This is my solar powered MEPT/QRSS transmitter.
I started with solar cells extracted from some cheap garden lights.
These were assembled into a panel built from Corflute sign material.
The corflute was obtain from a discarded sign after the state elections. Here is the back of the panel -
the origin of the sign seems appropriate (and purely by accident) !
The circuit uses an AVR CPU - specifically an ATtiny24. The CPU is clocked at the carrier frequency (10.14MHz).
The CLKOUT pin provides the carrier signal which is feed to logic gates which act as drivers.
The signal drives balanced outputs - inverted and non-inverted output gates.
This provides twice the voltage (four times the power) compared to a single signal operating against ground.
The CPU can adjust its own crystal oscillator frequency. LEDs are configured around the crystal to act as varicap diodes.
The PWM output from the CPU generates a DC voltage which is used to control the diodes and hence the oscillator frequency.
The QRSS band is only 100Hz wide (10,140,000 to 10,140,100 Hz). Temperature compensation is therefore necessary.
Compensation of the oscillator is achieved through software.
The CPU includes an ADC device and an internal temperature sensor.
The temperature is read, and the value used to correct the oscillator frequency.
The CPU is mounted upside-down under the hot-melt glue. It is in contact with the copper PCB to which the crystal is attached.
This provides a good thermal bond between the crystal and CPU, allowing the CPU to get a reasonable indication of the crystal temperature.
The whole CPU and crystal PCB is mounted separate to the driver PCB. The CPU is powered by a 3V regulator.
The coloured cable is the CPU programming interface.
The outputs of the driver are feed through a balanced low-pass filter to remove harmonics.
The output after the filter:
As you can see the filter does its job very well!
To calibrate the temperature to frequency, the transmitter was frozen into a block of ice!
It was then placed onto a heater pad. As it warmed up the frequency was measured as the transmitter reported its temperature reading and adjusted the oscillator control voltage.
This produced a large table of data from which a suitable calibration could be obtained to keep the transmitter close to the desired frequency over the operating temperature range.
The block of ice, and the calibration setup.
My GPS referenced frequency meter, and the PC data log.
The transmitter was wrapped in insulating material (Hollofill fibre) and tape.
It was mounted in the housing behind the solar panel, along with the batteries.
The antenna was fitted - two wires, forming a dipole.
Once tested, the housing was filled with insulation and closed up. Just a hole to access the power jumper and programming cable was left.
The sides were reinforced with fibreglass strips (pieces of PCB). Nylon line was threaded through the corflute to support the housing.
When installed (picture to come) the system will consist of a dipole with the transmitter supported in the centre.
A feed-line-less transmitter!
The software is written in C. It varies the transmitter frequency as required, taking the temperature reading into consideration.
The CPU has only 2kbytes of ROM, but this is enough!
The CPU sleeps as much as possible to keep the power usage at a minimum.
The CPU watchdog timer is used as the timer. This allows the current drain during carrier off to be very low, as the main CPU clock can be turned off.
'Carrier On' current = 88 mA (varies depending on antenna position). 'Carrier Off' current = 0.1mA.
As well as the calibration software, there is a Morse transmitter version and a Hellschreiber version.
It's up and running! The antenna is suspended between the house and the fence at an average height of about 3 metres.
As of 0900 9 July 2009 UTC it is transmitting.
Now to watch the grabbers!
This is what my transmissions should look like on a grabber:
The transmission shows my callsign, gridsquare, and the nominal temperature reading of the transmitter.
The message is sent twice and then followed by a 10 minute pause. Whether this power-budget works is yet to be determined!
For the receiver I will be relying on others with their QRSS grabbers. However I am looking at a grabber of my own.
I don't want to tie up a PC and receiver, and their associated power usage, but I think I have a solution for this.
This is a PPC405 based processor board. Once I get Linux running on it,
I should be able to port suitable applications to it and make a grabber operating with just 10-20W of power.
It should be cheap too. The board is only $US20 on ebay (plus another $20 to get it shipped!), plus another $AU35 for the peripherals.
I will build a suitable direct conversion receiver to interface to the sound card.
These PCBs are available on ebay. Search for "PowerPC 405". They start at $20, but don't seem to be selling very fast.
I think the supplier has quite a few available, so be patient and don't out-bid each other too much!
Email me if you want to know more, or help, with this low-power grabber.
Here is the initial installation, observed by some local wildlife.
My Initial test lasted only a few days before the battery went flat. The power budget needs some work!
The MEPT schematic is http://clayton.isnotcrazy.com/mept_v1/VK1TKA_SOLAR_MEPT_V1.pdf
I have added a connector to allow me to monitor the internel voltages and easily charge the battery.
opps - I think my 7 year old son is getting the Knack! Here is his MEPT - no electronics in it (yet!)
The installation has been tidied up - designed for easy installation and removal of the transmitter.
The software has been enhanced to not only send Hell, but to also send 'humps' and castle - these modes having a better chance of getting through.
These changes soon paid off! This image was captured from Bob VK7ZL's grabber. My 'humps' are clearly visible.
Later that afternoon conditions improved making my Hell signal clearly visible also. Not bad for 850km on about 200mW. (correction. It is in fact only 100mW)
Software modifications now have the MEPT duty cycle self adjusting based on the available power. However with the current installation (little direct sun) it is not transmitting very often.
Here is its full footprint (as seen by VK7ZL). It is sending:
Humps Hell (VK1TKA, gridsquare QF44, tx temperature 27C) Castle Humps
True power out (under load) is about 100mW. Frequency is about 10.140070MHz
I'm quite pleased with the results of my efforts. For only 2K of code space in this micro I still seem to be able to pack in enough functionality and make my own distinct shape on the bands!
The power source could be better, but that's something to work on in the next version!
The MEPT is currently offline. The power management software needs some improvements - it maintains the battery voltage okay, but it could transmit more often.
I am also considering a rebuild of the solar panel to increase the power output and hence the transmitter duty cycle.
There have however been a few notable events over the last few month:
Jiri Jetmat (from Zurcher) did some analysis of the LPF. This is the response he got:
His original files are here and here . It looks like the filter is doing just what it was designed to do!
In September my MEPT was seen by Peter, ZL2IK. Peter is at RF74ic - near Whangarei in New Zealand. That's 2315 km on 100mW!
Not much activity over the past few years on this project.
However I have recently had a request for the code, so it is now attached below (SolarMeptV1.zip).
Please read the readme file!