LoRa Communications Problems with Long Airtimes

I am building a 900 MHz LoRa system from Miniwireless Anarduino modules and the RadioHead library.

I am looking to test for maximum range with very small packets (17 octets).

I find the system works fine when the total frame airtime is less than 1 second, begins to deteriorate (high packet error rates) when the frame airtime is about 1 second and becomes totally blocked (no packets successfully received) above that.

This seems to be correlated with high spreading codes and narrow bandwidth.

For example: 31.25 KHz @ Sf9 (.85 sec airtime) appears to work fine (as well as many other mode combinations with shorter airtimes).

I begin to see packet errors (50%+?) at 1.1 sec frame airtime (125KHz @Sf12, 62.5KHz @Sf11, 32.25KHz @Sf10). I cannot seem to get communication at 31.25 KHz@Sf12 (4600 msec), 62.5 KHz @Sf12 (2300 msec), 31.25KHz @Sf11 (2640msec). 31.25 KHz@Sf10 (1320msec)seems a bit anomalous - one software does nothing, the other gets a few packets thru.

Any thoughts?

Hi Ken,

Curious what library you are using for the LoRa's.
We have used the radiohead library quite successfully in several projects.
Seems to be of solid quality, and there are several working examples.
Here's the link for their LoRa class for the RF95.
http://www.airspayce.com/mikem/arduino/RadioHead/classRH__RF95.html
Here's the header file:
http://www.airspayce.com/mikem/arduino/RadioHead/RH__RF95_8h_source.html

I suggest taking a detailed look into the register settings, rate, bandwidth, spreading factor, etc.
Lots of trade-offs. It seems to me this will hold the key for improvement. It sounds like you're already doing quite a bit of this already.

Another place to look is doing some searches on the Semtech LoRa parts. For example, the RFM95 is based on the Semtech SX1276.
I would bet the documentation for the Semtech might have additional details.

Very curious to find out how things are going, and hopefully you will post more here for the benefit of others.
I've heard rumors of excellent ranges, but what range are you finding? What obstructions to you have?

Rick:

We are using the RadioHead library. The question is about programming the wide variety of RH95 encoding modes. Their standard (four) modes for the RH95 work fine, but are not the longest range modes (about 8-10 dB in link budget short). So we have been exploring using the RadioHead library to directly program the configuration registers to get the highest range possible. That seems to work well, but we find a consistent problem with RH95 modes in which our short packets (17 octets of payload) exceed ~1 second of airtime. These VERY long packets should have the best link budget and best processing gain from Semtech's spread spectrum. Should give the very longest range.

For the modes I indicated in the message above, we see problems. That is, we either get no communication, or rather high packet errors.

We have an NDA with Semtech and they tell us to ask the radio module manufacturer. Can you answer the questions or refer us to HopeRF support?

We will share more of the range data as it becomes ready to publish ... but we suspect much longer ranges are possible with the right antennas and equipment location.

Ken

Hi Ken,

Thanks for your post, there is additional information on this in other post, at:
http://forum.anarduino.com/posts/list/110.page