fix: backport SX1262 PHY optimisations for RAK4631 from Meshtastic

[drupol]

This PR addresses several physical layer misconfigurations for the SX1262 transceiver on the RAK4631 (drawing upon proven optimisations from the Meshtastic project and RadioLib):

• TCXO Voltage: Corrects the DIO3 TCXO voltage from 3.3V to 1.8V for the RAK4631, aligning with the hardware specification to prevent frequency drift and potential component degradation.
• RTC Stop Errata (Semtech 15.3): Explicitly writes 0x00 to register 0x0902 during initialisation. This prevents the internal RTC from triggering spontaneous wakeups, avoiding silent hardware lockups.
• OCP Tuning: Increases the Over-Current Protection (OCP) limit to 140mA (0x38) for both RAK4631 and RAK3401.

This ensures the 22dBm transmit power is not artificially clipped by the default 100mA limiter, maximising range and signal integrity.

[drupol]

@attermann Can I ask a review for this PR too ? FYI - Using this on 2 RAK19003s, 2 RAK19007s and 2 Wismesh Tags successfully.

[attermann]

Apologies for not getting to this (and other) PRs yet. Life is a little demanding atm. This looks good, I just need to find the time to go over it in detail.
One question I have related to another report (link below) of LNA/PA issues on the 1W RAK13302 boards, any idea if this could be related to the OCP from your Meshtastic research?

6a485d5#commitcomment-194030534

[drupol]

I checked the current Meshtastic firmware and it applies the same 2 settings to both the RAK4631 and the RAK3401 with the RAK13302 module:

• The radio oscillator voltage is set to 1.8 volts for the RAK4631:
  https://github.com/meshtastic/firmware/blob/master/variants/nrf52840/rak4631/variant.h#L218-L220
• It is also set to 1.8 volts for the RAK3401 with the 1-watt RAK13302 module:
  https://github.com/meshtastic/firmware/blob/master/variants/nrf52840/rak3401_1watt/variant.h#L54-L57
• Meshtastic defines the radio current limit as 140 milliamperes in the common driver:
  https://github.com/meshtastic/firmware/blob/master/src/mesh/SX126xInterface.h#L34-L40
• That value is then applied during radio initialisation:
  https://github.com/meshtastic/firmware/blob/master/src/mesh/SX126xInterface.cpp#L113-L123

So yes, the increased current limit could plausibly be related to the reported transmission-side issue on the RAK13302. If the previous limit was too low, the radio output feeding the external power amplifier could have been reduced or unstable.

However, this would mainly explain transmission problems. It would be less likely to explain a genuine reception or low-noise amplifier issue for which I have no clue.

Now, take your time for reviewing, no worries, there's no emergency. I keep a local branch where I merged all my PRs, so, not an issue for me ;)