Highly effective Raspberry Pi cooling without the need for a fan.
If you have had any experience with the Raspberry Pi 4 and especially closer to its launch, you would have likely become aware that this iteration of the Raspberry Pi gets quite toasty under a moderate load. This initially resulted in some pretty crazy active cooling designs on the market for the Raspberry Pi 4.
Whilst many of the heat issues plaguing the Raspberry Pi 4 have been alleviated with firmware updates to dramatically reduce the power dissipation at idle and medium loads, the Raspberry Pi 4 still gets blisteringly hot when pushed with higher loads. This causes the ARM quad-core chipset to protect itself by reducing performance using dynamic frequency scaling (DFS) and dynamic voltage scaling (DVS). DFS and DVS are the methods used to reduce the performance of a chipset in order to reduce the junction temperature, and are often simply referred to as thermal throttling. On the Raspberry Pi 4, this thermal throttling begins at around 80°C.
With this in mind, when we heard that there was a passive cooling solution for the Raspberry Pi which would keep the temperatures low enough to prevent thermal throttling at stock speeds, we were a little dubious. With that said, we put the GemPro case to the test to see if it did what it claimed.
First impressions
The case itself is made from forged or cast aluminium and is a surprisingly hefty 262.7 grams when loaded with the Raspberry Pi and SD card. This heft is a good sign and adds quite a bit of veracity to support the claims made in relation to this case's cooling abilities. Simply put, this style of passive cooling case uses the entire enclosure itself as a heatsink, and thus, the more material and surface area the better the cooling potential.
The enclosure consists of two main pieces that are screwed together, and both have a sandblasted textured finish. The top of the enclosure has a pattern that appears to be milled into its surface that somewhat resembles a cobblestone path and certainly makes the design unique looking.
The holes to the micro HDMI, audio, and USB-C connectors are precision cut into the side of the enclosure, and there is no recess around the micro HDMI inputs to accommodate the overmolding on some micro HDMI connectors. Whilst this will likely work fine with many types of micro HDMI cables, ours struggled to make a reliable connection. Bumping the cable would result in losing the image on the screen. This could also be an issue with the cable rather than the enclosure, but it is certainly worth mentioning.
Overall, the quality looks and feels amazing and would look great on any desk.
There are, however, a couple of things missing that we would have liked to have seen included. The first is some non-slip rubber feet. The screws in the bottom of the case are countersunk so they fit flush with the case and won’t scratch your desk’s surface, however, we think rubber feet to hold it off your desk’s surface would be a great addition allowing airflow under the case as well as over.
It would also be nice to have some pre-tapped mounting holes in the rear or bottom of the case so you could mount the enclosure. This would make it much easier to create a VESA bracket, for example, which would allow you to mount the Raspberry Pi to the back of your monitor or Television.
The most egregious issue we noticed was that the GPIO header was broken out on the lower side of the case, requiring the connected cables to be manipulated at odd angles to exit the enclosure. This seems to have been a bit of an afterthought.
Finally, while assembly is fairly straight forward, instructions are included to get you going without too much fuss.
The enclosure requires a little assembly before use. It has two machined aluminium blocks designed to draw heat from various areas of the Raspberry Pi and into the aluminium of the enclosure itself. The blocks are attached to the case via a thermal paste.
This is a significant point of difference between the GemPro case and some of the more extreme active cooling methods we have seen for the Raspberry Pi 4. Some of the popular active cooling designs are only designed to cool the processor itself. However, if we look at the Raspberry Pi thermals, we can see there are several areas which get hot under load. The large white square is the hottest area reaching 75°C. This is not surprisingly the Broadcom BCM2711 processor.
We can also see there is a hot spot beside the USB-C input connector. This section of the PCB is the MaxLinear MxL7704 power management integrated circuit (PMIC) and the accompanying inductor. This IC is responsible for delivering the regulated supply to the processor and associated hardware. As we can see from the thermal image, under normal operation, these can easily reach temperatures around 70°C. This would increase with overclocking and if you are drawing a significant current from the GPIO, as this regulator would, then be under even more stress.
The GemPro case has an L-shaped block which connects to this area as well as the main processor.
The GemPro also comes with a smaller square block which connects to the VI Labs VL805 USB controller. Interestingly, this was not shown to get too hot in the thermal image, however, if you’re running several USB devices it's very likely that the chip will become incredibly hot. Thus, very much a nice to have addition.
All in all, the case appears to cover all of the requirements to keep the Raspberry Pi running smoothly without thermal throttling. With that said, the proof is in the results and we were itching to put the case through its paces.
Testing
To thoroughly test the enclosure and reach meaningful conclusions based on the results, we needed a way to test the enclosure’s thermal capabilities. For this, we created a fresh new Raspberry Pi OS SD card and installed the Stressberry program, available from: https://pypi.org/project/stressberry/
This program will allow us to simulate a load on the processor for a predetermined period and graph the temperature during the process. This will show us how hot the Raspberry Pi 4 gets above ambient temperature, and thus, can be used to gauge the effectiveness of a particular enclosure.
To minimise variables, we set the air-conditioning in the office to 25°C and ensured that the unit under test was not in the path of any draughts from opening doors or the air-conditioning unit.
The Stressberry test will load the processor so that it is maxed out for 30 minutes. It will measure and display the temperature at idle for 5 minutes prior and after the test to show the idle temperature and the recovery.
Our first step was to obtain a baseline result which we can use to gauge the effectiveness. For this, we tested the Raspberry Pi 4 without any cooling or enclosure and at stock speeds.
As we can see, the processor temperature started at around 47°C (22°C above ambient) at idle and rose a degree or two at idle. When the load was applied, the temperature rapidly rose to 80°C and the device thermally throttled the processor to prevent damage. Within 15 minutes of the load being applied, the Raspberry Pi had reached 80°C. Once the load was removed from the processor, the temperature dropped quickly down to 60°C. It then reached a temperature of 52°C after about 5 mins at idle.
With the baseline test done, we installed the same Raspberry Pi into the GemPro enclosure and repeated the exact same test. This produced the following results.
With the GemPro case, our idle temperature is already significantly lower at 30°C, which is only 5°C above ambient temperature. When the load is applied, the temperature shoots up to around 36°C. After 15 minutes at full load, the temperature has reached the maximum temperature of around 43°C or 18°C above ambient temperature. This is 5°C cooler than the Raspberry Pi 4 without any cooling at idle!!! Once the load is removed, the temperature of the processor rapidly drops from the max temperature of 43°C to about 37°C and holds that temperature until the test ends.
This was incredibly exciting, as we did not expect such significant cooling results from a passively cooled enclosure. The device was nowhere near reaching temperatures to cause thermal throttling, and was only 18°C above ambient temperature. This prompted us to try an extreme style of active cooling for the Raspberry Pi 4, which was an ICE tower cooler from Seeed studios.
With this monster bolted to the Raspberry Pi and under the same test conditions, we achieved the following results.
As we can see, at idle the ICE tower kept the Raspberry Pi 4 just a little bit cooler than the GemPro enclosure. At full load, it only achieved temperatures about 5°C less than the passively cooled GemPro.
This result was completely mind blowing and was not at all how we expected the experiment to go. This made us ponder what would happen if we extended the duration of the test. Would the GemPro enclosure continue to heat up? For this, we modified the test to run at full load for 1.5 hours. This test produced the following results.
We can see that the GemPro is more than capable of keeping the processor cool with a maximum temperature of around 45°C after being at full load for an hour and a half.
This was quite frankly just amazing. It prompted us to see how far we could push the case and hardware. We knew what we had to do... time to overclock the processor and see just how far we can push it!
Overclocking
IMPORTANT NOTE: Overclocking your Raspberry Pi will void your warranty and could potentially damage the device. We don’t recommend you do this unless you accept this risk.
We overclocked the Raspberry Pi 4 to 2GHz (500MHz increase from stock speeds) and repeated the previous 30-minute full load experiment. This produced the following results.
As expected, this had an obvious effect on the Raspberry Pi 4’s thermal performance. We can see at idle the temperature now sits at about 35°C, which is about 5°C warmer than at stock speeds, and 10°C above ambient temperature.
Once the load is applied, the temperature reaches a maximum temperature of around 58°C or 15°C higher than at stock speeds, and 33°C above ambient.
This is simply amazing and means that, in even the most hostile of situations, the GemPro case can keep your Raspberry Pi processor, power management and USB controller cool, even under heavy overclocking.
More amazingly, it does this silently and without needing to power other circuitry such as a fan. If you’re on the market for an enclosure to silently keep your Pi cool, especially if you’re wanting to overclock to squeeze out more performance, then this case is certainly worth a look.
WiFI Performance
A potential issue with metal enclosures for your Raspberry Pi is poor WiFi performance. The metal enclosure can attenuate the incoming WiFi signal, resulting in poor performance and dropouts.
We tested the WiFi connection on our Raspberry Pi 4 while in the Gem Pro enclosure using sites such as www.speedtest.net and by downloading large files. When comparing the results with the same tests performed without an enclosure, there was no significant or noticeable degradation in the download speed.
Conclusion
The GemPro enclosure is ideal for anyone who is looking for a passively cooled enclosure for their Raspberry Pi without compromising on cooling ability. If you want a passively cooled enclosure to keep your Raspberry Pi 4 cool (overclocked or not), then we are very confident that you will be impressed with this enclosure.
LIKES:
- Remarkable cooling capacity even when overclocking.
- Cools the power management circuitry.
- Cools the USB controller.
- Silent.
- No additional power usage.
AREAS FOR IMPROVEMENT:
- No rubber feet to prevent slippage on your desk.
- No mounting holes allowing you to mount the enclosure.
- HDMI ports not recessed.
- GPIO access is obfuscated.
GemPro Case available at SMALL FORM FACTOR
- GemPro Case for RPI4 by akasa® $46.50 AUD inc GST