motor to pump and all metal. That is not the case. You could get a used one cheap and inspect one. I incorporated a used Iwaki 20rzl that looked a little rusty on one base bolt and the seller said it had pumped chemicals and warned that might not be good for a fish tank or pond. As my PC might run only a few hours when I am there, I did not care about years and years of continuous operation such as if the pump ran 24/7 in an industrial process with no one to watch.

There are literally a ton on auction during a month if you just follow the eBay auctions.
Big factories overbuy, wholesalers are overstocked, stuff like that, and people sell them on eBay auction via that search method of putting the buyers and sellers together.

I saw an Iwaki model 10, a wimpy one, but get this: should you have the reservior a few feet above the pump and say over the height of the water block you merely are incorporating a 'circulation' pump and need only worry about tube friction loss and restrictions and hope that .5amp wonder will push enough. You can bench the pump with a poly bucket of H20 and a tube loop to watch the pump work out as I did before making an external system. I think the Iwaki Model 10 would work if you had good 'static head' on top of the pump {again, reservior water level elevated above the pump and waterblock level. I say this because I saw one start on auction brand spanking new from a copier company at $10. You know Iwaki makes a durable quality pump.

The "in the case packages" get away with doing what they do using a better cooling medium than just distilled water and they have as large and as restriction free a tubing setup as they can when using the little pumps that have to fit in the case, and at 12v cannot consume a lot of amperage/wattage to cause more heat in the case, not to mention the double PSU problem or the 700w PSU problem.

You could prowl eBay and get to the sweet spot and find a few generic 5GPM pumps on sale right now. Many sellers photo the name plate with the power of the pump right before your eyes.

So for an external you could feel the ball park would show:

5GPM
.75 to 1.25 amps power
'head' say 30 feet, or 15 PSI

*********do not worry about threaded inlet/outlet or barbed inlet/oulet size at all, just screw on Home Depot PVC bushings [most are NPT generic] or reducer barb to your system size********

Iwaki has all the pump curves at Walchem's [distributer's website] so those figures cut and paste nicely in a Notepad you can put on the task bar as you shop. The Iwaki was barb outlet fitting that are 3/4" on the 20RZL.

Some names to watch for regarding INLINE pumps:

Little Giant
I-R Aro
[of course] ...Iwaki
there are many more, probably a dozen makers, i do not recall them all

Used is cool. Used is cheap. Then, you have to be a bit of a gambler and might luck out. Surprisingly, since there are literally millions of these pumps in use, people are moving out excess and so forth. Fortunately, there is a sweet spot at eBay even I have to search for as the little guys are mixed in with the big boys, but they might have a capacity breakdown if you get lucky.

MOTO------------"If the head pressure is similar to the 30RZ then you have a much better pump I would think."

More than enough power to pump water from the room below in a normal house to the one where the water cooling system is, assuming the reservior is by the PC. Check those curves at Walchem's site.

MD/WMD Specifications (English units)
Polypropylene
MODEL CONNECTIONS MOTOR
OUTPUT
(HP) MOTOR
SPEED
(RPM) AMPS MAX.
FLOW
(GPM) MAX.
HEAD
(FT) MAX. SYS.
PRESSURE
(PSI) S.G. WEIGHT
(lbs)
HOSE
(in) NPTM
(in)
MD-6 1/2 - 1/250 3100 0.25 2.3 4.5 2.8 1.2 2
MD-10 1/2 - 1/125 2900 0.40 3.1 6.8 4.3 1.1 2
WMD-15R(T) 1/2 1/2 1/75 3100 0.82 5.0 11.1 7.2 1.3 6
WMD-20R(T) 5/8 3/4 1/38 3100 0.48 8.2 14.1 10.0 1.1 7
WMD-20RX(T) 1 1 1/38 3100 0.50 13.7 8.2 5.7 1.3 7
WMD-20RZ(T) 5/8 3/4 1/38 3100 0.53 2.9 22.6 14.2 1.1 7
WMD-30R(T) 3/4 3/4 1/16 3150 1.00 10.0 17.7 11.6 1.3 9
WMD-30RX(T) 1 1 1/16 3150 1.00 19.0 13.5 8.5 1.1 9
WMD-30RZ(T) 5/8 3/4 1/16 3150 1.00 4.6 36.1 24.2 1.0 9
WMD-40R(T) 3/4 3/4 1/12 3200 1.90 13.7 21.3 14.2 1.1 10
WMD-40RX(T) 1 1 1/12 3200 1.90 22.4 15.4 10.0 1.1 10
MD-55R(T) 1 1 1/9 3400 1.60 18.4 26.9 17.1 1.2 10
MD-70R(T) 1 1 1/4 3400 2.80 25.6 31.8 21.3 1.0 13
MD-70RZ(T) 3/4 3/4 1/4 3400 3.80 11.4 66.6 42.7 1.0 13
WMD-100R(T) 1 1 1/3 3300 3.40 35.6 39.0 25.6 1.2 varies w/motor

that HTML scrambles it

Yeah, the performance curves are still there at Walchem