Working on the G5 quad liquid cooling system

Let's keep this LCS thing simple. Seems that people are getting the wrong LCS makers confused in postings/replys.

1-pump Laing DDC LCS = Delphi

2-pumps Laing DDC LCS = CoolIT

2-pumps; one each atop water-blocks LCS = Panasonic

Silcone is an excelent material for the LCR. Delphi has been making liquid coolers for 75 years. Half of us on this thread most likely have Delphi parts in our cars Domestic or Foreign. The issue is so simple though. Placing a chamber of liquid directly above delicate electronics is a recipe for fails.

What I think really went wrong in these O-Rings is that it wasnt so much physical liquid leaking out..It was rather slow evaporation and vapors that seep through the o-ring while under high CPU heat loads, this then eats away at the heat plates and CPU. Yes Im sure many fails have happened from o-ring cracks and tube leaks..those being the extreme. I have my Dual 2.5 G5 running good. I check it from time to time and also frequent these cooling threads to keep on the finds of DYIers in the need I have to overhaul it. ATM I see no need too as my inspections come back good.

FWIW..Go with Viton material its wonderful when used with glycol and other presurized heated lliquids. In fact i might be overkill..lol

Silcone is a good material. However, I would not allow the silicone to cure around any electrical components or contacts within 30 feet. And even then allow the cure process to take two weeks to ensure the cure shape took hold.

Silicone contamination of electrical switch contacts can lead to failures by causing an increase in contact resistance, often late in the life of the contact, well after any testing is completed when the product is in the consumers hand. All that is requied is a silicone spray or silicone to cure around these contacts.

How many contacts in a basic computer .... hundreds !

Contacts are CPU to MB, graphics card to MB, the main power switch, the connections to each hard drive, the USB port and any port on the computer, etc ....

Also, I would agree Delphi (GM Harrison Radiator, ACG) has been making cooling products for nearly a century. However, like any company, they too have had recalls, quality spills, and liability blunders.

Wow, thank you for this detailed write-up! I will be using your information as a valuable reference.

I just purchased a used Quadcore G5 with dual Delphi cooling units and this issue hit me at complete surprise as I thought leaking was the only issue revolving around Powermacs. Consequently, leaking was the only thing I checked for during before my purchase of the machine. During my last successful boot-up I managed to download iStat Pro and I carefully watched as CPU B temperature reached 80*C before the computer shut itself down. CPU A only reached ~50*C max during that last operation, which lasted about 10-15 minutes total. The fans seemed to stay at about a medium speed from boot-up until shut down. Any attempts to start the Powermac after a successful boot results in an expected hang-up after the chime and the two signature red LEDs activated inside the Powermac. The Powermac must be at ambient room temperature to boot successfully. Interestingly enough, the previous owner had been running the machine with no problems or excessive fan speed. I have been thinking that something in the pumps or radiator must have been dislodged when I layed the Powermac on its side when transporting it to my house.

Anyhow, I will be attempting this cleanup of the internals when I get a chance. Thanks again!

Also, Heikki and those knowledgable on the subject, I was curious to find out what opinions are on replacing the Delphi pumps themselves during this whole process instead of cleaning them out? Would this create compatibility issues when running calibration from ASD disk since the pumps are not Delphi?

I have come across this product on another forum as a possible replacement for the Delphi units. http://www.svc.com/mcp355.html

Judging by the specifications for the Delphi MCP355 pumps listed here: http://www.sidewindercomputers.com/swiftechmcp355.html I am reading that the mean time between failures is listed as 5 years. This is almost reason enough to replace them both as the units are now over 6 years old. This is taking into account those that like to leave their Quad cores on for most of the machine's lifetime.

Just to note the difference between the arrangements. Hosing is reduced in the Quad core by a significant amount. I will be cleaning out the pumps and radiator within the net few weeks and I will be updating everyone on any performance improvements.

Chris323i wrote:

Just to note the difference between the arrangements. Hosing is reduced in the Quad core by a significant amount…

That's a most illustrative and helpful picture, thank you—especially from those of us who have been just too chicken to take our Quads apart. (Hope your thumb is OK now. 😉)

Chris323i wrote:

…and I will be updating everyone on any performance improvements…

Duly and truly appreciated. 🙂

Here are a few extra pictures I managed to grab during the teardown.

This photo was taken showing both processor boards attached to the cooling system.

Here you can see how large of a difference this plumbing system varies from the dual core setup.

Here I was taking a picture of the unique tubing that is located on each side of the copper blocks. This plastic material seems much weaker than the thick tubing located near the radiator. Also note the extra sealer. It can be found at every hose attachment point.