Saturday, December 12, 2009
Wake Up Boo! -The 50uA Super Regenerative Transceiver
For the last few posts I've been looking at low cost wireless links and how you can hack them at home.
You might want to look at the previous posts, Take Control, Radio Days and Radio Gaga.
I first got started about 10 years ago working with 433MHz short range wireless devices for reading gas meters.
50uA operational current from a 1.5V Cell
Right now I'm designing a low cost transceiver which will allow any microcontroller to communicate with its neighbours but using a minimum of current from the battery - allowing smart sensors to last autonomously on small batteries for years, or to make use of solar cells or other energy harvesting techniques.
Here's my prototype low power transceiver built on a small breadboard. It's based around a Telecontrolli RT4 transmitter module on the left, a 74HCT14 CMOS Schmitt inverter in the centre and about 20 other components. The FTDI serial cable providing output to the laptop is on the top right.
The aim of this project is to develop a very low cost and low power wireless transceiver which can be used for short range smart wireless sensor networks. The cost of the transceiver could be as low as US$ 1.00 with volume production.
Following on from some of last week's ideas, I decided to make a couple of changes to the receiver. I've also managed to get the power consumption right down to just 50uA and the ability to run on a single alkaline cell. I've also had a few thoughts about using this design as a wake up receiver - hence the title of this post.
I've also got the circuit of this receiver captured in Eagle CAD and I am working on a board design - approximately 25mm square. If you want to share the circuit, please leave me a comment.
Changes From Previous Design
Previously I had been controlling the RF transistor in the super-regenerative stage by pulling down on it's emitter. This required an NPN transistor to turn the RF stage on and off.
It then occurred to me that rather than going in through the emitter, as per the Bolling and McEwan patents, the obvious way to control the RF transistor was via the base connection. This was even simpler - a direct connection to the data input pin on the Tx module, which connects through a resistor and capacitor filter stage and then onto the base of the transistor.
Thus I could eliminate the extra NPN switching transistor and simplify the circuit considerably. The advantage of this change, means that the receiver can now be made from an unmodified Telecontrolli RT4 module - further simplifying the amateur construction.
Even Lower Power Operation
I then decided to see if the receiver worked at lower battery voltages. Previously I had being using 3V, but I wanted to see how the receiver performed at 2V. After a bit of fiddling I can confirm that the receiver will still receive and decode 1200 baud data from a test transmitter located 10m (and 2 brick walls and a floor) away at 2V drawing just 98uA from the battery. I had achieved my first personal goal - the sub-100uA receiver, and one that would work in conjunction with a microcontroller running on 1.8V.
Not content with 100uA, I decided to find out what the lowest battery voltage was that would allow the super-regenerative process to occur- to determine just how low can you take the collector voltage on the Telecontrolli module before it just stops oscillating. With a fairly exhausted AA Duracell, I found that the lower operating voltage for the RF stage is just 1.2V. Any lower that this and the oscillator just won't start up.
A typical AA alkaline cell such as Duracell is 2700mAh capacity. With the receiver budget down at 100uA, this would mean about 3 years operating life on a single cell.
The bulk of the operating current is consumed by the RF oscillator. With a 100 ohm resistor in the supply rail to the Tx module I found that nearly 45uA was used by the RF front end, and only 5uA in the rest of the receiver circuitry. 45uA at 1.2V defines the lower operating point of the Colpitts oscillator, and this would be the fundamental limit to how low power this design could be made. Below 50uA, the receiver is getting to the point where it is no longer a useful receiver.
However, by lowering the baudrate of the signal, which means more energy per bit, it would mean that the receiver could be operated at a reduced voltage as a wake up receiver, waking the microcontroller, which then applies some more volts to the power rail, allowing it to become a fully functioning 1200 baud receiver.
A 300 baud signal would be suitable to detect in the wake up receiver. The filter capacitors needed for 1200 baud operation and 300 baud operation are likely to vary in value by a factor of 4. The correct capacitor needed for wake-up and normal operation could be selected using a pin on the microcontroller. I've now got the receiver running at 48uA and successfully receiving and decoding 300 baud packets.
Scope for Improvement
The scope trace shows the output from the second transistor running on a 1.3V cell at 300baud. You can clearly see the 1kHz nominal quench frequency superimposed on the 300 baud data. A better filter stage could be used to remove some of this making the signal easier to slice. With a bit of care and a suitable schmidt inverter it is possible to slice this data at the correct level and retrieve clean, square edged data. This raw signal is approximately 400mV peak to peak.
In each of these experiments I used my standard test transmitter which consists of another Telecontrolli RT4 transmitter module powered from a 4.5V battery and data pin driven by a PIC microcontroller. In order to make the tests representative of the real world, the test transmitter is located in the front room of the house and I work at the back - a signal path of 10m plus two brick walls.
Another thing I noticed whilst working at low supply voltages is that the relaxation oscillator slows down to about 1kHz, about 100uS on and 900uS off.
I tried sending a 1024Hz squarewave modulated RF signal from another test transmitter, and found that the output of the receiver was a squarewave signal equal to the beat frequency, between the relaxation oscillator and the 1024Hz test signal. This might prove a suitable means of causing a wake up. By choosing a wakeup signal that produces a strong beat frequency the resultant could be rectified, detected and used to wake up the micro.
Please leave a comment if you want to share any of this design or have any suggestions. I'd like to start a competition to see who can come up with the lowers power super-regenerative receiver. McEwan suggests that 1uA is possible - but I'm yet to be convinced.