There have been many
reports of
coolant leaks and subsequent damage to Apple G5 PowerMac computers. There is also
an effort to create a
database in order to document this problem and get help.
A recent very detailed investigation by
EagleFour
gave me the final push to investigate my own 2 X 2.5 GHz PowerMac.
Background
I got the machine new from a reseller in summer 2005. It was
manufactured in Feb 2005 in Cork, Ireland. It runs continuously at full
load, performing scientific calculations.
Because of the many web reports on G5 coolant leaks, I recently decided to open
mine and check for problems. This was also motivated by white deposits on the rear "foot" of
the case. At the time I was not sure if this was due to leaks, now I am sure!
Disassembly
I followed the procedure of
EagleFour
and the Apple guide, which is linked on the EagleFour site. The only
difference was that I also removed the bracket stabilizing the heat
pipe radiators, at the rear of the CPU/cooler unit. This gave better
access to the rear center screws holding the plate to the chassis.
CPU
HeatSinks
EagleFour found colored solid deposits on the edges of the CPU heat
sinks. I found this as well, but the amounts were smaller. I removed
them with a small screwdriver, as well as possible. Solid material had
not become so extensive as to contact components on the circuit board
below.
Power
Supply
EagleFour either did not investigate the power supply, or it was
OK. Due to many reports of "exploding" power supplies (loud bang), I
decided this was the time to check it.
There was a gray, flaky, multilayer absorbent material on the power
supply cover plate
below the CPUs. This is apparently the "diaper" noted
by others. When I went to unfasten the screw below it, I noticed
corrosion on the rear of the plate, under the plastic holding the
diaper. Now I knew I had a problem!
Power supply cover plate partly
removed, in the foreground. Note the
corrosion under the plastic support of the gray absorbent fabric
("diaper"), at
the
rear of the plate. Also visible is corrosion on the top of the power
supply itself, around the cable
(which has been detached from the logic board).
Another view with no flash to better
appreciate the corrosion and absorbent fabric.
Next I removed the power supply itself. The appeareance suggests a
minor problem:
Bottom of the power supply, showing how
leaking coolant had spread
under and around
it, leading to extensive rust. Note the white deposits on the edges,
this
appears
to be from the coolant.
Rear of the power supply, showing
how leaking coolant had spread
up the perforated region as well.
Now I feared the power supply was dangerously damaged, and opened it
with trepidation. Fortunately (for safety reasons!) the liquid had not yet reached any
components on the circuit board. However, it was very close- a mounting screw was
badly corroded, it was probably not long until electronic components were damaged.
Inside of power supply.
CPU /
Cooling Block and Pump Leak
Initially I could not understand how such extensive corrosion had
occured. It was difficult to see until the CPU and power supply
cover plate
were removed. Inspecting the CPU / cooler unit, I also could not find
any obvious source of such extensive leakage. The small leaks on the
CPU heat sinks were clearly insufficient, and no residue was present on
the circuit board. I think a serious leak here would quickly destroy
the CPUs.
Only when I inspected the block more carefully did I notice the path
of the leaking coolant:
Liquid track beneath coolant pump.
Different light shows the white residue
similar to that on the power supply case. The aluminum is not corroded
by the fluid, making the track
less apparent than on the steel power supply case.
Cause of
the Leak and Power Supply Corrosion
Problem 1
The immediate cause was simple:
>>> A
pump gasket failed <<<
Rubber gasket fragments found under the pump.
Problem 2
A major contributing factor to the extensive damage was that the
escaping liquid was not absorbed by the fabric.
This was evident from the fact that the absorbent fabric was not wet, nor did it
have any coolant residue
on it. In fact, there was corrosion
under the plastic backing of the absorbent fabric. Why did it not catch the fluid and prevent it from reaching the
power supply?
>>> The
absorbent material does not extend
under the leaking pump <<<
The liquid dripping under the pump does not fall on the fabric,
but instead drains down the side of the power supply. This is a clear
design error. This potential source of leakage was obviously not taken
into
account when sizing and positioning the fabric.
Final
Comments
Like many others, this experience has left me very disappointed. For
years, a justification for the substantially higher price of Apple
computers was their quality and durability. I have 2 older Macs, both
run, and one calculates 24 / 7. On the other hand it is inexcusably
noisy, a Quicksilver 2 X 1 GHz G4, upgraded to 2 X 1.33 GHz. This is
also a major design flaw, which Apple "fixed" with liquid cooling in
some G5 models.
My experience, and that of others, suggests that owners of liquid
cooled G5 computers should take them apart to check for hidden
corrosion damage. This inconvenient (Take out the processors to reach
the power supply?! Difficult and not good design.) and carries some
minor risk, but could avoid catastrophic and very expensive failures
and repairs.
Will I buy a Mac again? I dont know. I got a Mini not long ago (3rd Mac
doing calculations), because it seemed the only element of the Apple
line which had a good price/performance ratio. Then I had to upgrade
the processor and memory anyway, but at least it was possible. And
rumors persist that Apple wants to kill the Mini.
Im starting to think that I should invest in Linux. And Apple has
helped me by providing training on a unix-based OS!