MBP and the Thermal Paste Discussions
I've read with interest the postings on this web site regarding the MBP heat issues and the possibility that too much thermal paste is causing that problem. I've spent quite a few years designing computer motherboards using Intel processor chips, and want to share some thoughts on the topic. I only mention my background because one thing this business has taught me is that we can all be easily mislead by our test data if our process for gathering the data isn't rigorous (and, believe me, I've learned that lesson many times).
First, the excess thermal paste is definitely a problem, regardless of whether it is the root cause of the heat issue. Some of the photos posted on the various web sites show thermal paste covering bypass capacitors on the chip carriers (the green fiberglass boards that carry the CPU and GPU die). Thermal pastes can cause stray capacitances that can interfere with circuit operation. Standard industry practice is to use only as much paste as is necessary to achieve a layer that is about as thin as a single layer of tissue paper (or less). The amount used by Apple is far too large, particularly because the paste in coming into contact with other electrical components and conductors. A common failure mode associated with degraded bypass capacitors is random (and unrepeatable) crashes due to corrupted digital data in the processor chip.
In addition, all thermal pastes contain various types of fluids. Over time, these fluids evaporate, and frequently they condense elsewhere inside a computer. When CD drives were first introduced years ago, a common cause of mid-lifetime failures was condensates on the optics of the laser. The condensation fogs the lens and interferes with the ability of the laser to focus while reading and writing data. The contamination was traced to outgassing by thermal pastes, greases, sulfurized elastomers, and similar items used to build the computer. (Note that the Apple service manual for the MBP refers to the thermal paste as “thermal grease”.) Consumer electronics manufacturers learned to minimize the use of such materials in computer designs. The amount of thermal paste shown in the Apple service manual is very large, and I wouldn't rule out that it might contribute to future DVD drive failures.
It is difficult to say, based on the available data, that reducing the amount of thermal paste actually contributes to improving the thermal behavior of the MBP. Part of the problem is that the procedure to remove the pastes involves disassembling the laptop, and this introduces the possibility that the disassembly/reassembly process is inadvertently fixing the problem.
First, the high case temperatures could be the result of loose or poorly placed thermal sensors. If the disassembly/reassembly process fixed a loose sensor, or loose sensor electrical connection, or placed the sensor in a better position, you would get lower case temperatures, and possibly (inadvertently) attribute this to less paste.
Second, the cooling system design in the MBP requires that the die caps on the chip carriers be pressed tightly against the pads on the heat sink, so that heat is transferred efficiently. If the MBP is not assembled properly, (i.e., if the logic board mounting screws are not properly torqued, or the logic board is not mechanically aligned properly to ensure a tight fit to the heat sink), it is possible that this will manifest itself as high case temperatures. In the Intel processor chips there are two paths for heat to travel – out of the die cap, and through the electrical contacts to the motherboard. If the die cap isn’t properly cooled, heat will instead go into the motherboard, and the case temperature will rise. There is a possibility that the disassembly/reassembly process will fix a loose or misaligned logic board problem, with the result that the case temp will be reduced.
Also, if you work through the details of the thermal model that results from excess paste, it's hard to see how it would result in more heat to the case and less to the heat sink. Most of the photos on the web appear to show thin paste on top of the die, indicating that there was probably a low thermal resistance path to the heat sink prior to disassembly. The excess paste to the sides will simply further reduce the thermal resistance between the chip carrier and the heat sink. I may be missing something, but it's hard to envision how this lower thermal resistivity between the CPU and the heat sink drives more heat into the case. But it is also possible that the disassembly process destroys evidence of the actual connection to the heat sink, so this failure mode is hard to judge.
The best way to determine if removing the excess paste is actually helping would be to measure the exhaust air temperature from the cooling fans before and after removing the excess paste. I haven't seen such data, but if it were taken, and showed that the exhaust air was at a significantly higher temperature after the fix than before it, then we would know that the fix worked. If not, then the heat is going elsewhere, and it is possible that fix has actually made things worse.
Finally, all the symptoms we're seeing can be explained by poor software calibration and/or incorrect software control of the CPU clock speed. I wouldn't rule out a software fix at his point.
I'm not saying removing the paste is not a fix, only that there are still open questions that remain to be answered. It would be hard for me to recommend that anyone rework an MBP for the thermal problem in the absence of additional data.
For myself, I own an MBP, and though I have access to an electronics lab and some great technicians, I intend to buy Applecare, and in the meantime wait for a fix (or at least more information) from Apple.
Best of luck to everyone who has this issue - thats what we get for being early adopters.
G5 2.5 DP Mac OS X (10.4)
G5 2.5 DP Mac OS X (10.4)
G5 2.5 DP Mac OS X (10.4)
G5 2.5 DP Mac OS X (10.4)
G5 2.5 DP Mac OS X (10.4)
G5 2.5 DP Mac OS X (10.4)
First, the excess thermal paste is definitely a problem, regardless of whether it is the root cause of the heat issue. Some of the photos posted on the various web sites show thermal paste covering bypass capacitors on the chip carriers (the green fiberglass boards that carry the CPU and GPU die). Thermal pastes can cause stray capacitances that can interfere with circuit operation. Standard industry practice is to use only as much paste as is necessary to achieve a layer that is about as thin as a single layer of tissue paper (or less). The amount used by Apple is far too large, particularly because the paste in coming into contact with other electrical components and conductors. A common failure mode associated with degraded bypass capacitors is random (and unrepeatable) crashes due to corrupted digital data in the processor chip.
In addition, all thermal pastes contain various types of fluids. Over time, these fluids evaporate, and frequently they condense elsewhere inside a computer. When CD drives were first introduced years ago, a common cause of mid-lifetime failures was condensates on the optics of the laser. The condensation fogs the lens and interferes with the ability of the laser to focus while reading and writing data. The contamination was traced to outgassing by thermal pastes, greases, sulfurized elastomers, and similar items used to build the computer. (Note that the Apple service manual for the MBP refers to the thermal paste as “thermal grease”.) Consumer electronics manufacturers learned to minimize the use of such materials in computer designs. The amount of thermal paste shown in the Apple service manual is very large, and I wouldn't rule out that it might contribute to future DVD drive failures.
It is difficult to say, based on the available data, that reducing the amount of thermal paste actually contributes to improving the thermal behavior of the MBP. Part of the problem is that the procedure to remove the pastes involves disassembling the laptop, and this introduces the possibility that the disassembly/reassembly process is inadvertently fixing the problem.
First, the high case temperatures could be the result of loose or poorly placed thermal sensors. If the disassembly/reassembly process fixed a loose sensor, or loose sensor electrical connection, or placed the sensor in a better position, you would get lower case temperatures, and possibly (inadvertently) attribute this to less paste.
Second, the cooling system design in the MBP requires that the die caps on the chip carriers be pressed tightly against the pads on the heat sink, so that heat is transferred efficiently. If the MBP is not assembled properly, (i.e., if the logic board mounting screws are not properly torqued, or the logic board is not mechanically aligned properly to ensure a tight fit to the heat sink), it is possible that this will manifest itself as high case temperatures. In the Intel processor chips there are two paths for heat to travel – out of the die cap, and through the electrical contacts to the motherboard. If the die cap isn’t properly cooled, heat will instead go into the motherboard, and the case temperature will rise. There is a possibility that the disassembly/reassembly process will fix a loose or misaligned logic board problem, with the result that the case temp will be reduced.
Also, if you work through the details of the thermal model that results from excess paste, it's hard to see how it would result in more heat to the case and less to the heat sink. Most of the photos on the web appear to show thin paste on top of the die, indicating that there was probably a low thermal resistance path to the heat sink prior to disassembly. The excess paste to the sides will simply further reduce the thermal resistance between the chip carrier and the heat sink. I may be missing something, but it's hard to envision how this lower thermal resistivity between the CPU and the heat sink drives more heat into the case. But it is also possible that the disassembly process destroys evidence of the actual connection to the heat sink, so this failure mode is hard to judge.
The best way to determine if removing the excess paste is actually helping would be to measure the exhaust air temperature from the cooling fans before and after removing the excess paste. I haven't seen such data, but if it were taken, and showed that the exhaust air was at a significantly higher temperature after the fix than before it, then we would know that the fix worked. If not, then the heat is going elsewhere, and it is possible that fix has actually made things worse.
Finally, all the symptoms we're seeing can be explained by poor software calibration and/or incorrect software control of the CPU clock speed. I wouldn't rule out a software fix at his point.
I'm not saying removing the paste is not a fix, only that there are still open questions that remain to be answered. It would be hard for me to recommend that anyone rework an MBP for the thermal problem in the absence of additional data.
For myself, I own an MBP, and though I have access to an electronics lab and some great technicians, I intend to buy Applecare, and in the meantime wait for a fix (or at least more information) from Apple.
Best of luck to everyone who has this issue - thats what we get for being early adopters.
G5 2.5 DP Mac OS X (10.4)
G5 2.5 DP Mac OS X (10.4)
G5 2.5 DP Mac OS X (10.4)
G5 2.5 DP Mac OS X (10.4)
G5 2.5 DP Mac OS X (10.4)
G5 2.5 DP Mac OS X (10.4)
MBP 2.0, Mac OS X (10.4)
