The Raspberry Pi Product That Shouldn’t Exist

When the Raspberry Pi 4 originally launched I wrote about how when enclosed in the official case sitting idle on my desk, the processor temperature rose to 77°C (@~22°C Ambient), and under even light load would quick hit thermal throttling. The improvements made have helped, with the Pi now idling at 56°C (@~21°C Ambient). Under sustained load, the Pi will still thermally throttle.

As you’d expect the 3rd party case market is awash with alternatives to the official Raspberry Pi 4 case, many of which can provide better cooling or better access to the GPIO and other headers on the Pi. The lack of access to the board headers without completely leaving the top half of the case off is a design decision that puzzles me. I’m also not a “fan” of active cooling solutions, having tested many they are for the most part noisy and over time the bearings typically make the fan noisier. All the Pi’s I have in constant daily use are using passive cooling, with the exception of the Pi with the PoE HAT which has a fan built-in.

On 30th November The Raspberry Pi Foundation launched a new product, the Raspberry Pi 4 Case Fan. The launch article by founder Eben Upton covers the heat issues with the Raspberry Pi 4, and how various updates since the launch of the Raspberry Pi 4 have helped to reduce power consumption and therefore heat generation. The article goes on to explain how the Pi was designed to work in a “sprint-and-recover” mode. Not for a sustained level of high load. The new Raspberry Pi 4 Case Fan aims to address the concerns around thermal throttling when using the Offical case for sustained loads or when overclocking the Pi.

Clever Engineering, But Why?

I have to admit it’s a clever bit of engineering to design something to fit within the constraints of an existing product design. The case fan clips in to the two half of the official clamshell case, blowing air across the processor heatsink. However air intake and exhaust is very limited, relying on small gaps around the various ports. The 25mm fan is a bit smaller than the 30mm versions found in most other case. I suspect the decision to mount the fan on the sloped baffle in this way restricted the sizes of fan that could be used.

As I embarked on my usual round of tests, comparing it against various alternatives, I was struck by the realisation that this product should never been released.

The Raspberry Pi 4 Case Fan was essentially created because the case it’s to be housed within is already seriously flawed. Adding a fan might help mitigate the heat issue, but introduces noise in its place and a very high pitched and annoying noise at that. In Jeff Geerling’s review of the case fan, he includes a recording of the sound. If interested I suggest you watch his review video, or you can jump straight to hear the sound of the fan.

If anyone has concerns about the heat, don’t compound the poor decision of buying the official case, by purchasing the official fan. Either take a drill to the case and make a large opening for a large fan or buy a different case, ideally one designed for the active or passive cooling you prefer.

Sunken Cost Fallacy

I think this Case Fan product could be a great example of the sunken cost effect, where the Raspberry Pi team have proudly invested in the design (with Kinneir Dufort) and manufacturing (with T-Zero) of the official case, continue to invest in the flawed product rather than admitting they should have done better in creating a product fit for purpose from the get-go. In fact, the design of this new addon is presented in another RaspberryPi.org blog post, “Designing the Raspberry Pi Case Fan” – 2 Dec 2020, which is an interesting read but sadly Gordon Hollingworth Chief Product Officer, doesn’t explain why they didn’t take a fresh approach and instead worked within the constraints of the poorly design (for thermals) case.

I’d have been a lot happier to see the Raspberry Pi Team invest their time and expertise in designing a new case which addressed the failings in the original design, specifically not designing for the thermal characteristics of a Raspberry Pi 4 in use in a range of ambient operating temperatures.

Alternatives

At £5 for the Official Raspberry Pi Case and £4.50 (+Shipping), it’s one of the cheapest ways of enclosing your Raspberry Pi 4. For £10-12, you can get a range of cases which maybe aren’t as pretty to look at, but are more functional in terms of cooling and access to GPIO. £15-16 gets you either Argon Neo (which is my favourite small Pi case) or the FLIRC case, both beautiful, elegant aluminium enclosures. If you really are serious about overclocking then larger cases like the Argon One should be considered, or if you don’t mind the board being partially exposed, then ICE-Tower is another option.

Testing

I’m somewhat reluctant to spend the time documenting the test results as I don’t think they really mater. But, the testing was already mostly done before I came to my conclusions I’ve expressed above. So might as well share for the sake of completeness.

Testing Method & Configuration

Three Pi’s were tested concurrently, each running the same image only reconfigured with unique host names.

• Image: 2020-12-02-raspios-buster-armhf.zip
• Updates: All updates as of 29 December 2020
• Kernel: Linux raspberrypi 5.4.79-v7l+ #1373 SMP
• Firmware:
  • Bootloader: Thu 3 Sep 12:11:43 UTC 2020 (1599135103)
  • VL805: 000138a1
• Software/Configuration:
  • WiFi configured and enabled
  • Boot to console
  • SSH enabled
  • Samba installed and configured
  • Stress
  • Stressberry
  • Python 3.7
• Load Test configuration:
  • Stress active concurrently on all 4 cores
  • 5 minute idle period (once the temperature has reached steady state)
  • 30 minute stress test run
• Overclock settings for OC test scenarios:
  • over_voltage=7
  • arm_freq=2147
  • gpu_freq=600

Test Devices

• Raspberry Pi 4 official case
  • Raspberry Pi 4 4GB version
  • Raspberry Pi 4 official case fan (and supplied heatsink) run at 5V only
• 219 Design 3D printed case
  • Raspberry Pi 4 1GB version
  • 30mm Pi-Fan (GF3007BS) run at 3.3V and 5V
  • Heatsinks on most board components, per review images.
• Argon Neo aluminium case
  • Raspberry Pi 4 4GB version
  • Pimoroni Fan Shim used in scenarios when testing with a fan

Test Results

Fanless

The official Raspberry Pi 4 case and the 219 Design case without any active cooling start to thermally thottle around the 1000 second mark. The Argon Neo with a significantly larger thermal mass is able to keep the system well below the thermal throttling limit. Based on previous 1 hour long stress tests with the Argon Neo, the temperature plateaus after the 2000 second mark.

Active Cooling

The impact of the 80°C trigger threshold for the Official case can clearly be seen, cooling by 10°C before turning off again. If the fan is able to run constantly, then the Pi is able to be kept in the low 60s. As you’d expect none of the scenarios here trigger thermal throttling. However the benefit of a case with more airflow (and a slightly larger fan) can be clearly seen with the 219 Design results. With the Fan running at 3.3V instead of 5V the 219 Design still out performs the Pi case and is very quiet, unlike the Official Pi Case fan.

Overclocking

The active cooling tests show all configurations are able to keep the Pi from thermal throttling at stock clock speeds (1.5GHz). To create a more challenging test, I overclocked the Pi to 2.147GHz as I’d previously done when testing out the ICE-Tower cooler. I did have to increase the over-voltage from 6 to 7 in order to ensure the systems were stable.

In this configuration, only 2 configurations didn’t thermally throttle, the 219 Design with the fan running at 5V and the Argon Neo with the Pimorni Fan Shim installed and the top cover removed (The Neo still covers the mainboard under a different cover).

Conclusions

Firstly I’ll restate what I’ve been saying since the release of the Raspberry Pi 4, don’t buy or use the Official case, and don’t compound that error by spending money on the official Case Fan. As you can see there are better cases for active cooling for stock or overclocked scenarios. The official Pi Case Fan nosier than others and a passive case can out perform the official case without the noise.

Product Links

• Raspberry Pi 4 Official case (Amazon) (ThePiHut)
• Raspberry Pi 4 Official case fan (ThePiHut)
• Argon NEO case (Amazon US, UK) (ThePiHut)
• Argon ONE case (Amazon) (ThePiHut)
• FLIRC case (Amazon) (ThePiHut)
• ICE-Tower Cooler (Amazon US, UK) (ThePiHut)
• Pimoroni Fan Shim (Amazon) (Pimoroni)

Amazon links are affiliate links which help support the site, where possible I’ve used links which should take you to the product in the Amazon store in your region. Links to other suppliers are included for your convenience.