2008-08-16
30 Metre QRSS Beacon Updates
The MEPT Beacon has gone through some steady changes over the past few days. Most importantly it is now in somewhat of a "finished" state, stable and boxed-up ready for real transmitting experiments. A fourth piece of foam surrounds the circuit for insulation once the top is closed, the open ends of this "oven" don't seem to hurt. Modules are (clockwise from the microcontroller board); The ATtiny13V based controller, the fine tuning and FSK modulator, the carrier oscillator and buffer, the amplitude modulator (on wall), the driver amplifier, the low pass filter (on wall), and finally the class-E power amplifier.
Carrier Oscillator
Essentially unchanged from the original circuit, except I have stripped out the diode switched modulation trimmer and replaced it with the frequency modulation and tuning board. Conventional Colpitts oscillator using a 2N3904 buffered by a J310 with a low-pass filter delivering 900 uW into 50 Ohms.
That 3.1 uH inductor in the filter is implemented by 32 turns on a T37-6 toroid. The 78 pF caps are two 39 pF in parallel.
The reverse-isolation of this circuit is pretty good. It might be used as-is for a QRPp beacon or with the driver amplifier for 14 dBm output which is quite usable.
Frequency Modulator and Tuning
A pair of 1N4007s are used as varactor diodes. I picked 1N4007s not because they are particularly well suited, but simply that I had some on the bench. One of the varactors is directly coupled to the xtal circuit through 1 nF so most of its capacitance change is seen, the other is coupled through a small trimmer to adjust its maximum effect. This gives two independent channels, one for relatively coarse frequency tuning, the other for FSK modulation. Thanks to John for his suggestion around this part of the circuit.
The varactor modulators were characterised to design the bias network in the controller for best linearity and desired range of shift (etc).
Fine was measured with the coupling trimmer quite loose, and subsequently tightened on final assembly to give a good maximum modulation width.
The data looks characteristically parabolic as expected. I suspect the oddness of the fine is measurement error. The y coordinates are in Hz relative to 10.138000 MHz (x is reverse voltage) so it doesn't take much error in measurement to make the fine trace look lumpy.
Driver Amplifier
The driver amp is a simple 2N3904-based feedback amplifier delivering about 25 mW from the 800 uW output of the oscillator board. The amplifier input impedance is about 300 ohms when loaded with 50 at the output, this is a pretty serious mismatch for the previous stage's output filter (Return loss of about -2.7 dB). The gain from a 50 ohm source models at about 11 dB, but the circuit in practice measured about 15 dB transducer gain from using the oscillator as a signal source, probably because of the increased output of the oscillator with the lower loading offered by the input impedance (compared to a dummy load). DC input power is about 500 mW, so this stage is woefully inefficient, running at its limits, and should probably be redesigned. The supply line is keyed by the amplitude modulator to affect CW modulation.
Amplitude Modulator
The amplitude modulator is simply a BD140 PNP transistor with a 2N7000 switching its base voltage. A resistor network gives some envelope shaping to soften the keying. In theory PWM could be used for fine-grain power control from the microcontroller board, but I don't think I will attempt this.
Power Amplifier
The power amplifier is a single 2N7000 running Class-E. From its 25 mW drive it can deliver about 2 Watts output (33 dBm) through the output filter into 50 Ohms from a 13.8 Volt supply. It barely gets warm although it is biased very slightly on by the (selected for the device) resistor network at the gate to give good 50% duty cycle drive and sensitivity to the relatively small drive voltage. At the normal supply voltage (regulated 12.0 volts) it will deliver about 1.5 Watts - note also that optimal tuning changes noticeably with a change in supply voltage.
Impedance matching into the gate could be better but I suspect acheiving a good return loss would reduce the drive voltage amplitude and kill the output power, another active device would likely be required - in any case the impedance doesn't appear to upset the oscillator buffer LP filter even though the driver is partially "transparent" being a feedback amplifier. The mid-stage filter is largely redundant, but was useful for initial experiments, you might like to omit it.
I wrote a calculator to help design class-E amplifiers. The design was based around 1 Watt out from a 12 Volt supply and a loaded Q of 5. The result gives a load of about 75 Ohms, but empirical tests showed 50 Ohms was suitable loading giving a sightly worse efficiency and more voltage at the drain (without changing the shunt capacitor). The shunt capacitor value was decreased from the calculated value to match the output capacitance of the 2N7000 (about 18 pF), experimental measurements closely agreeing with theory. Even better efficiency is likely achievable, but the circuit works quite well as-is.
The drain choke is 12 turns on an FT50-43, its precise value is not critical, it just acts as a current source and could be (and perhaps should be) much smaller - the calculator does make a (very tiny - often ignorable) adjustment for its admittance. The series resonator is 35 turns on an T68-6. The larger core was chosen because of the large circulating currents, but numeric analysis was not carried out, T50 or perhaps even T37 might be OK? The trimmer is probably not the best for higher Q circuits, but at 5 it seems to work OK and doesn't get hot.
Low-Pass Filter
The filter is a Chebyshev with its final peak at the frequency of operation. It has pretty good measured characteristics, the centre capacitor was tweaked to tune out the effect of using E12 preferred values. Normal ceramic capacitors were used with no measurable ill effects.
The 3.1 uH inductors are 28 turns on T50-6 toroids.
The Controller
An ATtiny13V is the brains of the MEPT beacon. The microcontroller code hasn't changed since the original build. The board also contains the resistor network for deriving the modulator and tuning signals from the "width" and "centre" pots, in addition to the green indicating LEDs which show the CW and FSK keying state.
There are some elements omitted from these diagrams, for example the 5 and 8 volt three-terminal regulators, some filter capacitors, decoupling, supply filtering at entrance to the box, etc. All are extremely non-critical and conventional so I haven't detailed them here. Please post a comment if you want anything clarified.
Vertical Antenna
The antenna has also changed, I've built a matching network to feed my 3-metre vertical used in the 80 metre beacon on 30 metres. I hope to eventually build a diplexer and matching network that can feed this same chunk of metal with both beacon TXs concurrently - but much design work remains before I can attempt that.
The matching inductor is 7.8 uH. A polyvaricon fine-tunes the match. Oddly enough the return loss into this simple network alone is quite good. This is suggestive of rather large ground losses... A lot more antenna work needs to be done, while I get excellent signals into David VK6DI and Bob VK7KRW's sites I am yet to be seen anywhere else by the online grabber network.
Results
David VK6DI has been able to copy my beacon in all its various states of construction and stability. One experiment in particular used just the 25 mW output of the driver amplifier. Signal to noise measurements from this test indicate I should be just visible above his noise floor running only 25 uW!
With the beacon output now at an easily measured ~30 dBm it is a simple matter to attenuate down for the real QRPp(p) experiments.
Bob VK7KRW has been seeing my signal as well. He is my first report from VK7, including the 80 metre beacon experiments. Next time I fire up the 80 metre beacon Bob will listen out for it as well.
QRSS Reception
My attempts to receive QRSS have not been as successful as my TX work. My noise floor is *horrible*, combined with poor antennas this limits my chances. I did accidentally see VK6DI's 500 mW signal during measurement experiments on my beacon.
2 comments.
Attachments
| title | type | size |
|---|---|---|
| Oscillator Circuit Source | application/postscript | 14.584 kbytes |
| Frequency Modulator Circuit Source | application/postscript | 11.443 kbytes |
| Driver Amp Circuit Source | application/postscript | 12.719 kbytes |
| Amplitude Modulator Circuit Source | application/postscript | 11.658 kbytes |
| Power Amplifier Circuit Source | application/postscript | 12.412 kbytes |
| Low-Pass Filter Circuit Source | application/postscript | 10.173 kbytes |
| Controller Circuit Source | application/postscript | 12.063 kbytes |
Parent article: 30 Metre QRSS Beacon.