I recently came across a problems with a MacBooks fan which caused it to stop working. I’ve just ordered a replacement, but in the meantime I’m using the MacBook passively cooled, with surprising results. Here’s what I’ve found so far, hopefully it would be useful if you’re thinking about tinkering with your cooling system or if something goes wrong.
Update: Just to let you know that this post just summarises my findings. I’d like to point that tinkering has its risks, so be prepared to take responsibility if yours doesn’t go the same way.
Passively cooled MacBook Pro
I’ve detailed how the MacBook Pro 13 cools itself in this blog post, so if you’re interested you can read on…
Right, back to the subject of this post. A decade or so ago, taking the fan off a system was suicide. AMD CPUs would literally burn out, taking the motherboard with them, while many Intel CPUs would quickly throttle to the point of becoming unusable. Thankfully, times have changed and most CPUs (especially mobile CPUs) almost never burn out and have safety mechanisms to ensure that they don’t overheat and damage the system. Lower idle power draw also means greater thermal headroom to use the system at full performance for longer.
For this particular system, the Core 2 Duo P8600 has a maximum operation temperature of 105°C, and under normal conditions with the fan, CPU temperature barely reaches ~90°C. I’m not sure what the Nvidia 320M GPU and northbridge are rated for, but they should have a similar maximum operating temperature.
At idle, both the CPU and GPU combined have a power draw of ~2.06W (about the same as a mobile SoC under full load) so they should have no problems with temperature. Under light workload (text editing with music etc), both CPU and GPU draw ~3.5W (similar to a Nexus 10’s Exynos 5 SoC under full load) which again should be manageable. Under full load however, the combined power draw of both CPU and GPU is ~27W, possibly more, and this requires active cooling to sustain this level of performance.
With Fan in Blue and Without Fan in Green. Temperature in °C
Under full load with the fan, CPU and GPU temperatures are never high enough to cause throttling, and the system remains responsive. Fans are spinning at 5300RPM, so there is still some thermal headroom (maximum RPM is 6000RPM) to allow for higher ambient temperatures. Battery temperatures are reasonable and palm rest warms up a little, but nothing to write home about.
Without the fan, the results are quite dramatic. CPU temperature hits 96°C (very quickly) and it is around this temperature where the CPU throttles resulting in a noticeable performance decrease. Once the CPU throttles, temperatures stabilises and doesn’t increase any further. The GPU temperatures (which are generally higher than CPU due to the nature of the heat pipe on the 2010 MacBook Pro) are slightly more worrying, with a peak at 103°C, considerably higher than with the fan (90°C). Although temperatures stabilise, the system remains almost completely unresponsive due to throttling.
Early Conclusion: without a fan, don’t stress the both the CPU and CPU, as throttling would make the system unresponsive. Tasks which are quite taxing but don’t fully utilise both CPU and GPU (Image editing, light gaming etc) can be done quite comfortably, albeit with a slight decrease in performance when throttling kicks in.
With Fan in Blue and Without Fan in Green. Temperature in °C
Idle temperatures show a similar increase in temperature as load. Interestingly, idle temperatures without pre-load (when the system has resumed from standby, then left to idle) with and without the fan are the same (not shown in graph). Placing load on the system and allowing it to idle results in differences in temperatures between the fan and fanless variables. Without a fan, residual heat “hangs around” and doesn’t seem to dissipate very quickly once the system is idle.
Clearly, running the system without the fan results in higher operating temperatures resulting in a shorter lifespan of the components. The CPU and GPU both throttle to prevent immediate damage to the system which allows the MacBook to be safely used passively cooled. The component which concerns me is the battery.
Li-Ion (and Li-Poly) batteries are especially sensitive to temperature, and higher operating temperatures results in a shorter life span. An increase in 15°C (as shown in the idle temperature graph) can accelerate capacity loss by 15% over a year. The battery tends to heat up to ~35°C during load regardless of whether a fan is present or not, but it takes much longer for the battery to “cool off” when idling without a fan. I would advise that if you are passively cooling the system, don’t tax the components too much while charging.
The heatsink can be thermally bound to the bottom plate of the MacBook by sandwiching a strip of metal between the two surfaces. I used thermal paste on the heatsink side to aid the binding of the surfaces, however the bottom plate has a handy plastic strip over the heatsink which seems to mould itself around the metal strip, creating a nice thermal contact. It’s not a perfect thermal solution, but it does the trick.
Thermal paste used to bind the metal strip to the heatsink. Thickness ~4mm. The fan is in there solely to help secure the metal strip.
Under load, the bottom plate now becomes much warmer, and the heat is much more noticeable around the heatsink area. Load temperatures haven’t fallen much (or noticeably so anyway), but general system performance has increased slightly…
…So I decided to measure how long it would take until the MacBook to throttle the CPU under heavy workload. I ran a few OCR tests at ambient temperature. OCR places a relatively consistent load on the system, and is limited to CPU performance so should be a reliable test. Between tests, the system was turned off to reach ambient temperature (30 mins).
Time in Seconds
With the fan, the CPU never throttles during the OCR test and maintains peak performance throughout the test. Without the fan and it took ~70 seconds before the CPU became too hot (peak = 103°C) and started to throttle. With the mod however, peak performance was maintained for ~100 seconds before any throttling occurred. This probably indicates that the mod was a success, albeit a small one. Better thermal binding of the heatsink to the bottom plate may have yield better results.
Passive Cooling, is it worth it?
From my short time testing the fanless MacBook Pro, I’ve been pleasantly surprised at the results. Problems didn’t arise as often as I expected, and as long as workloads are kept low (daily tasks such as e-mail, web browsing, videos etc), performance is virtually indistinguishable from an actively cooled system. However, demand more of the system (complex image editing, gaming, batch processing files, OCR(ing) etc) and throttling kicks in to slow everything down.
I find that removing the fan reduces the MacBooks capability to that of an iPad or tablet. If you are going this route for whatever reason, as long as you keep your usage expectations similar to that of an iPad, then it’ll work for you. Of course, occasionally, software updates or other important tasks require full utilisation of the hardware, and it is at these times when having a functional fan might be a good idea although worst comes to worst, tasks would just finish slower.
And finally, theres the case for noise. Replacing a hard drive with an SSD goes a long way to make the system quiet (most idle noise actually comes from the hard drive, not the fan). Removing the fan has a smaller effect, because at 2000RPM, the fans are almost silent anyway. Removing the fans presents a tradeoff, lower sustained performance in exchange for silent operation.