• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

PowerMac G5 Quad "The New Blood Mod" - A guide to flushing, modifying and refilling the dual-pump cooling system.

Quadraman

Well-known member
I only say non-conductive, because of the potential for short circuits if a conductive fluid comes into contact with anything electrical as the result of a leak. That's the problem with water as a coolant for PC's. A leak can kill everything.
 

CircuitBored

Well-known member
I only say non-conductive, because of the potential for short circuits if a conductive fluid comes into contact with anything electrical as the result of a leak. That's the problem with water as a coolant for PC's. A leak can kill everything.

You may be over-thinking this. Leaks and their resulting short-circuits are only a concern in a badly maintained or poorly assembled system. There are millions of water cooled systems out there that suffer no ill fate because they are properly looked after. Unless you're planning on doing a really bad job of overhauling your LCS then you likely have absolutely nothing to worry about.
 
Replacement RAM from OWC arrived a few days ago and all seems to be going fine now. CPU A's exhaust fan was spinning at 100% for some reason but a reset of the logic board (little button below the RAM slots) fixed that.

For the coolant I used clear EK Cryofuel. For repasting all chips I used Thermal Grizzly Kryonaut Extreme... way overkill but still had some left after building a gaming PC.

Images from left to right:
1. After resetting the logic board this little LED came on when I plugged power back in. It turned off when powering on the system and hasn't turned on since.
2. The temps (in C) after booting it up this morning.
3. Outer panel off. It doesn't go into fail safe mode and works fine without the inner plastic panel. The panel sensor does work since LED4 (trickle power) tuns on when machine is off with LED3 (main power OK) and LED5 (CPU arrived at open firmware) briefly turning on during boot. When I replace the inner door these lights stay off which is normal. However... I keep the inner door off since it's a tight fit with the new pump tops. Also, the PCI slot covers are off for better airflow.
4. A closer shot of the pumps. It took a while but most (if not all) air bubbles have been worked out.
5. Temps while updating Xcode to 3.1.4. Its not loud at all. I've noticed CPU B seems to run hotter then CPU A. The hottest I've seen CPU B was 67C while running Cinebench. The fans and pumps did ramp up to compensate and it was just purring along. After the benchmark ended the temps almost immediately dropped below 40C.
6. The Apple experience spanning about 11 years. The only computer not purchased new was the G5.

I still think this machine was barely used (if at all) since the original HDD has and installation of Aperture 2 and Final Cut Studio with no projects or anything indicating they were used. I've since deleted the original user account and created my own. My guess the original owner had issues with it overheating with kernel panics (same when I obtained it) and just left it. Most of the original 16gb of RAM was bad too.

This project made the Performa, G3 and G4 look like a cake walk.

Edit: i'm currently downloading Final Cut Studio 2 from macintoshgarden to see if I can put a real load (giggity) on this computer and see how it performa.
 

Attachments

  • 20230825_162554.jpg
    20230825_162554.jpg
    901.2 KB · Views: 19
  • 20230827_074006.jpg
    20230827_074006.jpg
    6.9 MB · Views: 12
  • 20230827_080315.jpg
    20230827_080315.jpg
    6.2 MB · Views: 19
  • 20230827_080345.jpg
    20230827_080345.jpg
    5.6 MB · Views: 19
  • 20230827_082317.jpg
    20230827_082317.jpg
    8.4 MB · Views: 21
  • 20230826_101225.jpg
    20230826_101225.jpg
    4.6 MB · Views: 21
Last edited:

CursedSilicon

New member
I've had to take a different tact with mine - currently trialling some generic CoolerMaster heatsinks and fans, with the dual Delphi pump unit in mine having completely corroded itself (radiator OK with ultrasonic clean but CPU blocks packed with solid corrosion), oh and the pumps failed. I'll post up a new guide should I get this air conversion over the line. Just mounted them in (with a lot of metal work), seems fine nothing at all warm, but fans ramping up with no shutdowns as before or error LEDs, calibration is next.


View attachment 57621

Have you got any updates for us on this? I've got the Gen 1 model cooler that seems to be slowly failing (but mercifully not leaking...yet) and just swapping it to some modern air cooling seems far more preferable for longevity of the machine
 
It has now been about 2 weeks since i've finished this project and in that time I got 4 more sensors running. Still not sure why they weren't working in the 1st place or why they all started working at once. For starters the drive bay sensor and 3 others from the back side were giving "general errors" during testing. The machine was running fine without these sensors so I figured i'd at least unplug and replug the drive bay sensor on the logic board. It was already fully seated so upon reboot the 4 sensors showed up in istats. However... I noticed the CPU temps climbing about 1C every 2 seconds or so creating a runaway thermal event. Taking off the side panel I noticed the Intake fans for CPU A/B and PCI bay were not spinning for some unknown reason. Istats showed them spinning at well over 2k rpm. At this point the CPU temps were in the 80s at idle so I shut it down. Figured there was something wrong somewhere so I reset the logic board with the little button on the bottom of the board. Apparently that worked.

I'm still not sure why unplugging and plugging in a sensor unrelated to the CPU/PCI fans would cause them to stop.

Attached images from left to right:
1. The 3 LEDs that should come on when the inner door is removed come on.
2. All the sensors are finally working.
3. Running Cinebench this morning. 64C was the hottest it got. The fans didn't have to spin much faster than idle to keep it there. I've noticed CPU B (especially core 2) seems to be hit harder and run hotter while installing software and other tasks.
4. A couple seconds after the test stopped the temps dropped to what it likes to idle at.
 

Attachments

  • 20230828_192840.jpg
    20230828_192840.jpg
    1.6 MB · Views: 8
  • 20230828_192706.jpg
    20230828_192706.jpg
    3.3 MB · Views: 6
  • 20230910_093945.jpg
    20230910_093945.jpg
    5.6 MB · Views: 5
  • 20230910_094054.jpg
    20230910_094054.jpg
    5.7 MB · Views: 7

killer

New member
Just a question, those look like Aluminium radiators are they? In my experience with systems with mixed noble and ignoble metals, it doesn't end well. Even inhibitors only work in the short term.
 

TCH

New member
Hi, do these acrylic tops work on single pump Delphis? Ie. is the pump same as in double pump systems?

I am planning to overhaul my two single pump G5's and wonder if I can use the same plan as the one in this thread?

Thanks.
 
Top