I reversed engineering the Wacom I2C EMR pad taken from Lenovo YOGA Book, first started by probing all pads connected to the Wacom board. There're 8 pins total connected to touchscreen chip board which is connected to YOGA's mainboard. the main character of this project is W9013 the REALLY SECRET Wacom proprietary chip that ONLY Wacom engineer team has access to datasheet(Try search it up, You won't find it ;P). I started by peeling of the copper sheet protecting the board against EMI, since this technology (Wacom EMR) rely on radio emission similar to RFID and NFC. take a look at the second pic in project gallery. I labeled 6 pins plus the main Vcc/Vdd rail. all the pins labeled with number are corresponding to FPC connector pin order
2. SCL (No pull-up D:)
3. SDA (No pull-up D:)
5. IRQ, goes low when w9013 is ready to send (pen present)
6. Chip Reset Active low, useful for re-establish/initialize I2C comm.
7. 3v3 Vdd
(I suspect that the remain pins are used for IRQ on other detection, Because this chip is very similar to one used in Samsung Galaxy Note 8.0 tab, btw my dad own one of these, might try probing later too).
The host seems to used 1MHz I2C, but I confirmed that the w9013 works with 400kHz too (on my F1C100s and STM32).
right now, I ported the Wacom i2c driver from mainline Linux kernel to STM32 (using STM32CubeIDE). I currently using the STM32F3 DISCO board, because I can share the I2C bus with onboard sensor (Pull-up already there 10k). also the onboard ST-Link provide the 8MHz clock for HSE and Serial ACM for debugging. It's worth buying board (I've been using this board for 3 years now Since Grade 10, 3 years ago).
the STM32F3 is capable of USB device (USB gadget I would say). And I use the USB HID feature and writing my own HID descriptor. The tool that I use is HID descriptor tool from usb.org (by the way, the latest version is younger that me just ~1 year xD). but the output of the program is in non-friendly-C format, so I literally wrote entire thing in C array. Right now I be able to get the HID partially working, the "evtest" in Linux can detect the HID event, but the problem that still persist is the cursor will never work on both X.org and Windows 10. I might need to take a look at HID descriptor more to fix this issue.