LoRa experiments in the 70 cm band

Two weeks have passed and the sensors are still sending data without problems.
On the graph, the water temperature in several lakes.

.

It's also cool to see where I used a lithium battery from a different company in one unit and the voltage fluctuates differently.

Why does the red have such sharp pins and the yellow is so smooth? Some depth of immersion or access to sunlight?

>>21633519 Sensors at a depth of about 1m, it varies sometimes 2m from the shore sometimes 10m, access to the sun is at this depth but that's not it. There are quite a lot of different vessels floating on this lake, I have a PTZ camera there and I have already observed that when something passes close by it just makes such temperature fluctuations.

In the graph the white colour is the water temperature and the coloured ones are the outside temperature next to the lakes.

.

There is a noticeable difference in transmitter range between the stick and the built-in "spring" antenna on the SX1276 ?

>>21742010 No. A stick antenna is also a spring only on a piece of wire which introduces loss, but the gain is due to being able to place the antenna in a more convenient location - away from metal parts or anything that will jitter it. My 'network' already has several gateways and 9 transmitters within a 15km radius. I've set myself a page where I can keep track of what is being received and how - http://meteo.lipowa.net/lora_log/

Fridge_TP nice diagrams.

I see on the charts that the average RRSI over -100 and no problem with reception, I have a problem receiving all the packets from the basement to the fifth floor of the same block, at my place RRSI between -102 and -107, SNR -20. Out of 3 short packets sent 1-2 arrive.

>>21742985 I already have a total of 6 gateways with different antennas in different locations, from observation the most important thing is the background/noise level, with a shorted antenna input the SX1278 indicates -137 dBm with a 125 kHz channel width set, the worst radio conditions I have are on the chimney of a heating plant, even though it is very high there - about 100 m, in a very short distance there is a whole bunch of different devices, radio links, sector antennas of probably all operators, a few of my sectors at 5 GHz, a few radio links at 60 GHz, and a tangle of twisted-pair cat7 causes that on a small antenna the background does not go below -100 dBm and despite the fact that this gateway is the highest it receives the worst and the fewest packets compared to the rest. The gateway works best on a 50m pole where there is no mobile operator and only my toys at 5/60 GHz.

Six months have passed. 3 of the 8 temperature transmitters scattered around the site are still working properly, 4 have been destroyed by animals - mostly beavers. They knock over the trees on which the transmitter was mounted, bite through the wires to the sensor, so that there is a short circuit and the battery runs out, or so that only one sensor is working. One has been stolen, simply disappeared.

During this time I found a few bugs in the code - in the transmitter the temperature values were incorrectly sent from -1.0 to +1.0 which only came out in winter, in the gates I found a problem that caused unexpected reboots of the device, but there I have implemented OTA so I can do the update remotely.
By the way, I also made PCBs for the transmitters so I don't have to solder it on wires all the time.

Water temperature for the last 20 weeks



New tiles for transmitters

dktr wrote:

the wires to the sensor are snapping and it's so that it makes a short circuit and drains the battery,

Maybe some sort of mosfet power cut off? I've always wondered, where frequent readings aren't needed, would such a solution save some energy and reduce the sensor self-heating?

>>21790461 You don't need a mosfet, you can power the sensor straight from the microcontroller's GPIO, nothing heats up there, at idle it draws about 700 nA.

Then do beaver detection and disable the port after a few failed communications.

How to make a beaver detector?

Well that if it chews through the cables and you lose communication with the sensor then you take the power off.

I always enjoy such practical trials, the lessons and practice one learns after such long-term testing can be useful in many other devices. I am usually amused when malcontents attack a practitioner's idea on the basis of "it can't be done, it won't work, it's wrong" putting their assumptions versus the experience of someone who has tested their ideas in practice.

If there were to be more sensors I thought that perhaps a circuit from LoRa could collect measurements from nearby sensors with a cheap OOK/FSK radio. Such cheaper sensors within tens/hundreds of metres could, for example, be sealed and collect information close to the water. The only thing to prevent such a LoRa 'hub' from wasting energy listening for transmissions from simple sensors is that after detecting the first transmission repeated every X amount of time, the radio would go into listening mode one second before the expected transmission.

I don't know if this approach with cheaper sensors and hub sensors from LoRa makes sense. The only thing that seems tempting to me is the very low price of an OOK (possibly FSK) transmitter and such a module could have a shorter range, perhaps it could be partially submerged in water? A 'hub' type sensor placed in a tree could collect data from sensors in the area and send over LoRa.

No it won't work, listening to the RX all the time draws too much power, and it would be difficult to synchronise it all and only run it when needed. Currently I have changed the type of housings to ones with 4 holes and screw it to the tree with 4 screws, hardly anyone walks in the woods with a screwdriver, I changed the cable to the sensors from random thin wires (like in headphone cables) to a thick microphone cable. There is a sticker on the case saying not to destroy and a link to a page where there is a map with measurements.

A uniform sensor network has its advantages and the lack of listening will extend battery life. I have noticed the idea of listening at certain times being used in battery-powered weather stations.

Local reception of data from wireless sensors via a battery-powered gateway is as possible. The scan window has a small fill and the average current drawn by the RX is negligible. No synchronisation is required. The sensors only need to send data frequently enough.