Long range 700 metre RF data transmission using multiple repeaters (in progress)

This system is a simple wireless mesh, where every unit is both a reciever and a transmitter, and also some units measure things or control something or both. A wireless mesh means that units are redundant and that messages go via multiple paths.

The simple rule is that all units listen all the time. If a data packet is received, it is processed and then after a random delay it is retransmitted. The delay decreases the risk of data collisions. The unit then sleeps for a short time. In order to prevent messages bouncing round the mesh indefinitely, each unit stores the last few packet header numbers and if those packets has been transmitted will not retransmit any of those packets.

Commands can be sent to any unit to send a measurement - eg temperature.

Some units can transmit on a regular basis.

Units that control things can send back an acknowledgement that the action has been done.

The RF modules are  315Mhz 1000mW modules from ebay (search for rf module) company is e-madeinchn but a direct search for their store does not seem to work. 433 modules are also available and either will work fine. The 433 range is possibly a little more crowded in inner city areas. Experiments with yagi antennas result in a doubling of the range, but this is directional and the concept of a mesh means that units need to transmit in all directions equally. Hence a simple 1/4 wave whip is used (22cm for 315Mhz, 17cm for 433 Mhz).

Many of the remote units need to be self powered and must be able to survive periods without any charge such as in winter when there are many cloudy days. Rechargeable batteries have the property that they hate being reverse charged and this can happen when there are two or more cells in series. A single cell does not mind going completely flat, and hence this circuit uses single NiMH AA or AAA batteries. This is the concept used in solar lights, and the other advantage of using single cells is that only 4 solar cells are needed to charge them.

This stepup circuit is from Talking Electronics.

The circuit has been modified:

1) The solar panel from a solar light did not give enough current (20mA) and was increased to 100mA. In some cases this was using new bigger cells and in some cases 2-3 solar light solar cells in parallel. For a 2500mAH NiMH, 100mA is still a trickle charge current and should not produce gassing. It is possible to use a picaxe to measure the battery volts and dump power when overcharged. The solar cells used were   the new flexible cells which are becoming available on ebay for very good prices. Current draw from the battery is 20mA and a general rule of thumb with solar is 5:1 ratio to cover night time and cloudy days.

2) The 1k was increased to 2k2. This decreases the quiescent current.

3) The 2k2 and 15k.3k3 voltage divider was changed to a 330k/22k divider which increased the output to 10.3V which is needed to power RF transmitter modules.

4) The 100uF was increased to 4700uF 16V. This provides enough energy to send out a data packet as higher power (4000m) modules use 100mA when transmitting.

5) The 10.3V supply goes into a 5V regulator to power the picaxe and the receive module. Both these need smooth supplies, especially the receive module. A 78L05 could be used but a better component is the LP2950 5V regulator which has a 0.38V drop (not important here) and only 75uA quiescent current (very important). Same pinout as 78L05. Sourced from Ebay - West Florida Components $US50c ea including shipping.

6) There have been lots of dead ends with this project. Concepts tried and rejected have been 9V batteries charged from 12V solar car battery chargers (too expensive, and when goes flat some cells are reverse charged shortening their life), running everything off the 5V circuit as above and using a capacitor voltage doubler for the Tx modules (too much hash on the 5V supply), using a single 1A 0.45V solar cell and stepping it up to 1.5V (couldn't get the right inductor value - high L, low R, and not worth the extra components), running the circuit on much lower quiescent current and having it sleep most of the time (defeats the concept of an always-awake wireless mesh), powering everything off a 12V supply from a wire (if 12V is available then may as well send the data via a wire).

There are some important choices with Rx and Tx units as there are lots of these available. For the Rx unit, the most important feature is a very low quiescent current. There are units from e-madeinchn are 200uA (search ebay for RF module brings up their website). For the Tx units, there are two types from the same supplier - a 1000 metre unit and a 4000m unit. Both work, but the 4000m unit puts out quite a bit of RF energy and nearby electronic components need shielding (even computer monitors are affected). But the Tx is only on for 1/10 of a second at most and only very infrequently as any more transmitting will flatten the battery very quickly. The 1000m unit runs off 12V and the 4000m likes 9V, and these voltages are very important as running both off a 5V supply resulted in ranges of about 4 metres and about 20 metres respectively.

Range: After two days of experiments trying to work out why the circuit will work on a workbench but not outside, I have finally discovered something that seems really obvious - the units must be at least a metre off the ground, and ideally 2-3 metres. If they are within 20cm of the ground, a 4000metre unit is lucky to go 10 metres. The units work on the testbed because the bench is a metre high. The post the units are mounted on is also quite critical and should not be made of anything conductive like steel. Wood is also a problem when it gets wet. The best sort of pole is PVC pipe. Some research on the internet reveals a number of sites that describe how the distance from the ground makes far more of a difference than the type of antenna or even the power. Eg one site describes a data success rate of 0% at 1/2 a foot and 99% at 3 foot above the ground. Another describes how 10mW modules can go more than 1km on a model rocket (these are lucky to go more than 30 metres on the ground).

More experiments are in progress as of 30th May - next step is the software to handle forwarding on of messages.

Home