Every value here comes from the firmware's own pin map for this board, cross-checked against LilyGO's sources. Nothing is aspirational.
The nRF24 chip-select pins are triple-booked with UART0 and the external header, so serial logging and 2.4 GHz traffic can interfere, and LilyGO's own issue tracker notes some Plus units losing nRF range over time. And because the SD card, the CC1101 and the nRF24 all share one SPI bus with the display, every driver has to cooperate on that bus rather than assume it owns it — a lesson this port learned the hard way. The sharpest example: the CC1101’s SO pin is also its GDO1, so unless it is left three-stated it drives that shared MISO line even while its own chip-select is idle — quietly corrupting every nRF24 read, pulling one-bits down while zero-bits stayed clean. It presented as flaky hardware for weeks. It was one register.
One codebase, selected at build time. What changes is the shell around it.
| T-Embed CC1101 Plus | Cardputer-Adv | |
|---|---|---|
| Display | 1.9 inch, 320×170 | 1.14 inch, 240×135 |
| Input | Rotary encoder + 2 buttons | Full QWERTY |
| Sub-GHz | Onboard CC1101 | Hydra hat |
| 2.4 GHz nRF24 | Onboard (Plus SKU) | Hydra hat |
| NFC | Onboard PN532 | None |
| IR | TX + RX onboard | TX onboard |
| LoRa + GNSS | Not onboard | LoRa hat |
| Wired Ethernet | Not onboard | W5500 hat |
| Extras | RGB ring, fuel gauge | Speaker, mic |
Neither wins outright. The Cardputer keeps a real keyboard and reaches LoRa, GNSS and wired Ethernet through hats. The T-Embed folds sub-GHz, 2.4 GHz, NFC and IR onto one board and adds a fuel gauge. Pick the body the mission wants.