Hardware V2.3 analog input issue - Davis wind direction only north

[thezenox]

This mainly affects stations using DAVIS 6410 (lets call them 'analog') sensors

It seems TVS diodes in hardware V2.3 and earlier are not sufficient to protect the ADC input against electrostatic coupling in nearby thunderstorm conditions. I got reports of stations only showing 'north' direction.

This looks like partial analog input damage, most likely from a nearby lightning/ESD surge coupled into the wind sensor wiring.
The SAMD21 ADC itself is probably still working, but the input protection / analog front-end (ESD clamp diodes or input mux MOSFET) was likely degraded.
That creates leakage or extra loading on the ADC pin, so the input can no longer reach full-scale voltage.
The result is a compressed ADC range (e.g. max ~360/500 instead of 1023).

To check the ADC raw values use the Analog Test mode here
North raw ADC values should be close to 0 (a bit clockwise of 0°N or close to 1023, a bit anticlockwise to 0°N)

BDCFO3
N    0
O    260
S    540
W    788
NNW    960
N    1023
-----Working-----

BDFEF3
N    0
O    5
S    7
W    14
NNW    55
N    360
---North only----

BD7D81
N    0
O    14
S    20
W    40
NNW    105
N    500
------North only-----

To fix this issue, I replaced the TVS diode array by a TVS + clamping diode array (D1213A-04TS-7) on V2.4 and added series resistors in front of the inputs. Test show no distortion of the readings.
If this protection is sufficient now, I don't know.

To upgrade V2.2 and V2.2 PCBs I will create a external board that can be connected to the Davis and UART connector, archiving same level of protection.

Big Thanks to Alex & Knut for their efforts and time on this issue