(WIP) RTL8812CU/RTL8822CU Linux Driver v5.15.8-52-20230728 FPV Mod
Download pre-built OpenIPC firmware with this driver integrated here to help test.
Original driver from Realtek: RTL8822CU_WiFi_linux_v5.15.8-52-g35e62e7c9_COEX20221215-3130.20230728.tar.gz
BL-M8812CU3 datasheet: BL-M8812CU3_datasheet_V1.0.1.1.pdf -- A good replacement of BL-M8812AF1
The RTL8812CU has fully integrated RF frontends (PA, LNA, and SPDT switch), so you will rarely see any RTL8812CU adaptor with PA/LNA (adding the PA means an additional SPDT switch).
The internal PAs have really good performance (HT20, MCS7, ~16dBm TX, -28dB EVM), but it's still limited by the 3.3V power and poor heat dissipation.
LB-LINK's support said the chip's output (using only internal PA) can reach 23 ~ 24dBm while injecting at 6Mbps, based on their measurement. So, if you need something between the big square RTL8812EU and the smaller square RTL8731BU...
Tested:
- Build on kernel 6.8
- TX power unlocked (same as the 8812eu driver,
iw wlan0 set txpower fixed <0~3150>) - RX in monitor mode
- TX injection in monitor mode, HT/VHT rates, and STBC/LDPC all work well
- Narrowband support in monitor mode, 5MHz/10MHz bandwidth
Need test:
- EDCCA patch
- ~4K Maximum MTU
- DKMS build script
- ACK timeout (in procfs)
- 10MHz BW in AP/STA mode (seems that the firmware needs to be replaced: hal8822c_fw_10M.c)
- Thermal sensor default offset (the sensor readout code is working, but it needs a single-point calibration, so a default value needs to be set depending on some actual measurement)
- Single tone output (/proc, single_tone)
- Disable CCA (/proc, dis_cca)
Known bugs/issues:
- Short GI not working (not the radiotap header issue -- even when TXDESC is set by SET_TX_DESC_DATA_SHORT_8822C()). Maybe it's a firmware bug. But you should always use long GI on your drone so it doesn't matter
Set txpower byiwhigher than a certain (depends on the module, but significantly higher than the factory) value will cause tx to hang, and cannot be recovered unless manually reset the adaptor. Not sure if it's a power issue caused by poor PCB trace width inside the module (it's reasonable to not use an inner power plane for a ~300mA total chip...), some firmware bug, or even some analog silicon bug (latch-up?). Not caused by high temperature (water cooling tested, still), or any high TX traffic. Weird...- LB-LINK said the issue above can be a power failure, and a proper 3.3V power supply is needed. I was using 3.3V/1A, so I recommend at least 3.3V/2A