T096 - 1.17 repeater radio.fem.rxgain not working

[godric]

TL;DR: radio.fem.rgain is not persisted across boots and does not seem to have an effect at all.
Fixed instantly by flashing back to 1.16

Hardware / firmware

• Heltec Mesh Node T096 (nRF52840), repeater role
• 868 MHz, cavity filter, 3 dBi omni, solar powered, urban environment
• Clean 1.17.0 flash with data erased (or really any method I tested).

First problem

set radio.fem.rxgain off applies at runtime but does not survive a reboot. After restart the value reads on again.

Steps to reproduce:

• get radio.fem.rxgain -> on
• set radio.fem.rxgain off
• get radio.fem.rxgain -> off
• reboot
• get radio.fem.rxgain -> on

Which, in the context of a solar repeater, makes the setting pretty much useless.

Second problem

Setting radio.fem.rxgain off (without a reboot) does not seem to have any effect (tested over the course of a few hours)

Admittedly I can't test directly wether or not the LNA is enabled, my only evidence is the degradaed noise floor (~7dbm) & RSSI (~20dbm) associated with the RX gain.

Steps to reproduce (on 1.17):

• set radio.fem.rxgain off -> nothing
• Flash 1.16 (keep settings) -> Noise floor and RSSI go back to normal

[alustenberg]

was throwing a bunch of debug statement around the config read/write for issue #3151

looks the call to set radio.fem.rxgain off does apply to me, immediately shifting from -99 to -117 noise floor. but, the write back to config has the value set to true.

DEBUG: VL53L0X not detected at I2C address 29
DEBUG: BMP085 not detected at I2C address 77
DEBUG: read string key=name token=Unset T096
...
DEBUG: read uint8 key=rxgain token=1 val=1
DEBUG: read uint8 key=fem_rxgain token=1 val=1
DEBUG: read int8 key=tx token=18 val=18
...
DEBUG: RX Boosted Gain Mode: Enabled
DEBUG: RadioLibWrapper: noise_floor = -99
DEBUG: write string key=name val=Unset T096

(send set radio.fem.rxgain off)

DEBUG: write obj key=radio
DEBUG: write float key=freq val=910.525024
DEBUG: write float key=bw val=62.500000
DEBUG: write uint8 key=sf val=7
DEBUG: write uint8 key=cr val=5
DEBUG: write uint8 key=cad val=0
DEBUG: write uint8 key=int_thr val=0
DEBUG: write uint8 key=rxgain val=1
DEBUG: write uint8 key=fem_rxgain val=1
DEBUG: write int8 key=tx val=18
...
DEBUG: RadioLibWrapper: noise_floor = -117
DEBUG: RadioLibWrapper: noise_floor = -117
(reboot)
...
DEBUG: read float key=freq token=910.5250 val=910.525024
DEBUG: read float key=bw token=62.5000 val=62.500000
DEBUG: read uint8 key=sf token=7 val=7
DEBUG: read uint8 key=cr token=5 val=5
DEBUG: read uint8 key=cad token=0 val=0
DEBUG: read uint8 key=int_thr token=0 val=0
DEBUG: read uint8 key=rxgain token=1 val=1
DEBUG: read uint8 key=fem_rxgain token=1 val=1
...
DEBUG: RadioLibWrapper: noise_floor = -99
DEBUG: RadioLibWrapper: noise_floor = -99

poked arouhd, and in the NodePrefs.h file there is the following... which doesnt seem right.

      def("rxgain", _parent->rx_boosted_gain);
      def("fem_rxgain", _parent->rx_boosted_gain);

set radio.rx_gain off, restart:

DEBUG: write obj key=radio
DEBUG: write float key=freq val=910.525024
DEBUG: write float key=bw val=62.500000
DEBUG: write uint8 key=sf val=7
DEBUG: write uint8 key=cr val=5
DEBUG: write uint8 key=cad val=0
DEBUG: write uint8 key=int_thr val=0
DEBUG: write uint8 key=rxgain val=0
DEBUG: write uint8 key=fem_rxgain val=0
...
DEBUG: read uint8 key=rxgain token=0 val=0
DEBUG: read uint8 key=fem_rxgain token=0 val=0

so, updated that and reflashed.. and now my gps works?!?

DEBUG: lat 39... lon -84.... alt 240.400000
...
DEBUG: read uint8 key=rxgain token=1 val=1
DEBUG: read uint8 key=fem_rxgain token=0 val=0

simple fix:

       def("rxgain", _parent->rx_boosted_gain);
-      def("fem_rxgain", _parent->rx_boosted_gain);
+      def("fem_rxgain", _parent->radio_fem_rxgain);
       def("tx", _parent->tx_power_dbm);

in src/helpers/CommonCLI.h and examples/companion/nodesprefs.h

[alustenberg]

so this works with MESH_DEBUG set, but does not apply when it is not set. likely related to the possible SPI root cause on the TX power bug previously mentioned.

[alustenberg]

#3137 is cleaner and was opened up 3 days ago. fix should flow through that PR IMO.