Having had some experience of designing ultra low energy wireless network nodes, I was very excited to get hold of this Solar Energy Harvesting (SEH) development kit. Unfortunately my excitement was tempered by the poor documentation that comes with the kit. I wasted quite a lot of time trying to figure out why it didn’t seem to work much out of the box. I’ll try and spell out some of the issues here, so you can avoid some of the same problems.

Quick Start – not!

Firstly, take the Quick Start guide and put out for recycling! Then put the CD in your PC, open the User Guide and work from that instead. The first instruction is to “Remove the 'Battery Enable Jumper', J8” from the SEH module. Unfortunately the silk screen ident is so poorly laid out that it’s virtually impossible to identify J8. However, Figure 1-2 in the User Guide shows where to find J8.

The kit comes with two RF2500 target boards: one with a battery board and the SEH module; and one with the USB host module. If you want to get the system running out of the box, then don’t mix these up, because they are loaded with different firmware. The USB target board is programmed as a SimpliciTI access point, while the battery/SEH target board is an end point device.

I was surprised how big the solar panel is on the SHE module. It must be about five times bigger than the panels you get on typical solar powered calculators. Considering the battery capacity of the SEH module is only 100uAh, this seems way out of proportion. I guess they’ve done it like this so that (1) is works indoors, even in a dimly lit room, and (2) the battery charges in just a few minutes, given a reasonable light level.

The next step is to install the demo application software, either from the enclosed CD, or from the TI web site. Be sure to install and run the demo software before attaching the USB eZ430-RF2500 module, so that the drivers are in place. Now when you plug in the USB module, the central blob in the demo software, which represents the Access Point, should start blinking.

Although the User Guide encourages you to connect the other RF2500target board to the SEH module, at this point it’s best to try running the other RF2500 board connected to the alkaline battery board. That’s because you’re likely to run up against a few issues. Be sure to insert the jumper on the battery board to provide power to the RF2500 target board.

Range, wot range?

The first problem is range – there isn’t any, to speak of. My initial idea to test the SEH module was to seal it in a translucent waterproof box & leave it outside to transmit temperature data overnight. I soon realised this was an unrealistic goal when I could only get a reliable connection to the USB access point when the SEH module was in the same room! I always start to feel uneasy when wireless systems require the transmitter and receiver to be in the same room. Although it’s wireless, you’re not exactly untethered.

Battery capacity, wot capacity?

One thing this development kit taught me early on is that a paradigm shift is required when thinking about energy harvesting systems. I thought I knew something about ultra low power wireless networking, having designed transceivers that run for 10 years off a single AA size battery. But with a total battery capacity of 100uA, the SEH is a different proposition altogether. The battery may be fully charged one minute, but a few careless button presses later, accompanied by brief LED flashes, and you’ve got a flat battery! There is only one button on the RF2500 board, and what it does, apart from flashing the LED and flattening the battery, is to set the interval between radio transmissions. So it’s best to play around with this while you’re running off the (alkaline) battery pack. You have to have the demo software running in order to see which interval you've selected.

The radio transmission interval can be set to various values between five seconds and four minutes. When the SEH battery is fully charged, it has an estimated capacity of 400 radio packet transmissions. Setting the transmission interval to two minutes should allow the SEH module to transmit for over 13 hours without requiring charge from the solar cell. I figured this would be long enough for it to run overnight.

Demo Software

The demo software will look familiar to anyone who’s used the other eZ420-RF2500 development kits. A blob in the centre of the screen represents the SimpliciTI Access Point, and displays the time, temperature and voltage on the node. Once you get the SEH node to transmit, it will join the network, represented by a second blob on the screen. The blob will show yellow normally, or blue if it powered by the EnerChip battery. It’s only when running off the EnerChip battery that the demo software displays any interesting information. That is an estimate of the number of radio packet transmissions that the remaining battery capacity will support, and the terminal voltage of the EnerChip battery. So if you are trying the system out in a well lit room, you need to cover the solar panel, so that it doesn’t charge the battery and power the RF2500 board.

The demo software will also plot a graph showing temperature and, optionally, radio signal strength (RSSI). However, since it only ever seems to display the past one minute of data, this is pretty useless for data logging. The demo software will also log data to a text buffer. Since the Access Point logs every second by default, this swamps the log file. So if you want to log the data received from the SEH module, which after all is what we’re interested in, then you have to disable the Access Point data by selecting Edit|Toggle Access Point Data. Be aware that the data is only logged to a buffer. So if you want to be able to save your data for further analysis, be sure to select File|Save As… before you flush the buffer or close the demo software. Otherwise your data will be lost.

To test the system, I left the SEH module on a window sill overnight, and logged the radio packet data using the demo software. I then tried to import the data into Excel so I could plot some graphs. Unfortunately I found the format of the data makes this very difficult, and it took me a while to figure out how to do it. It’s the usual problem where the data has been made easy for humans to read, which also happens to make it difficult for dumb software to read. Two different punctuation marks are used as field separators that Excel doesn’t import by default, and Excel only allows one user specified delimiter per import. So I had to import the data once to translate one delimiter, export the data as CSV, and then import it a second time to translate the other delimiter. Then use some simple formulae to reconstruct the time and date fields. Here is a graph of the data I plotted.

It shows the SEH running off the solar panel until dusk, and then from the EnerChip battery overnight. Once the battery stops charging the available capacity for radio transmissions decays linearly (blue). This makes me think the software is just counting the outgoing packets, and not really calculating the battery capacity. The graph also shows the EnerChip battery voltage decaying slowly (green), but the hold-up is pretty good; only loosing 200mV overnight. At dawn the SEH tarts charging again. Within a few minutes the EnerChip battery is fully charged again.

Other Applications

Although this development kit comes configured as a solar energy harvester, the Cymbet EnerChip module can be re-used in any other energy harvesting applications, such as thermal or vibration energy harvesting. The SEH board even has terminals designated for a piezoelectric generator. I would have liked to have tried this, but unfortunately there is no support for any other power sources in the documentation. I dug out the datasheet and some application notes for the Cymbet CBC5300 module, but they just make some suggestions that other power sources could be used in principle. There are vague hints that it may be non-trivial to get the impedance matching right between the energy source and the energy harvesting module. I guess that Cymbet are worried people will give it a go & find they can’t get good results. However, I think it would be more helpful if Cymbet gave reference designs and recommended bill of materials for at least a few other power sources.