I have an esp32 gateway connected on a 8dBi collinear antenna. I am testing my gateway with another esp32 which is on the top of my car and whitch sends position indefinitely going trough sf7 to sf10 (antenna 3dBi).
Evering works fine and I reach a distance of 3.6 km with sf7 very easily.
For the other sf values, I can only reah 500…800m.
Do you have an expanation ?
is it due to these cheap boards ?
is it because the car is moving ?
yes every thing is 3d printed.
I like very much my shark antenna.
what could be better than the roof of the car as a ground plane ?
Yes, I would have expeted that my results would have been better with higher SF and here it is just the reverse !?!
Why ? If I reach 3.6 km with SF7, it means that my antennas are so bad than that, no ?
My range should be better with sf8 and above
Indeed, I do not like the very tiny coax provided with the esp32. I plan to replace it with a rg316.
I think the overall setup should be ok if functioning at SF7. Only a question - who knows - did you wait sufficient time at high SF? Because duty cycle limitations allow no more than one packet every 2-3 minutes at SF12.
The node transmit from sf7 to sf10 continiously with a delay of 2s between ecah SF. after SF10 it restarts at SF7.
It works because this procedure is ok for small range.
I borrowed an imst880 from a friend and with the same antenna setup I had good range capabilities with high SF. Now, as I am back to esp32 as gateway, I only get good result with SF7
That software as I recall, uses a single RFM95 and hops around spreading factors using CAD to detect whether there are signals.
I would strongly suspect the issue you are seeing is an artefact of the ‘proof of concept’ way in which the software attempts to act as a multi spreading factor gateway on a single RFM95.
LoRaTracker nailed it, I did not consider the gateway limitations . In the past I had a couple of ESP32 / ESP8266 single channel gateways and the SF hopping feature has never been really reliable (in the documentation a shorter range is also mentioned). Consider also that sending every 2 seconds is almost impossible (not only for duty cycle, but also because after sending the system is waiting for RX1 and RX2).
I should have more clear : I do not send something every 2s but I wait that a transmission is totaly finished before waiting 2s and considering start a new one at a different SF.
A complete cycle from SF7 to SF10 take about 2 min.
Ok, anyway to test range lock it at SF12; from the readme, “The CAD functionality comes at a (little) price: The chip will not be able to receive very weak signals as the CAD function will use the RSSI register setting of the chip to determine whether or not it received a signal (or just noise). As a result, very weak signals are not received which means that the range of the gateway will be reduced in CAD mode.”
(by the way, which RSSI and SNR do you have at SF7, 3.6km?)