Thanks to Greg Koenig for helping me during MacBook Pro surgery...
The MacBook Pro is an awesome machine. It combines the sexy aesthetics of the PowerBook with the power and speed of the Intel Core Duo processor. These performance characteristics have given us the portable Mac OS X machine we've been wanting for the last few years. The screen is bright and beautiful, and the built-in iSight camera is a nice extra touch. It's a worthy upgrade to the previous PowerBook G4 in almost every way. However, there are a few niggling problems with the first rev of the Intel-powered portable, one of which (and probably the most bothersome) is the amount of heat it produces. In fact, although we call it a laptop, it's not very usable for extended periods of time on your lap.
There are many MacBook Pro owners who have taken matters into their own hands, such as Interrupting Moss on the Something Awful website. He describes taking his laptop apart and finding an excess of thermal paste between the main chips on the logic board and the thermal pipes that transfer heat safely away. Several other people have opened up their laptops and found the same thing.
Now, what's the deal with too much thermal paste? That's a good question. In many system thermal-management designs, too much thermal paste apparently interferes with the efficient transfer of heat away from the components. Instead, you want to have just enough between a component and its heat sink to ensure solid contact and no more.
It all makes sense in theory. But you have to ask, why did Apple and their design partners, including Intel, spec so much goo in the first place? After all, they use a sophisticated heat-pipe system and have spent a lot of time optimizing every other part of the design. And these systems are obviously built with quite sophisticated equipment. Why would they ruin a great design with too much paste?
Interrupting Moss and many other folks replaced the messy stock goo with well-executed and much neater applications of high-grade thermal paste. They reported good results--so good in fact, they've tempted a large number of people to try it themselves.
Since I was willing to try anything to cut the heat being transferred into my lap from my MacBook Pro, I decided to give it a try as well. After all, I'm pretty comfortable with Mac hardware. I've cracked many a Mac, including the original Mac mini. I gave myself pretty good odds of not turning my expensive tool into a super-expensive brick. So, I sat down one evening with my friend and neighbor Greg Koenig, who is an industrial designer by trade and a MacBook Pro owner himself, to get to the bottom of things. Since we would operate on my machine, Greg's would serve as our control subject. Both MacBook Pros were updated to the latest and greatest software, including SMC Firmware Update 1.0, which updates the System Management Controller firmware.
Opening up the MacBook Pro was quite an experience. I was excited about seeing what makes it tick, as well as fearful of messing something up. It's not quite like jumping out of an airplane, but there was definitely a thrill to the entire process. After removing the battery, memory, and the top of the case (which includes the keyboard), we were greeted with the following view of the MacBook Pro's innards:
The most delicate item in this part of the process is the ribbon cable connecting the top of the case to the logic board. Next, we moved the right speaker out of the way because its cord runs across the top of the logic board:
You'll notice lots of yellow tape. That's called Kaptop tape, or more accurately, Kapton polymide film. It's a high-tech product made by DuPont that's ultra thin and designed for high-temperature environments. In addition to being used on laptops, it's also a staple of the space program.
Next, out came the fans:
The fans really caught Greg's eye. They're very high-quality in an enclosure that's made mostly out of cast magnesium. Once the fans and the SuperDrive were out, my MacBook Pro looked like this:
At this point, with a full view of all of the inspector's markings and handwriting inside the case from the factory, I was feeling the full thrill of the risky endeavor at hand. I'm not too proud to say that I really wanted a beer at this point--or maybe even something stronger. But there were questions to answer. So we pressed forward, undid all the logic board connections, unscrewed lots of screws, and lifted out the main logic board. Once the logic board was out and flipped over, this is what it all looked like:
Yes, that's what thousands of dollars of premium hardware looks like when you take it all apart. Personally, I'm amazed at just how much functionality is located in such a small package.
At this point in the process, with the logic board out, we could see for ourselves the stock from-the-factory thermal paste application. They sure didn't skimp on the goo. Here's what the thermal paste on the heat pipe looked like:
There was even more goo on the CPU, GPU, and main logic chips, which are on the bottom side of the logic board. Here is what they looked like:
If the theories about too much thermal paste were correct, then my laptop definitely had a problem transferring heat. Of course, the only way to find out for ourselves was to press on. So, we cleaned the chips down to their mirror-finished packaging to prepare them for a new, and supposedly better, application of paste.
The heat pipes also got the fine treatment. They were cleaned up so that their polished copper pads, where the logic board chips make contact, were impeccable. All it took was a bit of elbow grease and industrial-grade isopropyl alcohol, and then a finish with a chamois cloth to make sure that there was no residue.
Next up was applying a new coat of thermal paste. For this experiment, I bought some of the best stuff out there: Arctic Silver 5. It seems to be the favorite of the over-clocker and system modification community, and the Arctic Silver website has great instructions on how to apply it. Carefully following these instructions, as well as the recommendations of others who had done this job, I applied the Arctic Silver.
Comparing my work to the Arctic Silver site and to other people's pictures, I was pretty sure I'd done as well as I could. So, I put it all back together to find out the result.
To test out the result of replacing the Thermal Paste, Greg and I ran both of our laptops at max CPU for a while. It's actually quite easy to do. You just need to fire up two terminal windows and execute the following in each:
$ yes > /dev/null
There are probably better ways to load up the CPU, but this did the trick for both of our machines. More importantly, we were treating both machines in the same way so that we could make an apple-to-apple comparison, so to speak.
After heating up the systems, we used a fancy Fluke IR thermometer to take readings from all over both systems. All of the readings told us the same thing: my modified laptop was running a degree or two cooler than Greg's unmodified version. Here are some pretty typical readings taken from exactly the same location on the bottom of each machine:
Greg's MacBook Pro with the original unmodified thermal paste is on the left with a reading of 116 degrees Fahrenheit. My modified one is on the right with a reading of 114. Here's another reading from the top, centered on the keyboard:
On the left is Greg's laptop, reading 97F. On the right is mine, checking in at 95F. We took many more readings; some places were hotter than the pictures above, up to 121F. But the message was pretty much the same: all that work seems to have given my laptop a slight advantage--two degrees of advantage, to be exact. With that small of a result, you can actually argue that there are any number of other causes. For example, one flaw in our methodology is that we didn't compare the laptop temperatures beforehand to see if there was a similar two-degree delta.
No matter how you argue it, however, this wasn't the finding I hoped for. It didn't match the glowing reports on the web. Maybe other people's thermal paste skills are better than mine. Or maybe there's another reason. You see, there's a bit more to this story that I haven't told yet. It's time to admit a mistake I made.
What I left out of the description above is that I actually disassembled and reassembled my MacBook Pro twice. Yes, twice. When I fired up my laptop the first time, I was greeted by the fans whooshing at full speed. Now, it was nothing like hearing my G5 running full out, but it made a fair amount of noise, and you could really feel the air moving through the case.
How loud was it? It was loud enough to be too loud while hacking code in bed. It was almost too loud when I sat on the couch on a fairly quiet day. But it wasn't too loud for other situations. For example, I couldn't hear it at all when sitting at the local cafe enjoying a double-shot Americano. I can tell you this because I actually waited a day to disassemble the MacBook Pro again to see what was wrong. I figured there was a slight chance the temperature control hardware had to reset itself, or that the Arctic Silver was conducting heat so much better into the pipes that the calibration for running the fans was off.
These weren't likely possibilities, mind you, but I'd taken the experiment this far; I wanted to explore all the possibilities. Not to mention, I really needed to do some other things for a while. And, my MacBook Pro was cool; actually, it was cold to the touch. I compiled a fresh install of Subversion, which is known to strain the system, and left the laptop on my lap while wearing shorts. I felt only the barest amount of warmth from the laptop. The keyboard area was actually cool to the touch. It was the most amazing thing.
For a day I left my system like this. Except for the noise, it was enjoyable. My MacBook Pro was once again a true laptop.
But after a day of listening to whooshing, as well as resetting the PRAM and performing other voodoo tricks to see if the PMU and SMC units would notice the more advanced thermal paste in action, I realized it was time to open things up again and find out what had happened.
The answer was simple: the connection from the motherboard to the heat pipe sensor didn't make it through the reassembly. This meant that, at boot time, the system couldn't take a reading of the heat pipe's temperature. This caused the system to enter into an "Oh crap, we don't know what's going on so crank up the fans to save the ship" mode. It's the same kind of thing that happens when you take the inside door off of a G5. The system's temperature management system responds to a failure in its setup with full-throttle fans.
Once I corrected the problem and carefully reassembled my MacBook Pro, the noise went away. And with it went the enjoyable coolness, which was replaced by a familiar warmth. It was this second reassembly that was used to take the temperature readings above.
So, maybe the reports of cooler laptops from people who replaced their thermal paste are really reports of laptops whose temperature control units are simply responding to a disconnected temperature sensor. Or maybe they really did do a better job than I did. I can neither prove nor disprove any of those statements. I can only report on my findings. However, since several internet reports of cooler laptops are accompanied by reports of louder fan noise, it's a strong possibility in my book.
If heat is more of an issue than noise, you could open up your MacBook Pro and disconnect the thermal sensor. This will recreate the issue I had after my first reassembly and cause your laptop's fans to kick into high gear all of the time. It's actually easy to do. You don't have to remove the logic board or a lot of connections. You just need to pop the top of your case, remove the keyboard cable, and then disconnect the sensor where it connects to the top of your logic board. It's circled in the following picture:
Once disconnected, cover the end with Kapton tape so that it can't short anything else out out, and then tape it down out of the way.
Of course, I can't guarantee your results. You may blow your fans up, though probably not. And you'll notice that I haven't told you how to crack your machine. There are lots of guides you can find online with a quick Google search, but I'm leaving that to you as a further admonition to open your MacBook Pro at your own risk.
With all of that work, my MacBook Pro was transformed from a slightly too hot yet amazing workhorse of a machine to, unfortunately, an amazing workhorse of a machine that runs a bit too hot. How much is a bit too hot? Well, from a little experimentation, the bottom of the case gets about 10-15 degrees Fahrenheit too hot for comfortable use on a lap. I arrived at these numbers by sleeping my MacBook, letting it cool, then measuring its temperature every few minutes as it warmed back up. The threshold between "just warm" and "too hot" is amazingly small, really. A change of five degrees in the outside case temperature is all it takes.
As far as the thermal paste issue is concerned, my opinion at this point is that although the factory's gooey application may look horrible, it's probably a perfectly acceptable practice given the factory and service center's desire to ensure proper contact between the chips and heat pipe with a minimum of fuss. After all, I can tell you that it took much longer to carefully apply the Arctic Silver as instructed than it would have to just glob on the stuff. And if the end result is the same, why not?
I have another concern, besides comfort, when it comes to heat coming off of the MacBook Pro. That concern is battery life. Lithium-ion batteries hate heat. Apple even has a web page dedicated to discussing how heat affects laptop batteries. It advises keeping your Mac as close to room temperature as possible, and certainly below 95 degrees Fahrenheit. Somehow, I don't think that a laptop that regularly runs with an outside case temperature of 115-120F qualifies as a good environment for lithium-ion batteries. I'll have to do some more testing with my fancy thermometer to determine how much the batteries heat up.
Now, there is hope for the future based on my experience with maxing out my fans. It's obvious that the cooling hardware in the MacBook Pro is first-class. When the system running full out, I can load the CPU and GPU to the hilt and the machine stays cool, almost cold, to the touch. The problem with heat in the current MacBook Pros seems to be a software one; the system firmware designers are apparently optimizing for quiet over heat. A smallish change to the balance between noise and heat will probably cool the laptop quite a bit without needing to run the fans all out, and it's something that I would welcome. Unfortunately, this is something that probably only Apple can fix with another SMC firmware update.
As a final thought, if you are considering breaking open your shiny MacBook Pro and modifying its thermal paste, you probably shouldn't. You run a decent chance of turning your machine into a brick and even if you do the job perfectly, you're not going to get a big change. I gave myself 10-20 percent odds of hurting my machine. As it turns out, I did just fine and my machine survived with no harm done, but it might not have happened that way. You might not be so lucky.
James Duncan Davidson is a freelance author, software developer, and consultant focusing on Mac OS X, Java, XML, and open source technologies. He currently resides in San Francisco, California.
Return to the Mac DevCenter
Copyright © 2009 O'Reilly Media, Inc.