Add MuziWorks Base Duo board support (nRF52840 + LR1121) 🤖🤖

[zaycev]

Summary

• Add board support for MuziWorks Base Duo (Elecrow nRFLR1121 module: nRF52840 + Semtech LR1121, sub-GHz LoRa)
• New CustomLR1121 / CustomLR1121Wrapper radio helpers (RadioLib's LR1121 class extends LR1120, not LR1110 — reusing CustomLR1110 fails at chip identification)
• 5 build environments: companion_radio_ble, companion_radio_usb, repeater, room_server, terminal_chat
• Board-specific config: 2-DIO RF switch table (DIO5+DIO6), TCXO 3.0V, 8MB QSPI flash (W25Q32JV), SoftDevice S140 v6.1.1, UF2 bootloader

Hardware tested

• Radio init (LR1121 sub-GHz, 22 dBm TX power)
• BLE pairing via MeshCore companion app
• LoRa TX/RX with good RSSI and SNR
• Battery ADC reading
• QSPI flash persistence across reboots
• TX LED indicator

Note

This PR was drafted by Claude Code, then manually reviewed, fixed, and tested on physical hardware.

Test plan

• All 5 build environments compile
• Flashed and verified on physical MuziWorks Base Duo board
• BLE companion app connects and communicates
• LoRa packets sent and received on sub-GHz band
• Settings persist across reboots (QSPI flash)

[lbibass]

I've already implemented this. #2054

Are there any major differences between my earlier PR and this one?

[zaycev]

I've already implemented this. #2054

Are there any major differences between my earlier PR and this one?

Hi @lbibass, I actually tried your PR first, but couldn't get it work.

I believe there are a few hardware mapping differences preventing it from working. My implementation just borrows the values from the MT firmware variant for the Duo board:

• LoRa Reset Pin: P1.10 (P1.12 overlaps with SPI NSS)
• RF Switch: Uses a 2-DIO table (DIO5, DIO6), not 4
• Flash: 8MB QSPI (not 16MB)
• Radio Class: Must use LR1121. Using LR1110 causes RadioLib to fail the hardware check on boot (0x03 vs 0x01 device ID expectation).

I really don't care which PR gets merged as long as we get working support for the board :) Happy to close mine if these hardware mappings get pulled into yours, or vice versa.

[lbibass]

I've already implemented this. #2054
Are there any major differences between my earlier PR and this one?

Hi @lbibass, I actually tried your PR first, but couldn't get it work.

I believe there are a few hardware mapping differences preventing it from working. My implementation just borrows the values from the MT firmware variant for the Duo board:

• LoRa Reset Pin: P1.10 (P1.12 overlaps with SPI NSS)
• RF Switch: Uses a 2-DIO table (DIO5, DIO6), not 4
• Flash: 8MB QSPI (not 16MB)
• Radio Class: Must use LR1121. Using LR1110 causes RadioLib to fail the hardware check on boot (0x03 vs 0x01 device ID expectation).

I really don't care which PR gets merged as long as we get working support for the board :) Happy to close mine if these hardware mappings get pulled into yours, or vice versa.

Oh yikes. Yeah, I did miss a few of those things. But the flash chip is actually a 16 MB chip. I know the spec sheet says 8, but my actual board has a 16MB chip.

In my testing, I didn't see any errors with the radiolib hardware check. But I can add those files. I'll also be getting in a SuperIO board so I can at least try adding support for the display sometime this week / weekend.