Question: Why is noise_floor clamped to -120 dB?

[michal-konopinski]

I noticed that noise_floor is clamped to -120 dB in the code:

src/helpers/radiolib/RadioLibWrappers.cpp:

if (_noise_floor < -120) {
_noise_floor = -120;
}

Is there a reason why noise_floor is clamped to -120 dB?
I’m working with devices that report a noise floor of exactly -120 dB, so this makes it impossible to tell whether the value is real or the result of clamping.

Is this limit based on hardware constraints or historical assumptions?

Thanks!

[luigi311]

Found the comment that brought that line in but doesnt really explain anything
9bcab09

Searching through the discord though there are a few people referencing that if the noise floor goes to -120 or lower it likely means somethings going on with the radio. I wonder if that was added in so it just doesnt go to a crazy number when that issue occurs.

Hopefully Scott Powel can shed some light on why that change as added in, not sure how to contact them

[IoTThinks]

I guess is due to the specification of SX1262.
The minimum receive sensitivity is around -120dBm.

You may remove that code to see much much noise floor you can receive?
For me, -100 dBm noise floor is too good to be true in my place already.

[Vasili-Sk]

i see -118 on some days, if clean frequency selected

[robrec]

hard caps dont make sense. there are a lot lora chips there that will work - with other specs.

[IoTThinks]

I think no harm fixing it.

I guess we can take a look at Radiolib for updated code for that part.

[mrlmuk]

If the device is a Heltec v4 saying -120 nf, then the RF switch is stuck in TX mode disconnecting the antenna during RX.

The current firmware lacks the code control the FEM GC1109 (PA/LNA) discussed here #1249 It was later discovered via email from Heltec and Semtech)that there is an undocumented register which allows the sx1262 to control the GC1109 correctly and completely sorts this out. leading to a simpler solution. #1398

When this first started happening to mine I found this limit code and wondered if it was right for the v4 how it was working out the nf.

[michal-konopinski]

In my case, not Heltec but E22P, but also containing rf switch, connected to DIO2 , nf after boot -96 db, after ~21 hours drop below -120, this is repetable after reboot. I will try SX126X_REGISTER_PATCH.

[mikecarper]

https://docs.google.com/spreadsheets/d/1WsCO8oMMIiYBY3XaWEMp2ufASr_gcWbFBHe3ATnXJbs/edit?usp=sharing
I got rid of the very slow presets at the top; anyway you can see for USA -127.5 is the floor. Can look at other settings