LR2021 sub-GHz TX power mismatch (configured vs measured)

[hlym123]

Hi, I found a TX power mismatch issue on LR2021 in sub-GHz mode.

At 868 MHz, the configured power and measured output differ a lot (especially at higher power settings), beyond normal tolerance.

I have fixed it locally by updating LR2021 PA/TX configuration logic according to datasheet tables and half-dBm TX params.

Reference: hlym123@f5a23d9

If needed, I can submit this as a PR.

[jgromes]

Thanks for reporting this. Actually the primary issue is incorrectly implemented setPaConfig command, after fixing that the output power is a lot closer to what it should be.

Regarding the tables for optimized confiuguration from datasheet, can I ask how did you verify the configuration? In the past (on SX126x), the default configuration from datasheet was actually not very good and we had to do quite a bit of measurements to arrive at an actually optimized PA configurations - see #1628 for details.

[hlym123]

We verified the optimized configuration using the following methods:

1. Measured the RF output power at the RF port using an RF power meter to ensure the actual transmitted power matches the expected value.
2. Measured the module power consumption using a power analyzer to evaluate the power efficiency under different PA configurations.
3. From practical testing experience, we also observed that the RSSI on the receiver side was clearly incorrect with the previous configuration, which further indicated that the PA configuration was not properly set.
4. Regarding the optimized configuration tables from the datasheet, I apologize that I have not carefully verified them yet. I mainly focused on fixing the setPaConfig issue and validating the RF output power and power consumption.

By comparing RF output power, power consumption, and receiver-side RSSI behavior, we concluded that the corrected setPaConfig configuration is closer to the expected performance.

[jgromes]

Do you have some more details on the measurements, like the actual values measured, the equipment that was used etc.? I currently don't have access to a calibrated spectrum analyzer to measure the output power accurately, but in any case once I do, I can run those measurements again so that we can check the appropriate PA settings, maybe find some more optimized ones like we did with the SX126x series.

[hlym123]

Here are more details about the measurements:

1. RF power meter used: RPM 20GS
2. Device under test: my own LR2021 module
3. The module IPX RF output is directly connected to the RF power meter input
4. The maximum TX power was configured (868 MHz: 22 dBm, 2.4 GHz: 12 dBm) and data transmission was performed

The actual measured output power is shown in the figure below.

[jgromes]

Thanks for the extra info. Using data transmission to measure the output power is not entirely accurate as the signal is modulated so you should take the bandwidth of your power meter into account. In any case I will try to repeat the measurement with calibrated equipment, and once that is done add the optimzied PA configuration to the library.

[lewisxhe]

@hlym123 The patch can restore normal power levels.

The attachment includes the latest original driver reference. I checked the transmit power setting by looking up a table to select the appropriate value.

LR2021.zip

[jgromes]

For future reference, this is being worked on in the linked branch.