Build of a Micropower AM 'Pantry' Transmitter

[kalee20]

This is a little project which, on and off, occupied me for most of 2009. I wanted to use my AM radios more, for Radio 2 and Classic FM, which of course are not available on LW or MW. So I designed and developed a micropower 'pantry' transmitter. I designed with the following aims in mind:

* Good audio frequency response

* Low distortion up to high percentage modulation levels

* Good frequency stability

* Minimal radiated harmonics

* Long life, high reliability

* Vintage technology

* Auxiliary power output, to operate low-voltage signal source (FM radio, CD player, etc)

I really wanted to use valves, and shall do in a future version. But for now I succumbed to a transistorised circuit. For vintageness, germanium is obviously preferred, but I don't have any with reasonable voltage and current ratings which work above audio frequencies. So I settled for silicon. Keeping technology as vintage as possible, there are no IC's in the signal path, although I did cheat by using three IC voltage regulators.

Early experiments showed a modulated oscillator is not the way forward. So, I have a master oscillator, buffer, and (micro) power stage operating in Class E (resonant switching) which is where modulation stakes place. AF input sensitivity is sufficient for full modulation from 100mV or so, all AF amplification being Class A. I considered operation in the LW band but I could see a few complexities arising because the carrier frequency is little more than a decade above the highest audio frequency; SW would have been nice, but here I was worried about radiation being too effective and being detectable down the road. So I opted for MW.

I included an 'overmodulation' LED which operates at about 95% modulation, to ease the setting up and adjustment of input level. This works particularly well, giving a well-defined indication even on brief transient peaks of speech or music. Although distortion is pretty low, I added negative feedback (about 6db) at AF, mainly to straighten out the frequency response. Otherwise, sideband cutting in the RF tank circuits clobbers the treble (though it is still better than 'official' broadcasts).

The circuit is built on a piece of 0.1" plain matrix board, and housed in an Indian wooden carved box, which suits the job admirably. It is powered from the mains via a plug-top type power supply. It has been running almost continuously for several months now, with no failures, driving a few feet of aerial wire, and is one of the most useful projects I have ever made!

[Guitarist28]

Hi Kalee20,
What a good looking bit of kit. Did you think about an internal battery to keep the unit even more compact or was this not possible?

Regards

Rob

[kalee20]

Hi Rob,

I wanted something running all my waking hours (actually it's been running 24/7), and mains operation was dictated by this. Batteries were never a consideration!

[Boom]

That trace is really clean. Ps what's that on top of the box?

Dave

[ppppenguin]

Looks like a good design though I admit I know little about transmitter design. In particular you haven't sought to keep down the parts count. Transistors and passives cost very little so there's no need to prune the design unless it's intended for mass production.

I wonder if there's an easy route to providing a synthsised RF oscillator so the output frequency can be simply selected in 9kHz steps.

ISTR a MW TX by Dinosaur, the same guys that did an earlier successful 625>405 converter. Date possibly late 1980s, early 90s, certainly not available now. I have no idea what methods they used but I suspect that results would have been very good.

[kalee20]

Originally Posted by Boom

Ps what's that on top of the box?

Just an inlaid brass embellishment, part of the box. Unusual in a transmitter, sets it apart from others, don't you think?

Originally Posted by ppppenguin

In particular you haven't sought to keep down the parts count. Transistors and passives cost very little so there's no need to prune the design...

No, I didn't minimise the part count, although as each circuit segment was optimised and added bit by bit, by the end I had my fingers crossed I'd get it all on the board!

Originally Posted by ppppenguin

I wonder if there's an easy route to providing a synthsised RF oscillator so the output frequency can be simply selected in 9kHz steps

Probably, and I have a few ideas. However, using a frequency counter (with input waving in the air near my transmitting aerial), I've seen a variation of less than 80Hz over a month (taking readings at various times of the day to encompass temperature changes. Not up to broadcast stability, but good enough!

[dominicbeesley]

Hi peter,

Any chance of a peek at the circuit, always interested...

Dom

[Kat Manton]

Dom, see 'Attached Files' in post #1

[paulsherwin]

Very nice job.

Originally Posted by kalee20

No, I didn't minimise the part count, although as each circuit segment was optimised and added bit by bit, by the end I had my fingers crossed I'd get it all on the board!

It might be interesting to see how far you could go stripping elements out and still finish up with a good quality design. For example, the overmodulation indicator is a nice feature but adds complexity, as does all the power supply stabilisation.

Presumably all the BC547s could be replaced with any general purpose Si NPN transistor?

Paul

[G8HQP Dave]

I'm pleased to see you used a Class E PA. It modulates nicely and generates less harmonics than Class C, although high efficiency is not an issue here!

[kalee20]

Originally Posted by paulsherwin

Presumably all the BC547s could be replaced with any general purpose Si NPN transistor?

Yes - BC107/8/9, ZTX107/8/9 etc. And the PNP's similarly. The BC639 is the only slightly funny one, as needs to run to peak currents of a few hundred mA, plus withstand 60 volts, and switch fast, but ZTX451/2/3, or 651/2/3 will also be OK.

I standardised on TO92-encapsulated devices, makes for a circuit board more pleasing to the eye, that's all!

The voltage stabilisation to the oscillator is to help frequency stability (one of my design aims). The 15V main stabilised line is just so that I had a known voltage to run from - if I was adding a mains transformer plus rectifier I'd probably dump this, although at the cost of one capacitor and one IC it won't reduce the parts count by much! The auxiliary power output is only there for when I use a cheap portable CD player as signal source, of course, and make it independent of batteries without using multiple mains adaptors.

The parts count could be reduced more - C1 could go, so could R29. And the demodulator/negative feedback could all go (I ran for 5 months without it - but it does improve frequency response).

Detailed write-up will appear later this evening for anyone interested (Dom perhaps?). But it runs to a couple of A4 sides so it'll be as a pdf attachment!

[Brian R Pateman]

Very nice.

I'd certainly be interested in a write up. I've been developing ideas along similar lines for a while now - you've got further than I have!

[kalee20]

Here is a photo of the set-up all set up. This uses a (modified) REPLICA Bush DAC90a FM radio as signal source, powered from my auxiliary DC outlet (normally I keep them a few feet apart, hence the coiled-up cable. Maybe this isn't so much a 'pantry' transmitter as a 'sideboard' transmitter...

Originally Posted by paulsherwin

It might be interesting to see how far you could go stripping elements out and still finish up with a good quality design.

Apart from C1, R29 already referred to, R33, D8, D5, D6 could be dumped as they are intended to make it more bullet-proof. D3, C10 give a bit of negative bias to Q13 to assist a fast turn-off, and these could go also. But I wasn't interested in minimising component count at the expense of protection - as long as there was room on the board, in they went!

Originally Posted by Brian R Pateman

I'd certainly be interested in a write up.

See pdf file below, Brian. And do progress with yours!

[MichaelR]

Nice piece of work, Peter. I love the vintage trimmer caps.... you just could no resist could you.

The 100 Hz drift that you are getting, I know it is adequate but since you aimed to get the minimum of drift from the beginning , do you think you have achieved that or could it be bettered.

Mike

[JimMac53]

Beautiful, great job. Nice write-up. It's on the project list.

[murphyv310]

Hi Peter.
Perhaps you should do a production run of PCB's I certainly would buy two off you.
Cracking work!

[kalee20]

Originally Posted by murphyv310

Perhaps you should do a production run of PCB's I certainly would buy two off you.

Thanks! But other projects call (one of which DOES have a PCB layout as I'm making several of them)!

Originally Posted by Brian R Pateman

I've been developing ideas along similar lines for a while now - you've got further than I have!

Ah, but then you've been spending your time helping to keep this Forum on track, Brian. If this isn't a public service, I don't know what is!

Originally Posted by MichaelR

The 100 Hz drift that you are getting, I know it is adequate but since you aimed to get the minimum of drift from the beginning, do you think you have achieved that or could it be bettered

I've seen a total change of 100Hz over a few weeks, earlier this year. However, since then I've done 2 things: First, I've impregnated the oscillator coil in polyester stoving varnish; second, cleaned all the resin flux off the underside of the circuit board.

Both of these might help with stability, in that moisture absorption due to changing humidity levels can cause frequency to vary. An impregnated coil won't absorb moisture; and whether flux residues are hygroscopic or not I don't know. But hi-rel applications always require flux removal. Down side is it will take a few weeks before I can assess whether there's been any improvement!

If I wanted any serious improvement now, I would stick the thing in an oven and plot frequency against temperature, and use this to decide whether I can compensate for temperature changes by a suitable TC capacitor. However, for the mo, it's satisfactory!

[HamishBoxer]

Could i ask where most of the bits came from?

Thank You David

[dominicbeesley]

Thanks for the write up and the circuit (and sorry I didn't spot it been a bit distracted with family matters this week).

A very interesting circuit and as ever an excellent description of how it works! I'm going to have to (try to) analyse how you got your NFB working! I never got mine to work that well, demodulating the signal. When I tried as I was struggling to get stability with all the various poles needed to remove the RF from the AF etc...

Thanks!

Dom

[ukcol]

Originally Posted by HamishBoxer

Could i ask where most of the bits came from?

Thank You David

Hi Peter

This looks a very nice design. Most of the components look straight forward to source with the possible exception the coils; where could I get them or will I have to wind them myself?

Regards

Colin