I’ve recently worked again on my MUTA project, and the first two modules are now almost done !
A microchip 18F27J53 (3.5€, 128kB of flash, 3.8kB of RAM, USB 2.0 without external crystal) is powering this unit, along with an MRF89XA8M module (7.5€, 868MHz, 1-10mW RF module, ‘free’ MiWi mesh networking stack).
Both are running with a voltage as low as 2.1V, and feature an ultra low power sleep mode, which is not particularly interesting for the operator unit, powered by the USB 5V power supply, but that comes handy for other units that are working on batteries !
The RF module is also operating in the good frequency range and within the maximum allowed power limits here in France (other modules of the same family are compatible with other countries frequencies).
Finally, I’m using an HID profile to have it communicate with my PC (no need to install any driver to do that, it uses the generic OS HID driver).
I’ve put a lot of work on the integration of the Miwi stack, and now it seems to be quite usable for my needs… 🙂
Miwi is an IEEE 802.15.4-like network stack, close for example, to Zigbee, but unlike this last stack, it’s provided free of charge by Microchip. More info here: Miwi info @Microchip or here (in french): Miwi info @blog.domogy.com.
The operator unit is my PAN (Personal Area Network) coordinator, the interface between the computer world and the mesh network of sensors and actuators…
When connected, it scans the frequency channels for noise, chooses a good one and starts a mesh network… But the very same unit (without USB connection) can be used as a coordinator, that will act as a communication relay between the end devices and the PAN coordinator, extending the physical range of the network.
I’m using the Miwi mesh mode of the stack. Thus, it is limited to 127 end device connections by coordinator, which is more than enough for my automation needs… 🙂
Furthermore, there’s a limitation of 4 hops in the mesh network, and no more than 2 hops away from the PAN coordinator. It means that a message could travel as far as from a PAN to a coordinator and to an end device. Considering that every device can emit/receive @10mW, it is not a limitation at all, for home automation.
This is my main sensor unit, and it features the same microcontoller and RF module as the operator unit, but adds a few features…
Firstly, it runs on two AAA rechargeable batteries (I’m using 1.2V 750mAh ENELOOP, they have a very low natural discharge in time). So there’s naturally an on-board voltage reference that allows the microcontroller to measure precisely the battery voltage.
A light sensor has been added too, and a precise temperature sensor (tmp75C, 0.25°C accuracy).
When the scout unit is waking up from sleep mode using its RTCC function, it will bring power to its sensors, read them values and then wake up the RF module to send those values to the PAN coordinator.
The scout unit could then fall again in deep sleep mode… And in this mode, it is using only 4.3uA! Yes, that’s 0.0000043 Amp ! With RTCC running… Wonderful! 🙂
I’ve even had to buy the excellent uCurrent from eevblog to be able to measure this current precisely! 🙂
Please note that all the unused I/O are exposed on the PCB, so one can power and add more sensors very easily on the unit.
Now that the hardware/PCB part is stabilized, I’m now working on having a clean and efficient software running on each module. I’m also testing the security feature of the Miwi stack, to encrypt the data sent/received between nodes.
Once everything is ok, I will share the PCB, schematics and code, and then switch my current home automation ZWave network to my very own custom MUTA network ! (^_^)