`RTL8812BU` Launch Report

[MuratovAS]

Initial conditions:

• transceiver: tp-link archer t4u plus (RTL8812BU)
• wfb-ng: v24.08.2
• linux: kali-2024.3-qemu-amd64
• driver: see below
• tools: iPerf3

Result: After a series of trials and errors I managed to achieve operability on the RTL8812BU chip. The system works in two ranges, 2.4Ghz and 5Ghz.

I checked the functionality with the rtw88_8822bu driver built into the kernel. With default settings the system refused to work, data is not sent or received. After I changed the line to frame_type = 'rts', the connection worked, but the transfer speed was unacceptably low, about 50Kb/s.

I was able to achieve stable operation using a custom kernel module from RinCat. Using this code, the connection speed through the tunnel was about 8Mb/s. On standard wfb-ng settings.

Based on the data I received, the chip can be used. As soon as I have more data, I will let you know here.

[svpcom]

@MuratovAS Thanks!

P.S. I've seen your portfolio - could you advice dc-dc for powering such wifi cards? Most of powerful 8812au and 8812eu cards consume > 3A in pulse when transmitting and generic dc-dc converters from aliexpress requires additional low-esr capacitors to survive these pulses. Are any better (an cheap) solution to power such cards from 3S-6S batteries?

[MuratovAS]

I would advise:

• Take large cross-section wires, multi-wire (многопроволочный) (reduce AC/DC resistance)
• Power wires should be well twisted together (reduce inductance, increase peak pulse current)
• Make this wire as short as possible, a couple of centimeters.

I would build a converter on RT6365, simple, reliable, cheap, powerful. If you need a ready-made solution URB2405LD-20WR3 or URB2405LD-30WR3

[svpcom]

@MuratovAS Thanks! I had very bad experience with this dc-dc:

It dies with smoke when connected to 24V and connection is unreliable: https://t.me/wfb_ng/4457/11336

[svpcom]

@MuratovAS Do you have any idea why glitching of input voltage can lead to dc-dc burnout? It is IC design flaw or bad schematics? All three dc-dc in my test dies identically: case near +Vin pin was burn out

[voanhcuoc]

Thanks to this report, I've conducted an experiment:

• transceiver: ASUS USB-AC53 Nano
• wfb-ng: v24.08 (official pi 3 image)
• driver: RinCat
• linux: Raspberry Pi 3
• camera: Chinese no-name webcam (640x480)
• stbc = 2
• tx power : default

I don't have a DSI monitor for Pi, so I connected an Android phone with the Pi by a USB cable. Then, I used adb reverse to tunnel RTSP connection over USB in order to show the stream in Android VLC.

The test is conducted on the outdoor ground with no obstacle.

Result: The video is fairly good at close range. It start to pixelate at 15m and then lose connection at 25m.
Pretty bad result, I guess due to limit of small onboard antennas, AC53 Nano is a smallest-possible USB wifi, anyway.

[MuratovAS]

@MuratovAS Do you have any idea why glitching of input voltage can lead to dc-dc burnout? It is IC design flaw or bad schematics? All three dc-dc in my test dies identically:

I am seeing the power mosfet burn out. IC: tps5430 Possible causes:

• Transistor not fully opened (transistor control system did not have time to enter operating mode)
• Load starting current is excessively high (output capacitance is too high + load is too powerful).
• Internal oscillatory processes have occurred in the system.

Solution:

• Do not load DCDC at full power at the moment of starting (in reality, this is a very rare scenario)!
• Choose an IC for DCDC with a soft-start function

P.S. I have never used this power supply at a current greater than 2A. The recommended current is 1.5A

[svpcom]

@MuratovAS The main problem is that most of wfb-ng users are hobby users and unable to make custom PCB design themself. Most of ready to buy dc-dc from aliexpress has awful design and overstated parameters. It would be great if someone who specializes in this topic would make such an opensource design, which could be easily repeated by yourself or ordered on pcbway