ELECTRONICLE

I am no longer using Blogger. You can find my new Blog by going to TheArduinoGuy.org or by clicking HERE.

See you there.

Mike

In an attempt to increase battery life further I've now got the HTU21D humidity and temperature sensor powered by a digital pin from the ATMega328p. In code the sensor is powered up and then back down again in-between reads in an effort to squeeze some more juice out of the battery. I've left the transmits/reads at every 2 seconds again so I can compare the voltage loss with what I've been having over the last few days to see if there is an improvement.

A test of 24 hours should be enough to ascertain if the voltage loss is the same or less than it was previously. Before I made this change the circuit was losing about 0.27 volts per day which would give the battery a life of about a week with 2 second reads/transmits. Ultimately the sensor will be set to only wake up every 60 seconds, which will give a much greater battery life. 

I've had the wireless sensor node with sleep functions transmitting every two seconds for a week now to see how fast or slow the power loss was. After about 5 days the battery lost 1 volt which is more than expected.

After reviewing the code it would appear I was powering down the micro-controller but not the radio module. So after amending the code to ensure that the radio module was also placed into sleep mode (using the Lowpowerlab libraries, not the Jeelib libraries) I ran it for testing purposes again. Overnight the voltage has dropped 0.13 volts. Something is obviously still not right.

After some discussion on Twitter it would appear I should also be powering up and down the sensor. So instead of having it powered from the 3.3v power rail it should be powered by an output pin so the power can be turned on and off in-between reads and transmits. So, I will be amending my code to allow for this and then running further tests to see how this impacts the power loss.

I am aware that 2 seconds is a lot and as such the battery is not going to last years like that. However, I would expect even with 2 seconds per reading that the battery should still last a few weeks at least. Once I am happy that I am achieving this then the node will be set to transmit only once every minute to conserve power.

Today I decided to change from using the Jeelib library for the RFM69 and instead i've started using the RFM69 library from LowPowerLabs. Not only is it easier to use than the Jeelib library, but it is optimised for using the RFM69 whereas the Jeelib library is designed for use with the RFM12B. I can now set baud rates and other settings which were impossible using the Jeelib library and in decreasing the baud rate down to 9600kbps i'm getting a much increased range (at the expense of slower speeds and slightly more power use).

I've made some great progress with the sensor node project over the last two days. After finally getting around the problems I was having with the software bugs and the SPI restrictions I settled on the circuit you can see on the left.

This is a breadboarded Arduino made up of the ATMega328p chip with appropriate resistors and an oscillator. Connected via SPI is the RFM69CW radio Tx/Rx module and via I2C is an HTU21D humidity and temperature sensor.

I've also added an LED to Digital Pin 9 to give a brief flash after each transmission as feedback the unit is working,

Everything is powered by a 3.7v 850mAh lithium rechargeable battery. The battery passes through an MCP1702 voltage regulator to give a nice stable 3.3v for all of the devices. I've added some sleep functions to the code so that in-between transmissions the MCU powers down for 10 seconds before waking and repeating the data transmission. This means for the majority of the time the transmitter is using micro-amps and then only milli-amps for a few milli-seconds. Overall a very efficient state which should easily give a battery life over well over a year.

The data packet contains the node's unique ID number, the battery voltage and the two sensor readings (Temperature and humidity).

I've got the circuit as small as possible on the breadboard (see pic). The next stage would be to transfer it onto a stripboard and solder it all together. Finally i'll design and etch a PCB. I could use SMD parts to make the entire thing even smaller still and swap out the ATMega for an ATTiny which is a smaller chip.

I want to make 2 or 3 different sensor nodes with varying sensors on them (temperature, humidity, pressure, luminosity, etc.) so i've got packets coming in from several nodes. This will then allow me to start writing the code for the base station which will receive the data, time stamp it, store it on an SD Card and then sporadically upload it to a web server.

I've got 433MHz and some 868MHz RFM radio modules and also an SRF radio module from Ciseco which claims to have longer range. These will be all tested out in time to see which is best for penetration through walls, etc. 

If you've been following the blog lately you would have read about the issues I had with transmitting data packets over a wireless netwrok between two Arduinos using an RFM69CW radio transceiver module.

A quick recap - I first had lost packets and lots of CRC errors in transmissions. These turned out to be a bug in the Jeelib library. Then I put an L7805 5v to 3.3v rectifier on and found it wasn't suitable. This caused lots of problems with the radio. Changed that to an MCP1703-33 which is much more stable and better suited to battery operation too. That fixed those issues.

Then I found out that I couldn't connect any other SPI device to the BUS as the RFM69CW also uses SPI and is controlled via timer interrupts. This was corrupting comms over the SPI Bus to any other devices as the interrupt was hijacking the bus on regular intervals. I tried a DS3234 Real Time Clock IC and the data were all wrong.  I could bit-bang another SPI bus on non SPI pins but this is a lot of extra work. Instead I opted to switch to an I2C RTC (DS1307). I'll make sure all the sensors I use are I2C to avoid any further problems.

So, now i've discovered the above limitations with the hardware and software and have a better understanding of what i'm up against i've made a lot of progress. The sensor nodes are now transmitting data successfully with no lost packets or CRC errors. I have yet to do any range tests yet or run them of battery (with power saving sleep modes), that is the next stage in the project..

I have no idea why but my experiments with the RFM69CW have descended into chaos and frustration.

At first I had the units talking to each other and sending out test packets for nearly 24 hours with no issue. Then I started to modify my code and add functionality and they stopped working. So, I presumed I had broken something in the code and so reverted back to the RF12Demo sketch and my original code to test them again and they still didn't work.

So I rebuilt the circuit and started again. This time I had some limited success but it shortly started to fail again then stopped working altogether. I've swapped the radio modules over and still nothing.

I then started to use the RF12_easySend() function instead of RF12_sendNow() and the transmissions started to work. Then the receiver stopped receiving the packets, not even packets that failed CRC, nothing.

I've wasted hours and hours on this and it's nothing but frustration. Is it the Jeelib library? Do I have faulty radio modules? Who knows. I am at a loss what to try next.

It turns out that all of the CRC errors and the packet loss I was experiencing was nothing to do with interference at all. It was a bug in the library!

After contacting the guys on the Jeelib Cafe forum I was made aware that the CRC errors was a library bug and this had been fixed in a fork to the original Jeelib library (which is now about 2 years old). So, after installing the forked library I was indeed able to confirm that the issues had been solved.

I left two units, one transmitting and one receiving, running overnight sending gibberish test data and this morning it was still running without one single lost packet of data.

A lot of time was wasted with that. Anyway, now it is fixed I am able to get down to the nitty gritty of connecting up a sensor and transmitting real live data over the network.

I've been playing around with  RFM69CW radio tx/rx modules lately for an environmental sensor node monitoring project. Using the Jeelibs RFM12 library I was trying to get the Arduino Yun to talk to and configure the device over SPI to no avail. No amount of messing around solved the problem. I finally tried swapping out the Yun for an Uno and voila! It worked!

After doing a little bit of research it turns out that the Yun does not have its SPI pins broken out to the digital I/O pins as they are on the Uno and instead are only connected to the ICSP connector. So no wonder it wasn't working. After hooking up two modules to two Uno's they were able to communicate to each other successfully.

I have no idea why the ISP pins are not broken out on the Yun like they are on the previous Arduinos but i'm sure there is a valid reason for it.

So, now I have two device communicating the next stage in this project is to hook up some sensors to one of the device and have it sending the data back to the base station device. After that, the circuit will be powered by battery and using sleep modes will sleep in-between sending data to conserve energy.