Morse key and Code Practice Oscillator
This project got started as a sort of "design challenge". I was asked to make a straight key, with as smooth action as possible and a practice beeper to go with it. The kind that doesn't disturb other people, so it had to use headphones only. And yes, purchasing and passing it on as one's own creation was not permitted.
I have worked with the straight key, but that was some time ago... back then I didn't think about it much. What should "as good as possible" be like? What makes a good key? And if there is an ultimate one, is it possible to fabricate it at home, without the use of lathes and mills? To find answers to these questions, I first explored the Internet. These things have been around forever, surely someone has written them up.
However, I found that most designs were either too basic or too complex, the latter lending it itself poorly to mechanically unequipped workshops like mine. Here's a good professional looking design but one next to impossible to copy without extensive machining. From this and others alike a set of criteria emerged:
1. Good solid base
2. Smooth action
3. Zero play (neither sideways nor up-and-down)
4. It has to adjustable for both travel and tension
After a bit of consideration, I implemented mine on a piece of hardwood (I think in previous life it was part of quality wine box), influenced by the fact that wood is easy to work with. I decided to leave the far end of the board a little longer than necessary to accommodate everything so that the board could be clamped to something. For the key body, I used an aluminum bar that pivots on a metal pin. To ensure silky-smooth movement and absolutely no play in any direction, I opted to put the pin (3mm hardened steel rod from an old CD-player, they all have one) on ball bearings. It may seem like overkill but this addresses points 2 and 3 100! Besides, most (but not all) computer fans (easy to obtain ODS item) have them, and the inner diameter is 3mm. The bearing blocks for both sides of the rod were cut out of an old Walmart kitchen cutting board and secured to the base from underneath with two small wood screws. I drilled 8mm diameter recesses into blocks which I then slightly tapped the bearings. If you wish to copy my design exactly, try to find a board that seems soft and slightly see-though. This type of board will not become brittle or "tire" and develop play. I used a hacksaw to cut out my pieces.
The rest of the workings are apparent from the picture. A piece of PCB board serves as a ground slab. To make sure it doesn't flex and alter the gap, it's screwed to the base in four places. To make the gap adjustable, I drilled two 2.5mm holes vertically through the key body, at equal distances from the pivoting point, which I then threaded with a 3mm tap. The screws are fixed in place with 3mm nuts. The reason behind going metric was that I have loads of 3mm metal screws, all taken out of foreign-manufactured computer equipment. Also, a good friend of mine, Kundar ES6KW, supplied me with a few quality 3mm metric taps. I guess metric is the sign of the times. The key knob came from Home Depot, probably some kind of replacement cabinet knob.
Tension is adjusted by means of a helical spring, hung from the back end of the bar. Finally, an island is cut out in copper foil and connected to the key with a piece of flexible, multi-stranded wire.
Now on to the noise maker. I knew, I myself didn't want to nor did I want others to listen to the harsh-sounding rectangular signal. There are myriads of these circuits on the net, most based on the 555 timer that generate it. I wanted to use pure sign wave only. After some thinking and a number of Spice simulations, the following circuit emerged.
It's simple, which is how I like it. U2 stabilizes oscillator voltage. Even though the schematic states LM7805CZ as the regulator IC, I actually used a LM2950CZ micropower LDO regulator here because of the 7.4V power source. If you plan to power yours from a 9V battery, then the cheaper LM7805CZ works, too. The left side of the circuit is a classical Wien Bridge audio oscillator, implemented on 1/2 of the general-purpose RC4558 op-amp. R5, R6 and C7 split the +5V rail in the middle, creating artificial AC ground. The frequency is controlled by the ganged R13, R14 pots. Trim pot R3 sets the op-amp gain which in this application has to be a little greater than 3. The other half of the IC works as a buffer shielding the oscillator from commutations and load variations. I used RC4558 in SIP-package, simply because I happened to have these. You can use any general purpose op-amp, say a LM358. The MOSFET is normally held open, shorting the signal to the ground. When the key is closed, the transistor closes and allows the audio to pass to the power amp's input. To drive the headphones I used LM386. I don't like them because they tend to be noisy and oscillate at RF frequencies, but I had quite a few in stock. On mine, I had to leave out the datasheet-specified 10ohm resistor in series with the C8, that was the only way to stop it from oscillating. Finally, the current draw is 9.5mA at idle, 10.5mA and 30mA at min and max volume settings respectively. I prefer to spare my ears and listen to it at ~13mA, that's about 1/5 of a turn up from minimum volume.
This is what the buffer output looks like. A tiny bit of distortion is visible but not too bad. Reduce the trim pot R3 slowly while observing the waveform on the scope. Find a spot with minimum of distortion. Unfortunately, on my prototype, I used a low quality ganged pots R13 and R14 that didn't track each other all that closely. This resulted in the oscillator not always starting reliably and I had to advance the trim pot back a little. But as I said, I can live with it.
It's interesting to note that in this project, simulation results almost 100% matched the actual circuit output.
Here's a dit, enjoy! Note the R9C10 nicely rounding off both the rising and falling slopes, all but eliminating the nasty key "click". Did you like it? Well, if you did, have another one, it's on the house :-).
Here is the semi-completed device, viewed from the right side. The front panel (a 65mm X 50mm piece of thin plywood) holds both the pots and the circuit board. I didn't bother making a PCB, this being such a simple project. All components are soldered on the breadboard, including the two phone jacks. They are held in place against the front panel by another 3mm screw, thus holding the board itself too. The battery is housed under the board. I left enough room to be able to use an ordinary 9V battery.
The battery shown here is a leftover from my "airplane flying". My good friend John has so graciously lent me a number of his batteries, not to mention other equipment. This particular specimen is one of his Lithium-Polymer two-cell 7.4V 1020mAH unit, made by Kokam. I didn't have a 9V battery at hand so it's a perfect fit for a project like this! I like Li-Po's and use them whenever I can.
Completed key and the beeper. Electronics are housed in a plywood box, painted with black Plasti-Kote and coated with polyurethane for esthetics. The white audio cable on the left goes to a PC with CWGet or similar running on it so that the sender can see how good or bad his or her wrist really is.
Final PCB artwork. I was asked to upload this as it was felt that without it, the project would be incomplete. The single-layer board measures 2000 x 2400 mil or 50 x 60 mm. To reproduce, download a pdf file here and print to scale.
Component assembly layout.
You can test what a short sample with this gadget sounds like (really slow, pardon my amusical wrist, I am by no means an expert. But hey, I AM getting better.)
Alternatively you can download the file here.
One of my readers sent me an article on how homebrewers built their keys in 1940-s, long before my time, as a kind of a reference. It's in Microsoft Word *.doc format and being that the content is so old, I don't think it's copyrighted anymore. You can dowload it here.
Happy practicing and don't forget to have fun, 73 de Brian.