I am currently working on getting a MM6108 based solution (Using the Quectel FGH100 module) running. I have update my Ras-pi based development environment to use the 1.14.1 release drivers/firmware/patches and currently am using the default bcf_mf08551.bin firmware (because as best as I can tell that is the closest to the FGH100MABMD module) as I don’t have a 1.14.1 BCF from quectel.
Using the Wavemon tool I noticed that the signal levels are in the yellow and red (depending on which antenna I use) and am looking to vary the TX power to see if I can boost this setup. The 2 development units are across the room from each other (~10m) and have practical line of site (Only a low cubical divider in the way) so the lower signal levels are surprising to me.
Is there anything in the morse_cli that I can use to adjust this?
I have a question, you said you are using “Quectel FGH100 module”, does it work in 860 MHz frequency, I am working on another project and I need a module that works in this frequency. thanks for the answer.
There is nothing in morse_cli which will let you boost the power. The power is dictated by the design and the BCF. Your best option is to contact Quectel for a 1.14.1 compatible BCF.
You are however, correct that the mf08551 is close to the BCF required for the FGH100M, so it is a bit alarming that the reported signal is lower than expected. What are the measured signal and noise values? What is the reported transmit power?
@Asma, the FGH100MABMD is US/APAC only so will only operate in the 902-928 MHz range. Have a look at the FGH100MAAMD which will be capable of operating across a wider range, but with a reduced transmit power (13 dBm), to meet European regulatory requirements.
Of course, please confirm this with Quectel, as they will have the most up to date information on their modules
We have contacted quectel for a compatible 1.14.1 BCF and received that. Using wave mon we are now seeing a link quality of 69/70 and a signal level of -41dBm (0.08uW) with a tx power of 18dBm. (It is using 924MHz, Channel 44 with 8MHz bandwith as that is the Australian frequency with the most bandwith possible).
Using the morse_cli stats function we are seeing received power (dBm) of -44 and noise of -80. Running UDP based (as it gives a better indication of dropped packets) we are seeing spikes of upto 16Mbit/s but usually when it spikes up into MCS6 we see the retries starting to jump causing it to go slower again.
Is there anywhere I can be looking for tuning settings to reduce this or to get an indication of where this fault is occuring (eg is the module detecting an errror, is the client rejecting packets etc?)
There’s a number of places throughout the stack which could contribute to behaviour like this. To rule out the HaLow radio, the morse_cli stats command is a good place to start diagnosing.
Run morse_cli -i wlan0 stats -r to reset the statistics, then start your iperf. Run morse_cli -i wlan0 stats periodically during the iperf to dump the statistics. Periodic logging can be helpful in diagnosing the interference - if any - in the environment.
It could be the network stack itself contributing to this behaviour, so checking the output of sysctl -a to examine the rmem and wmem would be beneficial.
Also note, that if using fq_codel, tuning of the parameters can be required due to increased packet transmission time over higher speed network links. An easy test here is to temporarily switch to pfifo to see if there is any improvement.
