Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I had a similar experience @amishallin. I inadvertently failed to put the CPU intake fan block back in one time, and the machine booted and ran just fine. I figured out what I had done when I noticed the CPU temps were a few degrees higher than normal. Since I watch those temps like a hawk, I caught it right away. But... it seems clear that the (air cooled) machine will boot and run just fine without those intake fans.
 
What I find interesting is iStats3 still reports the intake fans as spinning. I'm not really sure if the machine really knows what it's doing. I was expecting them to either show 0RPM or not show at all.
 
Funny! iStat Menus shows temperatures for CPU B on my Half Quad, even though I have CPU B turned off. The iStat code may not have accounted for "phantom CPUs"!
 
I wonder if there's still sensor data that it can pull from, even though the CPU itself is disabled.

In regards to the fan report, it's probably just pulling from whatever the OS is trying to make it do, not what it's actually doing.
 
I agree. It was probably unheard of at the time to convert to air cooling or swap CPUs around.

Speaking of swapping CPUs around... I found a replacement for CPU B on Ebay and hope for the best when it arrives. I've done some digging and found CPUs with the same EEE code should in theory be compatible with each other. The EEE code is taken from the white sticker on the bottom of the CPU card in the format of "xxxxxxxxxEEEx." With the EEE being the code and mine being UPE. The specific response on 68kmla is here and that led me to the complete assembly at DVWarehouse.

I did find out that the June 2004 model (not relevant here but still interesting) had 3 CPU versions labeled V1 V2 V3 and those had to match for it to use both.

When the replacement arrives I'll install it and post my results here.
 
I found the single CPU card on Ebay. There are many quads on Ebay that may or may not work or with hidden liquid damage but all is needed is a CPU card. After talking with the seller it should work but has to survive shipment. It was the only CPU card the seller had. If the CPU card doesn't work I'll have to get a whole quad for parts and hope the CPU cards are ok and use the matching (EEE code) pair.
 
Actually @amishallin, the fact that there are a lot of Quads on eBay, whether their LCS' are damaged or not, is a real blessing. Now that air cooling is something that we can all do, we don't need to care about the state of the LCS. Certainly in my case, all I need is the CPU cards. The LCS itself is now irrelevant to me.

So, I am making a different gamble, and may buy one of those "for parts or repair" Quads you see from time to time on eBay and hope that at least one of its CPU cards are still operative.
 
@amishallin, your photos are very good! Thanks for your gracious permission; I will use a few of them in an updated version of the guide, giving you credit of course.

Folks, I have updated the Air Cooling Conversion guide to v0.4. This version includes some additional and excellent photos contributed by @amishallin - thanks @amishallin!

I have updated the link in the original post and reproduce that link below for your convenience:

https://inverary.net/LCS/Power Mac G5 Quad Air Cooling, A-Z Guide, v0.4.pdf
 
thanks for update! I just secured two FROZN A400. ;-)

Excellent! Best of luck with your conversion!

If you are using the guide to direct what you are doing, and you see any areas that could be made clearer, or any areas that would benefit from additional detail, don't hesitate to let me know. This will just make the guide better for all the rest of us here at MacRumors.
 
If you are using the guide to direct what you are doing, and you see any areas that could be made clearer, or any areas that would benefit from additional detail, don't hesitate to let me know. This will just make the guide better for all the rest of us here at MacRumors.
Yes of course! If I find something I let you know for sure.
But I will work on Quad cooling somewhere in September or even later:
I just finished Pegasos 2 CPU swap from 7447/1000 to 7447A/1666 - I was afraid about it, because there was no documentation about VCore. Luckily I succeed with correct VCore change and now I have a lot of tests and overclocking in front.
Second in line is my Powermac G4: CPU swap for 7457 versions, overclocking, flashing of FireGL X3 card
And after that Quad will come up next ;-)

Thanks again for your nice guides!
 
Quick question, what version of CHUD Tools is recommended/compatible with Late 2005 G5s and Leopard? I tried Xcode 3.1, but the "Processor"/"Hardware" preference pane is nowhere to be seen.
 
So the eBay purchased CPU card to replace CPU B finally arrived from Australia. And let me tell you what that box was absolutely MANGLED. However, the CPU card inside was packaged well enough to have survived the rough international journey and customs.

One thing I've noticed is the drastic temperature difference between CPU A/B and especially with CPU B core 2. There is definitely something weird going on. CPU A continues to run HOT while CPU B core 1 looks to have the ideal temp that is most comparable to when it was liquid cooled.

Attached are some images:
Image 1: The completed setup! minus the air baffle but that still fits. My existing CPU was designed for version2 of the LCS and the replacement for CPU B was designed for the version1 LCS cooler. I used double sided 3M tape to attach the bottom fan to the top fan.
Image 2: The system correctly identifies as a quad again.
Image 3: The temps sitting on desktop with reduced CPU power.
Image 4: The temps sitting on the desktop with highest CPU power.
Image 5: The temps doing a benchmark in reduced performance mode.
Image 6: The temps doing a benchmark in highest performance mode. Notice the temps of CPU A vs CPU B.
Image 7: This was taken a little bit after image 6 (notice the time). For whatever reason the temps, watts and amps dropped for unknown reasons. There was no noise or anything noticeable that happened. The numbers remained steady until the test completed.
Image 8: The results of the benchmark using highest CPU performance. #6 is the original benchmark after I rebuilt the LCS 2 years ago. #7 is the result of both CPUs being air cooled. #8 is a single CPU at highest performance and #9 is single CPU at reduced performance.
Image 9: Was taken once the test running for image 6 completed and everything normalized for idle.
Image 10: Fully put back together! The reflective air baffle and light from outside was confusing my camera.
 

Attachments

  • image1.jpg
    image1.jpg
    535.5 KB · Views: 18
  • image9.jpg
    image9.jpg
    376.8 KB · Views: 17
  • image8.jpg
    image8.jpg
    751.2 KB · Views: 17
  • image7.jpg
    image7.jpg
    323.9 KB · Views: 13
  • image6.jpg
    image6.jpg
    302.7 KB · Views: 12
  • image5.jpg
    image5.jpg
    353 KB · Views: 12
  • image4.jpg
    image4.jpg
    389.7 KB · Views: 11
  • image3.jpg
    image3.jpg
    377.5 KB · Views: 14
  • image2.jpg
    image2.jpg
    911.1 KB · Views: 13
  • image10.jpg
    image10.jpg
    260.1 KB · Views: 14
Last edited:
I just wonder if anyone has swapped CPU cards like I have before. The funny, and obvious thing, is that each CPU card was made for a different LCS (version 1 and version 2).

Sure that ominous red light is gone but the temps of CPU A being more than 10C hotter than CPU B during sustained 100% load bothers me now. I may do a remount of that cooler since reaching 91C is a little hot for these and it wasn't getting that hot before (topped out at 84C MAX). Maybe it didn't have a stable mount and I bumped it or something.

Then there is the comically low temps being reported from CPU B core 2... single digits during idle.

The wifi card still isn't detected but the bluetooth is detected and works. May have to look for a replacement.
 
I just wonder if anyone has swapped CPU cards like I have before. The funny, and obvious thing, is that each CPU card was made for a different LCS (version 1 and version 2).

Sure that ominous red light is gone but the temps of CPU A being more than 10C hotter than CPU B during sustained 100% load bothers me now. I may do a remount of that cooler since reaching 91C is a little hot for these and it wasn't getting that hot before (topped out at 84C MAX). Maybe it didn't have a stable mount and I bumped it or something.

Then there is the comically low temps being reported from CPU B core 2... single digits during idle.

The wifi card still isn't detected but the bluetooth is detected and works. May have to look for a replacement.

Re: CPU A being hotter, I’m curious what temps you would observe if you turned off CPU B in Open Firmware. I wonder if heat from the CPU B card is warming up that heatsink and offsetting some of its capacity to shed heat.

No idea what’s going on with that CPU sensor, obviously the numbers are wrong. Maybe the sensor has thermal paste or old coolant on it? I wouldn’t worry about it too much given the sensor for the other core seems to be functional.
 
Comical, indeed @amishallin. I hadn't seen the temperatures photo until this AM... lots of access issues yesterday, with only spotty connectivity. A CPU running at 4C is ... well ... pretty unlikely.

However, I have swapped CPU cards from one Quad to another Quad. The CPU cards now powering my Air Quad came from my LCS-cooled Quad, and visa versa. The swap was an experiment, but since it worked, I left well enough alone. So yes, you can do this.

My LCS-cooled Quad has the original CPU cards from what is now my AirQuad, and I have not redone the thermal calibration. However, both the CPU card intake and exit fans are now hard wired through a fan control card I bought, vs. being controlled by Apple hardware and software, so thermal calibration is actually irrelevant.
 
Found out why CPU A was running hotter after installing CPU B. I somehow got the fins of the cooler under the fins for CPU A. After gently moving the coolers away from each other just enough so they aren't overlapping I took the image below. The machine booted up just fine and the temps remained mostly the same with CPU B running a little cooler. However, when I ran a benchmark (reduced performance) CPU A core 1 was now running an additional 15C hotter than CPU A core 2. After letting everything cool back down to idle temps I shut it off and tightened the front/back mounting screws on both coolers about 1/4 turn. Now the overtemp light comes on right away when pressing the power button. I'm may have to modify the fins so they don't touch as shown in the image taken.
 

Attachments

  • 20250704_100523.jpg
    20250704_100523.jpg
    372 KB · Views: 7
Now the overtemp light comes on right away when pressing the power button.

Well, that is too bad. You are not "out of the woods" yet!

I am going to guess that the overtemp issue has nothing to do with that quarter turn on the mounting screws and instead results from the physical realignment of the two coolers to get the fins to clear each other.

My guess: the thermal paste "seal" between the cooler/shim assembly and the face of the CPU has been disturbed. You may need to repaste/remount both CPU cards to correct this...
 
Yes I could try repaste/remount them but I was thinking of taking the logic board out and replacing the front mounting pegs so the brackets are the same height. That should allow the coolers to be mounted flat and possibly eliminate the risk of damaging a CPU like I did earlier with the front being 1/16" lower than the back. I've taken it out before when I did the complete overhaul and repasted the heatsink on the back. It will be a pain and I think it's possible but it will have to wait for another time. When I do it I'll take plenty of pictures.
 
Indeed @amishallin, you CAN remove the logic board, but if you are going to do that anyway, you might as well try the repaste approach first. Infinitely simpler, and you have nothing to lose!
 
Yeah I suppose I could try repaste CPU A after modifying the fins. Could probably just mill 1/8" off the side at work and be done with it. At least they wouldn't be under so much pressure against each other or overlap. CPU B had good temps so I'll leave that be.
 
I'll say this @amishallin, your experience makes me less eager to try to add a second CPU card to my AirQuad. In its current 2 CPU configuration it is 100% stable, cool and quiet. I am reluctant to disturb that, but I have to admit that if I can find new CPU cards, I will try it!

Right now I have two "half Quads" (one LCS-cooled and one Air-cooled) that are 100% stable, functional, cool and quiet. Both are running just one CPU card (2 cores) so they are effectively 2.5 GHz Duals. I am now using the AirQuad one of these two for porting v4.4 of my VE text editor (Terminal-based) to Sorbet. I will absolutely finish that work before I attempt the adding of a second CPU card to the Air Quad.
 
Last edited:
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.