RMF98PW Spreading Factors

[Mikefly123]

We are currently messing around with changing the spreading factors in the the pycubed_rfm9x library and are running into an unexpected issue.

The Issue
It seems like whenever we increase the spreading factor (i.e. from 7 to 10) the RSSI for the link get worse or sometimes it doesn't close at all. My understanding is that increasing the spreading factor should add to the link margin, so this is pretty confusing behavior.

Setup
We are currently testing with two PyCubeds, both with RFM98PW radios in the Radio 1 slot. There are no differences in the configuration of either PyCubed, but one is connected to a custom dipole antenna and the other is connected to a yagi. The link closes consistently in the far field and near field at SF7 or SF8. It starts to miss packets at SF9 or SF10. The link does not close at all in any circumstance at SF11 or SF12.

[maholli]

Hey Michael!

By increasing the spreading factor you are significantly increasing the symbol time and therefore the time on air (assuming you're keeping bandwidth the same). Semtech addresses this with their "low data rate optimization" parameter.

1. Set cubesat.radio1.low_datarate_optimize = True after configuring your higher SF and I'm betting all your problems will disappear :) Always make sure both transmitter and receiver(s) always have the same low_datarate_optimize setting.

2. That begs the question, when do you need to set low_datarate_optimize? I did a quick search in the sx1276/77/78/79 datasheet and found that Semtech "mandates" low_datarate_optimize whenever symbol duration exceeds 16ms. Here's the excerpt:
   

   • At the library default BW125: SF7 has a symbol time of ~1ms, whereas SF12 is ~32ms!

So try setting cubesat.radio1.low_datarate_optimize = True in your code and let me know how it goes! :)

Max

P.S. fun fact: low_datarate_optimize wasn't part of common RFM9x libraries until I realized this and added it for PyCubed and pushed upstream. Yay open source!

[Mikefly123]

Just wanted to write a very quick follow up to this. The low_datarate_optimize toggling does indeed help catch the packets at different spreading factors! Another thing I had to implement as well is variable waitTime or else the receive function appears to actually time out before the packets are fully read. I've included some really basic code for this:

spreadingFactor = 7
waitTime = 2

cubesat.radio1.spreading_factor = spreadingFactor

if spreadingFactor <= 8: 
    cubesat.radio1.low_datarate_optimize=False
    waitTime = 2
elif spreadingFactor == 9:
    cubesat.radio1.low_datarate_optimize=True
    waitTime = 3
elif spreadingFactor == 10:
    cubesat.radio1.low_datarate_optimize=True
    waitTime = 5
elif spreadingFactor == 11:
    cubesat.radio1.low_datarate_optimize=True
    waitTime = 6
elif spreadingFactor == 12:
    cubesat.radio1.low_datarate_optimize=True
    waitTime = 10```

I pass waitTime into the receive function for the rx_timeout. For some reason SF12 still doesn't seem to work no matter how long you wait. Will update if there is any further development.