Quad G5 Liquid Cooling Pump Chugging, nasty thermal disparity between cpu's

Dear Samsara and all other forum members reading this thread,

I found the French site again which provides detailed information regarding the liquid cooling systems used in G5-computers and the reasons why these masterpieces of engineering eventually have to fail.

The problem is that the information is in French only and that it loads very, very slowly even with a fast internet connection. I hope the site is useful to you all.

Best regards,
Robbert

Lol, the image on the top of the first page should be a classic...

Thanks, Mechanic man, that's the best I've seen on the topic, really outstanding with the pictures shown at least. Thanks a lot for providing that link.

True it's in French, which probably most Americans will have problems with myself included, but with many Translation apps and services out there I bet it wouldn't be that hard, a little at a time perhaps, to understand what is said.

Thanks for being an example of what it means to be a member of the Mac Community. 🙂

Thanks for the link. Great information. Gasket sealant looks like a really smart addition.

Dear Samsara,

Thanks "for the flowers", but although the computer shown is a 2004-version, I found this site very interesting for all owners of liquid cooled G5s. I don't think Apple changed the concept that much afterwards, because the same problems seem to appear in the later Quads.

Pay special attention to the pictures on page 3, showing the thin bladed copper radiators firmly attached to the processor cooling plates which are dissipating the heat into the fluid inside the cooling circuit. The crystals and jelly partly obstructing the flow explain why some users have more or less sudden temperature problems, despite of the fact that the pump (or pumps) still works as it should.

Best regards,
Robbert

flowers

(sigh) In all honesty MM, they didn't really come from me... but from the many others who you have helped and will help.
I have a Dual 2.7 that has heat issues but the liquid cooling system seems fine. No sign of any coolant or corrosion at least. The Quad I bought use uses liquid cooling also but I never had one problem with it. The Quad is the finest purely Apple Mac I ever had, the last made. Although I have one and love it... Intel still means "icky" to me.

However if you like the flowers, consider a trip to one of those "special" bars you have there and send me a double order. 🙂

I didnt want to start another thread for this question, but I was wondering if the G5 I just bought (Its still in shipping, or I'd look inside and see) is liquid cooled or not. I bought the late 2005 2.3GHz dual core machine. I know the 2GHz was not LC, but I have no idea on the 2.3GHz

No, according to EveryMac , at least.
Yours is air-cooled. I'd try to keep it extra cool if I could, it's a G5, can't hurt..

Is there anyone who knows whether you can start the G5 with the processors disconnected to provide power for the pump. I've dismantled my processors (G5 Quad) and got the fluid out, but need to test the pumps and refill the system.

Yes, I have done that successfully. I tried using an old PC power supply to power the pumps, but never figured out the correct connections. It turns out that a G5 will run just fine, although with any remaining fans on a high setting, with the processors removed. By the way, I am still struggling to get my G5 running successfully. I've got the Quad with the 2nd version of the LCS and one of my two CPU cooling blocks is internally obstructed. I thought that I had it clean, but it clogged again after just a few hours of running.

Good to know someone else is ahead of me here. Thanks for the answer, it saves my main worry. I thought that one of my blocks was slow running, but they both seem to be quite restricted. When I blow through them there is a fair resistance.

I'm on the 2nd version two pump quad as well. It would be good to hear how you put it back together and bled it. I've cut the feed stub on ali pipe, but not the bleed stub on the radiator. Have you documented your process anywhere?

London,

I had planned to document my process, by I got in a hurry and did not do so. I used the stub on the aluminum s-shaped tube to fill and the stub on the top of the radiator to bleed. I found that 3/16 inch ID tubing (I think that's close to 5 mm) fit fairly well over those stubs and that I could achieve an adequate seal by wrapping tightly with fine wire. I used gravity to fill from a length of tubing hung from the ceiling with a string and I would control the fill by pinching the tube attached to the bleed port. I had to run lots of water through to get most, but not all of the air out. I could not get the last bit of air out, but that seems to remain in the top of the radiator and doesn't interfere during normal operation.

The fill and bleed ports can be re-crimped while the tubing is still attached, giving an adequate seal. then I removed the tubing and replaced the black caps filled with 5-minute epoxy. I replaced all of the black hose with clear plastic tubing. I refilled with distilled water only with a few drops of benzalkonium chloride to inhibit growth of algae or bacteria.

After reading lots of web sites run by folks who water cool PCs in order to overclock their CPUs, I think that the cooling blocks on our Quads are becoming corroded internally because the cooling circuit includes both aluminum and copper (or whatever alloy the blocks are made of). You might wants to try an additive that inhibits corrosion.

I'm going to run my quad for a while as a dual while I try to figure out how to clean the clogged block. I used a weak acid solution that is sold here to remove hard water deposits from coffee pots and pluming fixtures. Next time, I may try a phosphoric acid solution that is used to etch metal in preparation for painting. I am nervous about damaging the internal structure of the block, but I think that I may have to risk that. Do you have any good ideas for cleaning the blocks?

Good luck!

No ideas yet for cleaning them, I was hoping I might find a replacement in the watercooling market. It's such a shame the parts are not available. I had a massive thermal disparity so I'm suspecting the pump right now, since the blocks seem equally slow. I will try the pumps now I know I can power them from the case. In a rash moment I thought tapping the cooling surface might help, but the metal is very soft and dents immediately.

Thanks for the extra info about your process, I will use the epoxy on my bleed tubes. Did you get that from a watercooling site, I use a lot of different glues and epoxy in my trade but not sure where to source that one.

Also I'm thinking about using the ready mixed solutions that are supposed to be non-conductive and food safe, I measured the fluid in the system and it was 100ml per side.

As far as removing all the air goes I'm sure there was some air in both sides of my system and certainly it caused no problem in one side. The whole liquid cooling system is so well engineered in comparison with systems I've seen working, in principle it should be easy enough to repair. Anyone has links to comparable liquid cooling blocks, or spares?

I've uploaded some pics of the job so far, you can see the discolored waterblock:-

http://gallery.me.com/mrandyjones/100003

Andy

Great pictures! Thanks. The discoloration on the one CPU block is very odd. Almost looks as though something was splattered onto it during the original assembly.

The epoxy that I used is here:
http://www.loctiteproducts.com/products/detail.asp?catid=11&subid=49&plid=157

The adhesive used originally by Apple is tough, but brittle, which helps in disassembly. I guess that I'll find out soon how well I can disassemble the stuff that I used. I actually ran the machine for several hours before finally gluing everything back together and the lower temps and fan speeds made me think that I had the problem solved. My guess now is that I had dislodged the blockage, but not removed it. The internal structure of those cooling blocks seems to be very complex, based on the limited view available through the ports.

I agree that the mechanism seems well-designed and should be easy to re-build. Unfortunately, most of the stuff available for PC cooling is not designed for the very tight spaces available in the Apple design. I am watching Ebay for parts of broken Quads, but no one seems to sell pieces of the cooling system. If you find something, please share the information!

Thanks for the link to the epoxy, I tried the pumps today and they are both running. I will get some hose and test the volume they are pumping to see if it's equal.

Now it seems as if the resistance of both water blocks is roughly equal, I tried cleaning one with viakal (limescale remover) but no deposits in the liquid so I assume no blockages removed. Having a close look at the block I'm beginning to think it is a two tier design with complicated network of ducts close to the cooling surface leading to a radiator which is cooled by the liquid. It could be the very fine ducts that are clogged I suppose.

I can't understand why you would be having a problem with your system if you managed to test run it for a few hours. Did you flush all the parts before filling the system?

I'm trying to source some spares, if I have any success I'll let you know.

I have one pump only, and this is a delphi one ...

Actually, this is the cause of my only problem : noise ...

Calibration test and hardware test failed ON THIS ELEMENT only, during the test.
Causing the test to failed, the system bo te stuck on "checkstop" and "CPU AOVERTEMP"