Basic Medium Wave Box

[Albert H]

The attached is a simple MW transmitter circuit. You can build it on Veroboard or "dead bug" - it's up to you. The crystal comes from Farnell for about 50p. The output transformer needs to be wound on a 1½" diameter Type 2 ferrite ring (T150-2). You'll have to calculate the turns ratio - you're probably going to want to match into 50Ω - so you use the usual formula for output impedance and then remember how to calculate the ratio..... The output filter coils need to show a reactance of 50Ω at the output frequency, and the capacitors are 50Ω at each end and 25Ω in the middle. This is a "half-wave" filter and will get rid of the harmonic crud.

The bias adjustment preset that adjusts the working point of the op-amp can be adjusted for minimum distortion when you do a trapezoid test. There is no audio processing - it is assumed that the incoming audio will be compressed, equalised and limited.

The output FET shown is a "logic level" type, designed for 5V on its gate. You can use conventional MOSFETs (I used to use the 2SK135) but you'll have to tweak the resistors on the gate a bit. You might find that you need some extra grunt to swamp the huge capacitance of the FET gate: I just use a 4049 hex buffer with all its inputs tied together. All its outputs are similarly together and give more than enough push to switch the biggest FETs!

[Analyser]

What a cruel post.

[Banus_radio]

Defo cruel.
Ill post an easy to build AM transmitter he said..................................................

[Albert H]

If you want really, really simple, you'll have to suffer drift from a VFO circuit or you'll have to risk ordering a crystal for your frequency of choice. A crystal cut to frequency will cost you plenty (£20+ last I heard) and will draw the attention of the authorities - especially since the crystal-cutting companies are usually run by radio hams. The synthesiser uses a standard, off the shelf crystal (that's meant to be used in modems).

Medium Wave in Europe goes in 9 kHz steps. The three chip synthesiser gives stable squarewaves at your frequency of choice:

The 4060 contains a crystal oscillator and a binary divider. If you connect the 4608kHz crystal, a resistor and a couple of capacitors across pins 10 and 11, you get 9kHz from pin 13. This is your reference and is fed to pin 14 of the 4046 and sets our step size.

The 4046 contains two parts - a voltage controlled oscillator that's tuned by a combination of the capacitor between pins 6 and 7 and the voltage on pin 9, and a phase comparator which gives a voltage proportional to the difference in frequency (or phase) between pins 14 and 3. This control voltage is filtered by the capacitors and resistors and fed to pin 9.

The output of the oscillator is fed to the 40103. This chip is a binary-programmed down-counter. You programme a number on to the switches attached to the chip, and as the oscillator pulses arrive, the counter counts downwards from the number programmed. When it reaches zero, it puts out a "zero-detected" pulse which resets the counter and feeds the phase comparator in the 4046. The "zero-detect" takes one clock pulse.

If you wanted 999kHz output, you'd program 110dec which is 01101110bin since you actually want a division ratio of 111. If you wanted 1602kHz, you'd program 177dec which is 10110001bin. It's easy enough to work out: the number that you have to programme - it's just N = (output frequency ÷ 9) - 1. The -1 allows for the reset time of the 40103.

The squarewave at your frequency of choice feeds out of pin 4 of the 4046 to the gate of a power FET. The source of the FET is grounded and the drain is connected through a transformer to a modulated power supply. The transformer ratio depends on the supply voltage you're using - remember the magic output impedance formula:

Zout = Vcc²/2Po where Z is the output impedance of the FET in ohms, Vcc is the nominal supply voltage (unmodulated) and Po is the power output.

A realistic aim for power output with this circuit would be about 5 Watts blank carrier, 20 Watts peak. This would cover a big area with a reasonable aerial during the daytime. If we've got a 30V supply and 5 Watts carrier, the output impedance will be 22.5Ω (remember the resting voltage is half the supply voltage). To convert that to 50Ω I'd use 8 turns on the primary and 12 turns on the secondary. At this power level, I'd use a T68-2 Amidon toroid, with 18swg enamelled copper wire.

You need to calculate the output coils and filter capacitors too - not difficult to do, using the reactance formulae:

Zcap = 1/(2 * pi * f *C) and Zind = 2 * pi * f * L

The modulator circuit is simple. The output voltage of an op-amp tends to follow the voltage on the non-inverting input. We establish a DC voltage on the non-inverting input with a potential divider. There's a preset provided to allow trimming of the quiescent voltage. The output of the op-amp is coupled to the base of a Darlington power transistor. The collector of this transistor is connected to the positive supply, and the emitter feeds the PA circuit and feedback is taken from the emitter of the transistor to the inverting input of the op-amp. When audio is applied to the non-inverting input, the emitter of the transistor will go up and down in voltage in sympathy, varying the supply voltage to the FET, thereby giving us Amplitude Modulation.

This is about as simple as you can get for a synthesised MW box.

[Albert H]

Farnell sell their part number 1666958 4608kHz crystal for under 20p!

[Albert H]

I forgot to mention - the output filter needs two coils with Z=50Ω at the frequency of operation, and the capacitors at each end of the filter need to be Z=50Ω at the output frequency as well. The capacitor in the middle needs to be Z=25Ω.

A friend of mine just built two of these - 6 Watts carrier, 25 Watts peak. The FET he used was an IRF640 and the mod transistor was a 2N3055 with a BFY51 to make up a Darlington Pair. He included a simple compressor and audio filter in each rig, since the audio sources are going to be MP3s on SD cards, provided by several DJs. A quick test last Sunday into an inverted - L "Marconi" aerial, using local water pipes for the earth gave a good daytime range of about 40km, as there's nothing at all on the frequency he chose since the French switched off their MW transmitters! Each one cost about £35 to build and the heatsinks and mains transformers were the most expensive parts.

[hailstorm]

Would love to have a go at building one.
Have also seen this though:

http://www.marktplaats.nl/a/telecommuni ... usPage=VIP

[Albert H]

That's pretty, but it's expensive. You see that it only covers part of the band as well. They've used a PIC or an Atmel programmable chip for the synthesiser and to drive the display. It looks like the output FET is bolted to one side of the case, and the modulator IC is hidden at the other side.

I'd be interested in seeing the circuit of that Dutch box. It's close to mine in principle, but I've used generic, off-the-shelf parts, to obviate the need to programme ICs. Mine's also much cheaper!

I did a synthesiser today, using a MC145106 and a 4.608MHz reference crystal. It's a single chip solution, and you just have to programme the control pins with the binary representation of your required divisor - the frequency you want divided by 9 (between 59 and 178 for medium wave). Unfortunately, the IC is difficult to obtain these days since it's been obsolete since the 90s, but you can still sometimes find them on Ebay. The other old favourite that I was going to experiment with was the PLL02, that was widely used in CB rigs.

I use the PLL IC to lock a varicap-tuned Colpitts VCO, which is using a 2N3819 FET for the oscillator, followed by a 2N3904 and a BFY51 as a two stage buffer. The oscillator signal sample is taken from the output of the buffer and fed to the PLL IC, and the control voltage is fed round to the varicaps. Setting the synthesiser is easy, because the PLL IC gives a good "locked" output - programme the divisor you need on the switches, and then slowly tune the VCO coil core until you find the centre of the lock range. The circuit gives good sine purity, so it's applicable for all sorts of transmitter designs, and could even be used as the local oscillator for a medium wave receiver.

I think that I'm also going to write a PIC PLL, which will use a 20MHz crystal and go in 1 kHz steps. This will allow the frequency to be directly programmed (and easily displayed if required), and will be able to be directly used up a good part of the short wave bands, and cover higher frequencies with larger step size using something like a 74HC4017 (for example) as a prescaler. If I use one of the common, cheap PICs, this will be cheaper (though more complicated due to the need to programme the IC) than using a dedicated PLL IC.

The cheap CMOS synthesiser still can't be beaten on price!