Most amazing homebuild watercooler...

Toggle sidebar Toggle sidebar
G

Guest

Guest
Does any of you have any similar experiences????

I just finished my watercooling project, and amazingly it performs almost identical to my silverado!?!?!?

When I turn on my computer, it is pretty cold when booted to windows its about 35 Celcius. Then it goes to a steady climb, and flattens out at about 50 Celcius. The water is initially 22 degrees and climbs to a max of 33 degrees.
These are exit temps from the coolhead.

First the specs.
Waterpump: Eheim 3.8 L/min
Radiator: 4 meters of 8 mm coppertubing
Cooler: Modded P3 cooler with 4 watterchannels
tubing: 12 mm plastic tubing.

The weird thing is that the waterflow in the coolhead seems fine, I made an acryllic top, so i can see when the head is filled, the water passes right.
The system contains just under ½ a liter of water, so the water circulates the system about 8 times a minute. isnt that enough???
I use a big fan to cool the radiator, the fan is set at 5 volt, the difference when using 12 volts was only 3 degrees on the CPU???

Hope somebody can come up with a suggestion...
It was great fun to build, and it still performs like a silverado, and my only expense was the copper tube for the radiator, all the other parts was old obsolete parts.

There is much room for improvement, I just dont know what is the weakest link right now.

PS. Temps are Athlon 1000 Mhz @ 1400 Mhz 1.85 volt


---
Engage!
 
I think your pump is about 1/10th the rated speed of usual PC pumps, so it might be that the water is just getting too hot. Also people generally have large reservoirs (but not all) which increases the volume of water available.

Is your radiator just the cpooer tubing with no fins? You may do better to get an old car oil cooler or similar as just piping will not have enough surface area to cool effectively.

-* This Space For Rent *-
email for application details
 
Thanks for yuur input

I think you may be right that the pump is to small, I will try to borrow a more powerful pump.

No my "radiator has no fins, but is one 4m long copperpibe, I did some rough calculations on it, and it should be about 10 times the cooling surface compared to my Silverado. I chose the copper tube because its cheap, and I can shape it to fit the PC case. I have formed a spiral that fits under the drive bay, together with the fan and pump. Auto parts are extremely expensive in Denmark (taxes), even second hand parts.

Right now the water is 32.4 deg. celcius, ambient is 24.8, so the water is 7.6 above ambient. The core is 51 which is 18.6 above the water temp.
What temps are reasonable? It seems that there is most to gain between water and core temps?

---
Engage!
 
I think your water temperature is the issue - if you cannot get that down, you're not going to realise any benefit. There is no point taking any more heat out of the cpu unless you can get it out of the water. Either that or you are going to have to use a TEC on that and live with the warmer water.

The radiator/water cooling often happens OUTSIDE the case, that way you prevent adding heat to your case temps. Can you try your cooling setup without the cpu attached in the open air and see what you get the water down to? It should get to much colder temperatures. Until that water gets and stays colder, better flow and radiator performance, you are not going to do any better...

-* This Space For Rent *-
email for application details
 
I did try the radiator oustside the case, only 1 degree lower temp, I think thats because the case helps the airflow around the radiator.

I have borrowed a larger pump eheim 960 l/h, and I will try it out tomorrow.


---
Engage!
 
Two of the biggest issues in these watercooling setups is the efficiency of your waterblock or jacket and the cooling of your water.

Your water temperature is rather high - I don't think your copper piping is doing a good enough job acting as a radiator. Ideally, your water temperature after leaving the radiator should be close to room temperature.

You may also want to take another look at the contact of your waterblock with cpu and its surface. When you say 'modded P3 cooler', did you turn a regular heatsink into a waterblock - like the lexan mods offered by a website whose name I can't remember? I too was going to attempt to mod my old Monster II, but someone convinced me not to. It may be the transfer of heat from metal to water is not as efficient as in a professionally designed waterblock. Does it have a lot of turbulence in the block, or does it flow smoothly through?

And as someone else mentioned, your pump is a little weak. 6 to 8 liters/min is a better number to shoot for.
 
It seems that my pump, radiator and waterblock are all the problem.

So I plan to change it all, I will change one element at a time, to see the effect of each unit.

Yes it is a standard cooler modidfied for water, read more at overclockers.com. In Denmark a watercooling kit costs almost 200$, the cooling head alone 60$, I have used 33$ on epoxy, silicone, copperpibe, tubing and tighteners. So compared to the 70$ for the Silverado its not that bad.

You should definetely try making your own block, its great fun, and if done right, you may get even better results than with a commercial block. But mine seems not to be right...

Just tried a bigger pump, results are almost identical. This pump is an Eheim Professional and pumps 960 l/h or 16 l/min. that is 4.2 times as powerful as my old pump, but results are not better?? The pump has a 10 l resovoir. So it takes longer to heat all the water, but endresult for water and CPU are the same, it just takes longer to reach max temp.

So back to the coolhead and radiator, I have build a new coolhead with a new approach. The current one is big and with a thick bottom. The new one is smal and with a thin bottom. Maybe its better to have the water closer to the core? I will try it as soon as the silicone dries. (at least 24 hours!)

Waterflow in both is good. The current has very little turbulence (a little in corners opposite to entrance and exit holes), the new one has absolutely none. The new one is an old cooler for 486 or pentium I don't remember. Both coolers are build by making an acrylic frame glued together with silicone and epoxy. Drill and patch holes strategically.

I agree that my radiator could be more efficient therefore I will try 10 mm copperpipe instead of 8 mm, that would reduce waterresistance and increase cooling surface area, I think I can make room for 6 meters woth of 10 mm pipe, with reasonable airflow, This will double the surface, and reduce resistance a little.

Right now im running 1410 Mhz 1.85 volt at 47 degrees, the machine has been on for almost 2 hours, but this is not max load.

PS Previous stated temps are max load. Water temps are exit from coolhead.

---
Engage!
 
BillHS,

Your waterblock is not the problem - you just said yourself, the water is okay at first, but eventually it all heats up... The heat IS getting from your CPU to the water, it just isn't getting OUT of the water again - you need a radiator - that piping alone, I fear, Is not going to do it...

As a comparison, I think your last cooling figures are below a fop38 and stock fan ~ $30...

-* This Space For Rent *-
email for application details
 
i advise you get something thinner than 8 mm for the radiator and a more powerful pump... but then thats rebuilding the whole setup... the temps arent TOO bad and thats not a bad oveclock at all 40% is quite impressive i think... but then again maybe not... if i were you id rebuild it... have all the piping made out of copper if possible and have a small copper block with pipes running through it instead of a modded heat sink... that could bring the water (which is colder significantly than your cpu in your case) nearer the cpu increasing energy transfer...

you do not strengthen the weak by weakening the strong
 
I also think that its the radiator part of the setup thats the problem- if the water is heating up- which it obviously is, it needs to get cooled, which it isnt, how about using a heater matrix from a car, maybe with a fan on it, I cant imagine copper pipe on its own (unless theres miles oif it) will be able to didipate the heat quick enough.


Next time you wave - use all your fingers
 
In my opinion the radiator is the weak link in your system at present. Using a larger copper tubing should be beneficial but it still will not compare to the surface area of a typical finned radiator. Is the copper tubing radiator passive or do you have a fan blowing over the tubing? You did note that it worked better inside the case where there was likely better overall air flows but probably higher temperatures. You should see if you can find an inexpensive radiator from an auto parts store (heater core or oil cooler), preferably with copper piping to limit the potential for galvanic corrosion.

The reduced friction from larger copper turbing will likely be insignificant compared to other friction losses in the piping system. The flow rate does not need to be high in order to provide for the heat transfer as water has a very high specific thermal capacitance (heat/temp rise). And turbulence is good as it provides greater mixing of the fluid flow and almost all piping systems will have turbulent flow unless there is exceptionally low flow rates to allow for laminar flow conditions.

These are just my thoughts as I prepare to build my own water cooling system... but based on many many hours of pleasurable browsing many great web sites sharing other's experiences and a decade of professional experience as a [municipal] water system designer.

Regards,

-PT
 
Ooooohhhh Thanks gyus, that a lot of qualified answers since last I was here.

I'm pretty bored with my computer right now, because I cracked my Athlon....
I just sold my old system, so its 2 generations back to the old old system,
An AMD K6-2 266 running 337 Mhz (no agp). Originally it never got stable beyond 300, so the new cooling is definately better than the old old one.

After browsing the net some more, and reading a lot of tests, I have come to wonder if anyone has made a reasonable comparrison between newer aircoolers and watercoolers. I can't find any, and everybody meassures in different ways, and im sceptical about comparing results from 2 different setups.

Before my Athlon core cracked, I tested the radiator, the previous results were from the radiator intake, when touching the radiator the temps felt the same at in and out hoses, but I meassured a difference of more than 4 deg. C. So the radiator is not totally inefficient. Remeber the water only stays in the radiator for less than 2 seconds, then its out the other end! With the bigger longer pipe, that would be something like 5 seconds. If the cooling was linear, that would mean that water comming out would be below room temperature, that is of course not going to happen, but I think it will be sufficient. The 8 and 10 mm. are the inner diameter of the pipe. Thats the standard way yo meassure copper pipes in Denmark, I don't know about other countries. Unfortunately I never got to meassure my new cooling head, but I hope to get a new Athlon by the end of the week.

By the way, My AMD K6-2 @337 Mhz meassures 36 deg. C. on an Asus TX97-X motherboard.
Athlon Was Asus A7V133,
CPU meassurements by motherboard monitor 5, using the motherboards own sensor in both cases.

I had an old K6-2 200Mhz I decapped it, and will use the cap to protect the core of my new Athlon while testing. Then remove it when i'm confident I have a reasonably good cooling system, this way I hope to be able to prevent another cracked core. This is of course a sign that my clip is not very good uuups.

Thanks again for all your input.
I'll be back with results ASAP.

---
Engage!
 
I wouldn't advise using a K6-2 cap on an Athlon - I've no idea what sort of thermal impact it will have, but it can't be good. Very difficult to know if you've got a good contact under the cap, plus it's yet another thermal interface. You'll be running the risk of frying your new Athlon, which is no better than cracking it.
 
Good point, the cap is aluminium, and not very thick, I have sanded it to fit in hight, its fairly simple to check the contact, putting some grease to the core, and check if it sticks to the cap. Furthermore I have the option to underclock. I think the cooling will be less efficient, but not by more than a couble of degrees. But thanks for the warning.
If others want to try this, you should also pay attention to the bridges, theres an obvious risc of shorting them, but you can just tape them over.
Maybe if somebody made a silver cap, it could be used for both protection and improved heat transfer? That would be neat.

---
Engage!
 
Just got my new CPU, I decided to go low budget and wait for Athlon 4 for desktop. So my system is now Duron 750 @ 940 Mhz. Its surprisingly cool, only 38 degrees Celcius normal workload 23 ambient, so its become much less interesting to make an extreme cooler now! Maybe its not so bad after all.
Thanks for your interest and participation in this thread.

---
Engage!
 
With what mobo, CPU, Mhz, ambient temp?
I just finetuned a little, Im now running 950 Mhz at 1.75 volt at MAX 39 Degrees Ambient 23.
But of course, if I could get it to 32 it might be possible to go a little further.

---
Engage!
 
Right now it's an A7V133, TBird 900"B"@1050(140x7.5), and runs around 32-33C whatever I am doing. Before this I had an A7V with a Duron 750 @ 1.14GHz doing 35C.

---------
Grass is a beautiful weed
 
So, if I understand what I've read:

-Try and get the same metal components for all parts.
-Use the swanky Water Wetter to reduce corrosion risks and improve cooling.
-Definitely use a fin radiator.

Now for the questions-

I am not building my own water-block. Which one do you recommend fo a tbird socket a?

I am also not bending my own copper- should I use surgical tubing? Clear plastic tubing? Thoughts?

I am doing this more for a quiet box than for over-clocking- Do fans on the radiator generate much noise? Can I live without them? If the fans end up being as noisey as the one in the case, what's the point?

What about a pump? Can you recommend low-noise with decent through-put?

Big thanks to everyone in advance. I'm a newbie, but I think this is too cool to pass up.
 
I do not know what is available in your area, here in Denmark it is almost impossible to find anything for watercooling. You have to do it all yourself.

You could go to www.overclockers.com to learn more, they have a number of tests.

Waterblock CoolChip
Pump Eheim 600 l/min or 1000 l/min.
Fan Pabst 12 cm. Low noise

And yes you can make a very quiet system, because you can use large fans at low speed.

---
Engage!
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom