The extra pin of each fan is a tachometer output. You can leave those disconnected or connect them to the unused GPIO pins for RPM monitoring (check out this tool). For my build I just cut off the fan connectors, soldered on extension wires and crimped female Dupont connectors onto the ends. I also reused the sleeve that came from the factory and secured it with heat-shrink to make the harness look tidier.
I appreciate your answer. I was able to make it work perfectly. I have a few more questions:
- Use this script for OpenWrt which reads all temperature sensors in hwmon and adjusts the fan speed according to the highest reading; you can dial in your own temp thresholds there
- Put the script into /opt (or any persistent directory) with the following command (if using Linux or macOS): scp -O /ā¦/fan_pwm.sh [email protected]:/opt/
- Make it executable: chmod +x fan_pwm.sh
- Make it run at startup: sed -i ā/^exit/i /opt/fan_pwm.sh &\nā /etc/rc.local
For a common mortal that is approaching this from a 0-knowledge perspective, how do I do this exactly? I was capable of making the script run from ssh, but I donāt know how to save that and make it run on startup.
- Rather than making the grounding strips by hand, order them to be laser-cut insteadāitās likely to cost less than your time spent measuring, drilling and cutting
Are there any CONS in not using the grounding strip? I installed everything without it, and itās working decent.
Oh, BTW, just for laughing a bit, instead of soldering the EMI shielding, I used superglue 
Regardless, before I was somewhere at 78/79 noise, now I sit at around 85. Itās still progress!
For a common mortal that is approaching this from a 0-knowledge perspective, how do I do this exactly? I was capable of making the script run from ssh, but I donāt know how to save that and make it run on startup.
If youāre creating the script on the system rather than transferring over:
apk add nano or ipkg install nano depending on your OpenWrt release, if not already installednano /opt/fan_pwm.shchmod +x /opt/fan_pwm.sh to make the script executablesed -i ā/^exit/i /opt/fan_pwm.sh &\nā /etc/rc.local to make it run on startupcat /etc/rc.local; you should see the line /opt/fan_pwm.sh &/opt/fan_pwm.sh & to launch the script without rebooting; you should hear the fans slow downluci-app-statistics and lm-sensors packages to track the temperatureAre there any CONS in not using the grounding strip? I installed everything without it, and itās working decent.
In my experience, the improvement from the grounding strips and EMI shielding together was marginal, but that could change depending on how congested your Wi-Fi environment is. I also hope the BE19 will be inherently less susceptible to noise to benefit from extra grounding.
Quick question how much difference did it make for noise, and how do you cool the WiFi card now?
Looking at my previous experiments, when everything else has been sorted out (SFP cages populated, pigtails routed properly), not using the stock metal case can lower the noise by about ~9 dBm.
I have a copper heatsink mounted on top of custom-made EMI shielding. Thermal pads under the shield can, thermal adhesive tape between the heatsink and the shield can.
Hmm, I was looking for a less custom and cheaper solution lol
Slapping three smaller heatsinks on the ICs with thermal tape will do just as well. One large heatsink wonāt work, though, as the ICs differ in height.
Ah ok, cool thanks for that info
Iāll probably adapt the top plate to support 1 90mm fan instead since I already have a 90mm fan installed
@Betonmischer Man, youāre out-of-mind! (in a very very good way!) 
If I can find all the pieces, Iāll give it a try with ABS (any clue for support material?).
Fans should be pushing or pulling air?
Out of my curiosity, what CAD did you use to design this beauty?
Thank you!
A guy on MakerWorld had success with ABS, but he didnāt post any photos and Iād be worried about ABS warping with such large flat parts. If youāre willing to give it a try, though, Iād pick the ABS-compatible support material from the same manufacturer as the ABS. A way better option dimentional stability-wise would be GF or CF-reinforced ABS/ASA, but those might require a larger than 0.4mm nozzle. BTW, at first I was apprehensive of PLA due to the low glass transition temperature of 55ā65 Ā°C, but the R4 should never get the air inside the case anywhere close to that.
As for the fans, they fit both ways, but Iād install them for intake so that the cold air hits the Wi-Fi board and the SFP cages first.
I used Fusion 360. Or rather, used this project to learn Fusion, which lead to a file with a pretty messy history, otherwise Iād share it and made the project fully open-source. 
Changes in v1.2.0
Would anyone want an add-on for mounting the case in a 10-inch rack?
ā¦would you be interested in adapting that model for the BPI R4 Pro out of the specs?
Yes. It would help if Sinovoip published DXF drawings of the board, which they havenāt yet. So Iām waiting until the Pro is available to buy, which Iād have to do regardless to verify the dimensions.
I oredered the board in the ongoing presale.
When Iāll get it, if thereās nothing better, Iāll be happy to send you measures.
The print will have to be tested to fit the real thing, so Iād suggest not rushing it. Iām personally holding off until the first reviews are out and no critical bugs are found in the board.
DXF files are published here Banana Pi BPI-R4 Pro Wifi7 Router board design with Mediatek MT7988A | BananaPi Docs
Excellent, they even have a 3D-model of the board this time.