[serr]

Has anyone gotten to the bottom of this thermal calibration error with the G5? I see lots of discussion of a possible bug in ASD thermal calibration and claims that it can destroy the eeprom that stores the calibration info on the logic board.
So what's the bottom line here?
Does this error in reality indicate that the eeprom is no longer writable?
Could this simply be an issue of the inner door not sealed properly (perhaps this calibration is critically sensitive to this)?

The fans are racing as intended since the CPU's have been replaced and the calibration stored on the logic board is no longer correct. Everything is fully functional and passes all tests. The inner shield is in place and the sensor (by the rightmost tab) sees it there. The intake fan is running. I also tried another fan with no change. I suspect that some overly high sensitivity to the fan speed is not really the issue here.

I have a dual 2.5GHz G5. The liquid cooling system is completely refurbished and the system passes ASD tests. I currently have ASD 2.5.8. I see reports of success using this version so there must be some hardware related issue here.

I'd be most grateful if someone would share the trick or workaround for this.

[reader50]

Did you run calibration in a room at 72 degrees? You might be getting the error because room temp is significantly higher or lower than expected. The service disk should include instructions on the test setup. And check the specific error against the results guide. Note: you didn't tell us the specific error, which kinda helps.

[serr]

Originally Posted by reader50 

Did you run calibration in a room at 72 degrees? You might be getting the error because room temp is significantly higher or lower than expected. The service disk should include instructions on the test setup. And check the specific error against the results guide. Note: you didn't tell us the specific error, which kinda helps.

The exact message is:
"CPU Intake Fan Speed Error. Check that inner door is secure and in place then rerun calibration. "
This is reported for both CPU's.
The room is about 68 F. Does this need to be climate controlled to EXACTLY 72F? I thought the instructions said "below 77F". There is a specific error message for environment too hot (Ambient Max. temperature exceeded. Move unit to cooler environment and retest.) and I'm not getting that.

[reader50]

Check for lint clogging the radiator fins. It probably means the fans were revved all the way up, but the CPU temps did not come down as expected. So it assumes the inner door is removed, allowing the airflow to skip the radiator assembly. With the inner door in place, an air blockage would be the next thing to look for.

Besides the radiator, check the case holes front and back for lint / grime buildup. You should be able to feel the air coming out the back of the CPU zone. If you can't feel air with the fans revved up, then locate the blockage.

[serr]

Originally Posted by reader50 

Check for lint clogging the radiator fins. It probably means the fans were revved all the way up, but the CPU temps did not come down as expected. So it assumes the inner door is removed, allowing the airflow to skip the radiator assembly. With the inner door in place, an air blockage would be the next thing to look for.

Besides the radiator, check the case holes front and back for lint / grime buildup. You should be able to feel the air coming out the back of the CPU zone. If you can't feel air with the fans revved up, then locate the blockage.

First, thanks for helping!
Well, I blew compressed air through everything before I reassembled it. It's not sterilized but it's pretty clean and you can feel a healthy gust of wind coming out the back.
I should mention that I have 2 identical G5 dual 2.5's and they both fail this calibration as described. I rebuilt these from 4 machines. I got 1 working CPU from each machine so there's no way to use an original configuration with the original calibration. My first web search made me say "oh no!" as this seems to be an issue many have had - that's why I mentioned the rumors I've read.
Should I try this in a 50F or 60F room?
When I fire up OSX and run temp monitor, it shows the CPU's at 38C, 32C for unit #1 and 29C, 30C for unit #2.

[reader50]

Your temps sound fine for testing. Check your gaps between fans and CPU modules - is there a big enough gap for air to avoid going through the radiator assembly? Air will take the path of least resistance. If it can, it will escape early and skip the radiators.

Hope someone else will comment - I haven't had to do a thermal calibration yet, so my answers are based on theory.

[Leonard]

Well, I'm just guessing here, but if the fans are working and the CPUs are working, how about what's in between them. The liquid cooling system and the connection of the liquid cooling system to the CPU. You're sure the pump or pumps on the liquid cooling system (lcs) are working? Was there a thermal paste between the lcs and the CPU, and was this replaced when you reassembled it?

The plastic inside shield/door - I believe you said the tab was ok, but just to check it is solid and not clear so it blocks the sensor right?

[serr]

Originally Posted by Leonard 

Well, I'm just guessing here, but if the fans are working and the CPUs are working, how about what's in between them. The liquid cooling system and the connection of the liquid cooling system to the CPU. You're sure the pump or pumps on the liquid cooling system (lcs) are working? Was there a thermal paste between the lcs and the CPU, and was this replaced when you reassembled it?

Brand new Swiftec pumps. Arctic Silver 5 thermal paste. Completely cleaned and referbed inside and out. Not the 1st time I've done this. My dual 2.7 G5 has been running flawlessly (and nice and cool) for about 2 years now (24/7) and periodic inspections show now problems. I trust my work better than the factory job.

Originally Posted by Leonard 

The plastic inside shield/door - I believe you said the tab was ok, but just to check it is solid and not clear so it blocks the sensor right?

Check

I've had experience before with programmable chips that are a bit flakey and need slightly different power for write mode. Could that kind of madness be going on with these logic boards with the ROM chip in question? Or is it more of a trick with setup conditions (temperatures, etc).