Understanding SX1262 PaConfig / TxParams

[devRomeos]

Hi There,
I noticed that the library defaults to 0x04-0x07 for the paDutyCycle and hpMax parameters regardless of the power output. As all in this library, I'm sure this was done like that for some good reason despite from the datasheet table, these do not appear to be the most efficient parameters.

Following the table it would seem that the ideal parameters for 14dbm (as example) would be 0x02-0x02, but it is a bit confusing as it seems that you should always put an output power of 22dbm (as the last row says) and modify paDutyCycle and hpMax instead?

Just out of curiosity. I attach the SX rf output to a powermeter (that should not be perfectly calibrated) I have gotten the following results:

Traditional way
radio.setOutputPower(14) +15.1dbm - Power draw +55.6mA
radio.setOutputPower(22) +23.8dbm - (not power measured)

Per PA table 14dbm recomended

 radio.setPaConfig(0x02, 0x00, 0x02, 0x01);
 radio.setTxParams(22, RADIOLIB_SX126X_PA_RAMP_200U);   +11.9dbm -  Power draw +38.5mA

Per PA table 14dbm recomended, increasing hpMax +1

 radio.setPaConfig(0x02, 0x00, 0x03, 0x01);
 radio.setTxParams(22, RADIOLIB_SX126X_PA_RAMP_200U);   +14.8dbm  - Power draw +43.6mA  

Per PA table 14dbm recomended, increasing paDutyCycle and hpMax +1

 radio.setPaConfig(0x03, 0x00, 0x03, 0x01);
 radio.setTxParams(22, RADIOLIB_SX126X_PA_RAMP_200U);   +15.2dbm  - Power draw +45.7mA  

I would still like to test if this translates into the same transmission distances, as on paper there seems to be an efficiency gain of +16%. Do not take the absolute consumption as a reference as it will depend on other factors such as packet length, ldo, frequency etc.. but it could be a good starting point.

This was test on a Heltec V3 board (heltec_wifi_kit_32_V3)

Looking forward to hearing your thoughts

[jgromes]

I'm sure this was done like that for some good reason despite from the datasheet table, these do not appear to be the most efficient parameters.

There certainly was an intent behind this - when I was first writing the SX126x driver, I didn't have any equipment to check the actual output power. So I defaulted to what seemed like the safest option, i.e. setting the maximum paDutyCycle and hpMax parameters to avoid having to scale the setTxParams value to reach what the user actually commanded.

it is a bit confusing as it seems that you should always put an output power of 22dbm (as the last row says) and modify paDutyCycle and hpMax instead

Indeed that is what the datasheet implies. But this confusion was why I wanted to avoid messing with the PA parameters, despite having the device running at a suboptimal efficiency.

We can revisit this and optimize - what I'm a bit confused by is what flexibility is there in the PA configuration. For example, can we select any paDutyCycle/hpMax value or should we stick to those in the table?

[devRomeos]

It definitely looks like you can play with the parameters to get different powers.

I made a long distance test, almost at the limit of reception and I remotely changed the power from ‘traditionally’ +15dBm to ‘recommended table’ +20dBm and I started to lose packets and/or receive them with worse RSSI.

 radio.setPaConfig(0x03, 0x00, 0x05, 0x01);
 radio.setTxParams(22, RADIOLIB_SX126X_PA_RAMP_200U);   

I have to confirm this as were a quick test out of my intentions, it would be nice if someone more experienced than me could test it as well.... but seems that despite there are more power being emitted (as per powermeter confirm) this translates into worse range. Perhaps this is where antenna balancing and other things come into play where maybe we shouldn't get involved.