Overheating problems and MFC- Check Important Screws!

[mfearon]

Summary- Check that important screws are present and tight!

I just replaced the logic board in an old MacBookPro4,1 with overheating problems.
It has now become a MacBookPro3,1 -but no overheating problems!
Macs Fan Control can help with overheating, but only up to a point.
If you have overheating, here's some steps to take to clean out your macbook.

Gently blow and/or vacuum out the dust in back vent areas near the hinge.
For an old one, opening the computer to clean out the innards and
blow/vacuum/scrape years of embedded dust/hair from inside may work wonders-
Ifixit has good tutorials on this: https://www.ifixit.com/Guide/MacBook+Pr ... cement/681

In my case, I found that crucial screws had been left out at steps 4*, 5, 6, and 7-
I didn't even have to open the case -as it turned out- as dust wasn't the problem.
These screws (esp. step 4!) are important in preventing overheating,
because they tighten the circuit boards, heatsinks, etc, to the metal case
in order to cool the computer properly- if screws aren't snug, parts can overheat!
Sets of replacement screws can be found an Amazon and Ebay.

If you proceed to pop the keyboard and go on to step 9, the fans can be inspected
and much of the logic-board-topside cleaned of hair & dust.

To remove the fans to clean under the logic board --WITHOUT removing it--
it is necessary to proceed to step 20-
NOTE- in step 19 you don't have to remove all 15 screws(!),
just the 6 marked YELLOW and GREEN!

If you remove the logic board, step 21, you can inspect & clean everything really well
-without removing the heatsink/pipes-
BUT AGAIN PLEASE NOTE in step 19: "Eight 4.7 mm silver T6 Torx screws..."
-- ONLY remove 3 of 8!!!
the ones to remove are the 3 closest to the front edge of the computer--
the other 5 hold the heatsink/pipe and the logic-board-CPU-GPU together as one unit-
there is a note to this effect in step 20(!) Unless you have some
"Thermal Paste" or "Compound", you DON'T want the heatsink/pipe to come loose!
This could reduce/break the "thermal contact" to the heatsink/pipe-
causing a new overheating problem!

With the logic board removed, you can inspect the GPU, CPU and circuit board area
nearby for discoloration due to overheating, or replacement/repair, of the GPU-
the GPU is closest to the RHS fan, CPU to the LHS fan.
In between is the Memory Controller (MEM).

The last step is to remove the heatsink/pipe from the logic board,
by removing the last 5 screws from step 19. Before replacing the thermal paste,
inspect the areas of old thermal paste for any "bubbles" or spots of bare metal-
these may have caused the overheating problem-
by preventing a good thermal contact between the heatsink/pipe and the CPU & GPU.
A good thermal contact could also been lost at an earlier time by:
(1) dropping the laptop, but most likely
(2) a technician who didn't properly clean and replace the thermal compound after
removing the heatsink/pipe- it is crucial that the thermal paste be replaced properly!
The Ifixit instructions referenced on step 21 about "Applying Thermal Paste" are:
https://www.ifixit.com/Guide/Applying+Thermal+Paste/744

The MacBook Pro logic boards for the MacBookPro3,1 (A1226, 820-2101-A) and
MacBookPro4,1 (A1260, 820-2249-A) were notorious for problems with the GPU,
thought to be caused by solder joint failures to the GPU,
from heating/cooling stress, dropping the laptop, etc.
Many were repaired by resoldering or replacing the GPU IC,
and there are repair services with the equipment to "reflow solder" the GPU
on your logic board- I did not have good luck with such a service.
Other people have tried a DIY approach to reflow, baking it in the oven:
http://russell.heistuman.com/2010/04/27 ... gic-board/
Others don't bother removing the logic board- just leave the whole laptop
powered-up, wrapped in a blanket, upside-down, hot for several hours(!)
I haven't tried it- can't recommend it- it seems like a fire hazard- or a weird joke:
https://www.ifixit.com/Answers/View/904 ... video+chip

Good luck!

[Steven]

mfearon
Thanks for sharing this information!

[fbl]

mfearon
As a MacBook Pro 3.1 owner, using MFC when running Windows 7 SP1 32 bit Pro, via Bootcamp 4.0, I read your information with interest, having recently cleaned out the inside of my machine.

Possibly because I am running Mac OS 10.6.8, I was unable to run the runTest process that Steven has requested, which would tell me the maximum temperatures permitted with my hardware - did you have any success with that? Could you share the values you found? I'd be particularly interested in the CPU Diode and GPU values.

Thanks.

[Steven]

fbl
Max temp for your CPU is 100C

[mfearon]

I run Xplane to stress the MacBookPro3,1 -I'll look into runTest!
When I use the "Automatic (controlled by system)" values,
I can get CPU and GPU Diode temperature of 180 degF,
and the fans tend to repeatedly "surge" over time.
I can easily set up "Sensor-based Value" for the L & R fan
that will keep the CPU and GPU Diode temperature around 140 degF,
and the fans run at a steadier speed.

Last January I was trying to use MFC to get another logic board stable
-it probably had a GPU with damaged BGA solder joints-
the logic board would fail at GPU Diode temperatures around 140 degF.
The replacement $2.24 logic board seems stable over a good solid range.
The "Automatic (controlled by system)" fan controls can allow Diode
temperatures to get really really high before the fans blast.

A better use for MFC than CPU/GPU are areas of the computer
more sensitive to heat, and not designed to endure it.
I've been having rotten luck with the DVD drive in a MacMini3,1
that is used for a home entertainment system.
Sometimes the disks come out hot!
MFC may allow internal drives like this to last longer,
even if they do not have their own temperature sensor,
by keeping the inside of the MacMini from getting too high.

[fbl]

As runTest requires at least Mac OS 10.8, hopefully you'll be able to get it to run on your Yosemite system. I'm still using 10.6.8, because subsequent OS versions seem to me to be more restrictive and less capable, in what I take to be Apple's drive to become less computer- and more phone-like.

My MBP 3.1 also serves as a media centre for me, when it's not being used for location music recording. The tasks I give it under Mac OS 10.6.8 occasionally see the CPU getting over 90 C (194 F), but it usually hovers around 60 to 70 C (roughly 140 to 160 F) when idling.

When I run Windows 7, it's primarily to run Pyramix 9, a multi-track audio workstation. Just starting this program puts the CPU diode temp up to 90 C. I started with MFC set with both fans responding to the CPU diode, with the temperature range set at 84 to 94 C (183 to 200 F). I've since modified this scheme so that the right fan responds to the GPU diode with a range of 75 to 85 C (167 to 185 F), while at the same time doing everything to keep the system running a cool as possible - no Aero display scheme, and limiting graphic display demands from the program. Plus, I use a laptop cooler with two fans powered by USB. With these measures in place, I can keep the maximum CPU temp limited to around 75 to 80 C, which I'm treating as the "sweet spot" between restricted capability and meltdown.

This machine has a checkered past, having had three motherboard replacements (since the GPU is mounted right on it) done to correct display problems. When the second fix failed after three weeks, it was determined that the only way to be sure was to replace the motherboard and the screen, plus all connections between them. From that point on, it's been rock solid. Consequently, the only original parts left are the lower case, keyboard and the optical drive, which has performed without issue for six years, never showing the temperature issue you describe. Go figure.

My thinking is that in these machines, everything is so jammed into such a restricted space that just about any MFC cooling scheme is going to have more or less the same effect, so whether both fans react to the same sensor, or to different ones, the results are more or less the same. The primary sources of heat, as far as I can see, are the CPU and GPU.

[fbl]

fbl
Max temp for your CPU is 100C

Thanks Steven, that's just what I had guessed.

[mfearon]

Trying “runTest" in Yosemite- the console log sez:
ERROR invalid signature for crystalidea.macsfancontrol.kext, will not load
The chown seemed to do OK...
It sounds like fbl got an answer of
100degC for a CPU in a MacBookPro3,1 -hot!

My GPU is nearly always hotter than the CPU-
the mysterious part is even when it has been asleep overnight-
I would have expected all the sensors to start the same- room temp;
my GPU diode starts out at 93 degF -when everything else is 73degF.
Can’t do this from a cold-boot as things are hot by the time
MFC and other programs get going!

The tasks I give it under Mac OS 10.6.8 occasionally see the CPU getting over
90 C (194 F), but it usually hovers around 60 to 70 C (roughly 140 to 160 F) when idling.

This MacBookPro3,1 is a 2.2GHz T7500, Logic Board- 820-2101-A (c) 2007
Everymac Identifiers: Mid/Late 2007, 2.4/2.2GHz* -MA895LL -MacBookPro3,1 -A1226 -EMC=2136

Max temperatures when the system is idling w/ display on are:
CPU diode =110degF GPU diode=130degF
I like to look at the difference between the diode temp and nearby heatsink temp,
to see if I put the thermal compound on properly:
CPU heatsink 8degF below the CPU diode (when idling)
GPU heatsink 15degF below the GPU diode
(Maybe this indicates the GPU didn’t get properly “thermaled”!)

...run Pyramix 9, a multi-track audio workstation. Just starting this program puts the
CPU diode temp up to 90 C. I started with MFC set with both fans responding to the
CPU diode, with the temperature range set at 84 to 94 C (183 to 200 F).
...I can keep the maximum CPU temp limited to around 75 to 80 C…

The MFC strategy I came up with was to start the fans just above the idle temp,
and have them max out at the top temperature I was hoping to stay below.
This would allow the fans to return to idle at the min fan speed.
The temperature range I came up with for this had these properties:
1. Assigned the L fan to the CPU, and the R fan to the GPU,
as these are the primary heat sources to each fan, even though heat from one can flow to the other.
2. The idle temp plus a few degrees was set as the:
Temperature that fan will start to increase from: L=CPU=104degF, R=GPU=122degF
3. The temp to stay below minus a few degrees was set as the:
Maximum temperature:: CPU =140degF, R=GPU=149degF
I can probably go much higher than this, based on a 100degC max for the CPU.

When the system is running iTunes playing & converting an aiff->aac,
QT playing an mp4 movie and Xplane w/ lots FS graphics going on:
CPU diode 208degF138 GPU diode 163degF143
CPU heatsink 70degF below the CPU diode (!)
GPU heatsink 20degF below the GPU diode
The system Activity Monitor showed both CPUs close to being maxed out.
Perhaps I can stress the GPU further by hooking up the ext. monitor port.

...the optical drive, which has performed without issue for six years,
never showing the temperature issue you describe. Go figure.
My thinking is that in these machines, everything is so jammed into such a restricted space
....The primary sources of heat, as far as I can see, are the CPU and GPU.

Most of my recent experience has been with MacMinis,
in which the DVD drive is right above all the heat producing circuitry!
The MacBookPro3,1 has the DVD in a cool corner, which should be fine.

I don’t have much experience using a laptop,
but I have put together a hackintosh D820, and got this MacBookPro3,1 working well.
It is interesting to hear your experiences, having kept your MacBookPro3,1 running
thru thick and thin for so many years- and it will run MacOS 10.10 with probably
several more years of security updates at least! I got this MacBookPro3,1 for
my daughter, whose old Cube has been stuck at 10.6.8 for a long time,
and security updates stopped last year.

[Steven]

mfearon
The kext is not signed, but I had no problems with loading it under OS X prior to Yousemite (the only issue was OS X warning)

[fbl]

It sounds like fbl got an answer of
100degC for a CPU in a MacBookPro3,1 -hot!

I got that number from Steven further up this thread, as I can't run runTest on my 10.6.8 machine. However, experience with temperature reporting software under both the Mac OS and Windows (XP and now 7), led me to assume that this was the case. As you've pointed out, these machines run hot - although not as hot as the Apple laptops just a couple of generations before, which had the reputation of being too hot to run them sitting in your lap. I fully expect the Minis to have the same issues - too much heat producing hardware crammed into a very tight, albeit stylish, space.

My GPU is nearly always hotter than the CPU-
the mysterious part is even when it has been asleep overnight-

GPUs generally run pretty hot, and in recent years there has been a move among programmers to offload some tasks, even if they're not graphics related, from the main CPU to the GPU because it's usually so fast. Add in the fact that they are generally smaller physically than a CPU package, and you wind up with a busier processor generating more heat in less physical space. It doesn't surprise me that yours gets hot so quickly, although you might want to recheck your heat-sink mounting and whether the thermal coupling goop got distributed properly.

This MacBookPro3,1 is a 2.2GHz T7500, Logic Board- 820-2101-A (c) 2007

Mine runs at 2.4 GHz., but otherwise, I believe it's the same model.

Max temperatures when the system is idling w/ display on are:
CPU diode =110degF GPU diode=130degF

Those don't seem like unreasonable numbers to me.

The MFC strategy I came up with was to start the fans just above the idle temp,
and have them max out at the top temperature I was hoping to stay below.

Well, isn't that the crux of the issue: the trade-off between improved cooling to extend the life of the processors, while not over-using the fans to preserve their life. That said, I'd rather pay to replace a couple of fans, than to have to replace the whole motherboard (again!) because one or both processors have burnt out. I'm hopeful that with care, neither situation will occur for a good long time.

the Temperature that fan will start to increase from: L=CPU=104degF, R=GPU=122degF
The temp to stay below minus a few degrees was set as the:
Maximum temperature:: CPU =140degF, R=GPU=149degF

So,
Left 104 to 140 F, or 40 to 65 C; and Right 122 to 149 F, or 50 to 65 C.

This is certainly a lot lower than what I've set up for my machine under Windows 7/Bootcamp 4 where the values I'm using are:

Left (CPU Diode) 83 to 93 C, or 181 to 199 F; and Right (GPU Diode) 75 to 85 C, or 167 to 185 F.

In Windows 7, I've adopted the use of a "Classic" display theme, rather than a more processor hungry "Aero" one. With all this in place, my machine will idle with the CPU Diode hovering around 59 C (138 F), with the GPU around 67 C (153 F). As long as I don't use any processor intensive plug-ins (which I'm not too likely to need while doing location recordings, i.e., capture only), I can run Pyramix for hours with the CPU Diode hovering around 80 C (176 F), which really makes me feel far better about using this machine for things like concert recordings where I only get one kick at the cat.

I can probably go much higher than this, based on a 100degC max for the CPU.

Well, that's been my experience so far, but of course, YMMV. I think your idea of running an audio format conversion at the same time as a graphics intensive process is the way to max out your machine for testing purposes. I can always expect to see the fans come on under the Mac OS whenever I'm using XLD to convert FLAC files to ALACs. When I do this, I take care to boost my fan speeds ahead of time (currently using SMC Fan Control on the Mac OS, because I'm concerned about the memory leaks and other issues reported for using MFC under Mac OS 10.6..

Other things to watch out for include whether the battery is charging or not (always good for an extra 5 or 6 degrees C), and wireless networking to carry a lot of internet traffic. I got myself a nice long Ethernet cable, and now my machine runs much cooler for having Bluetooth and Airport turned off. Running Plex to handle media (TV shows and Movies), driving my 720p TV and my surround sound amplifier, I see CPU Diode temps hovering around 65 to 70 C most of the time. Of course, in all cases, I'm using an active laptop cooler which has a couple of fans and runs off USB power.

It is interesting to hear your experiences, having kept your MacBookPro3,1 running thru thick and thin for so many years- and it will run MacOS 10.10 with probably several more years of security updates at least!

I have two machines, the other is a 2006 24" iMac 6.1. I really have no desire to update either of them beyond Snow Leopard, and as I've just dragged myself kicking and screaming from WinXP to 7, I think that's all the upgrading I can stand for the moment.

[mfearon]

The kext is not signed, but I had no problems with
loading it under OS X prior to Yosemite (the only issue was OS X warning)

In 10.9.4 it is only a warning, but...
Intel(R) Core(TM)2 Duo CPU T7500 @ 2.20GHz
MacBookPro3,1
ctl_id: 0x7 for ctl_name: crystalidea.macsfancontrol.kext
TjMax T: 0

[Steven]

mfearon
It failed, but TxMax T for T7500 is 100