June 23, 2010:
Revised: v2.0


Two cheap CMOS ICs and a few transistors was all that was required to rebuild my elderly TTL automatic morse keyer into a low current lightweight battery powered iambic morse keyer.
No microprocessors were harmed in the construction of this keyer!


Sometime over the past year or two, my elderly Accu-Keyer gave up the ghost and died. I built it more than 30 years ago (!!) from the original TTL IC-based design published in the ARRL Handbook. It was an accurate copy of the original, right down to the printed circuit board which I made using nail polish as resist and ferric chloride etchant. As I was tearing the old keyer apart, I instantly recognised the PCB production method I'd used by the quality of the resulting PCB. It was not very pretty, and I certainly could not reuse the PCB simply by replacing the dead ICs with new CMOS devices.

In those days, the Accu-Keyer was the most popular design. Most of those keyers used TTL chips, and the current drain worked out to be a few hundred milliamps. Running it off batteries was hardly practical, and I never tried. But there were times when it would have been useful.

In fact, I was never very happy with the keyer. The power supply was my biggest gripe, and my own fault. I made it using a rewound transformer. I was a poor high school student in those days, without much money to spend on the hobby. To get the right voltage for the power supply - I needed 5VDC - I found a transformer somewhere, pulled it apart, and rewound the secondary by hand. The problem was that the darned transformer used to hum like crazy because I was unable to clamp up the laminations adequately. I guess I could have dipped it in varnish or something, but I never got around to it.

The hum wasn't too bad when it was all enclosed inside the little box. Even so, every time I turned it on, I would hear that hum, and I'd mutter to myself about doing something about the transformer 'one of these days'. But while the keyer was working and usable, I just couldn't be bothered. There were too many other things of more interest to build. A few months ago, however, I went to use the keyer again (You can tell how much CW I do!) only to find the keyer had died. I checked out a number of the chips, fixing my old TTL logic probe in the process (The first dry joint in 20+ years), only to find that almost all of the seven chips used in the old design had up and died. I was prepared to replace one or two, but the widespread deaths of ICs across the PCB made me forget that idea.

It was time to rebuild my keyer.

What is a Keyer?

Keyers, or "Elbugs" as the Europeans call them, are used to send Morse code. Most keyers use a two-paddle key for operator inputs. Pressing either paddle towards the opposite paddle results in the keyer sending a series of either 'dits' or 'dahs', depending on which paddle is pressed. Unlike typical manual morse keys, keyers are designed to send near-perfect code, with 'dahs' equal in length to three 'dits', and with precise inter-symbol timing.

The really neat thing about the more advanced versions of these keyers, called 'iambic' keyers, is that these allow you to insert a 'dit' in a stream of 'dahs', or vice-versa, or send a string of 'dit-dah' or 'dah-dit' characters. To send an 'a' ('dit-dah') merely requires a single squeeze of the paddles. Sending a 'c' ('dah-dit-dah-dit') take no more effort - A single, slightly longer, squeeze does the trick.  The objective of these keyers is to reduce the effort of sending Morse code manually. While computerized keyboard keyers also exist, they are not as portable as these keyers, and certainly nowhere near as popular.

Modern keyers are almost all microprocessor-based, and include memories to hold standard messages. These provide storage for 'CQ' calling messages, and routine sentences which regularly occur during a call such as the operator's name, location and equipment details. Most also add specialized auto-numbering messages for use in contests. Some older TTL or CMOS designs also supported memories, but these designs were so complex, with dozens of chips, that you'd have to have a lot of time, and parts, on your hands to want to build one of those.

Choosing a Design

I had a quick look around my books and magazines to see what I could come up with by way of keyer designs. A common choice for much of the past thirty years was the Curtis range of keyer chips. These came from a US company of the same name. Those chips are obsolete, no longer available, and so those designs were definitely out. (Of course, nothing ever really dies. It turns out that the descendent of those original Curtis chips is still available, the 8045, from MFJ.)

The rear view of the keyer shows the old AC power cord entry, now empty, and the AC fuse, now unused, and the connectors for keyer output (to the transmitter) and a standard phono connector for use with a manual morse key
. (Right click with your mouse to get a closer view)

I could have rebuilt the keyer substituting TTL with CMOS devices, using the same basic design, but I was never really happy with the Accu-Keyer. It used a two-transistor dot clock which was switched on and off by the keyer logic, and that led to slightly shortened leading dits on occasions. My original keyer never exhibited that defect, but others did. The more important problem was the dot oscillator itself. It had the reputation of being hard to get going, a fact I recalled on all three units that were built by a group of us back then. We ended up scouting around the city to purchase exactly the right type of transistor specified in the original article to get that oscillator going properly. Very annoying.

Others attempting to build the Accu-Keyer in CMOS reported similar problems in that area. And the idea of finding seven chips of the right type in my junkbox, (Yes, that was the number of ICs used in the Accu-Keyer) was hardly likely. No. I wanted simplicity, and I wanted to build the keyer right now.

I was briefly tempted to design a keyer using a microprocessor. Other pages on this website demonstrate that I'm no stranger to these devices, but I really was after something with a touch of the original keyer. Something that was easy to build in an afternoon so I could get back to other tasks. I just wasn't keen on spending a couple of days writing software for a keyer I only used from time to time.

Naturally, I could have purchased one of those little ready-coded keyer microprocessors which are available on the web. (For example, the K8 and K12 PIC-based chips from K1EL, VK1OD's PIK keyer chip, or the TiCK1 and TiCK4 chips from Kanga) But I wanted a quick fix, and I did not want to have to wait a few weeks while an order went out and the chip was delivered by snail-mail.

 It was only after I had built the keyer described here and used it for a few weeks that, of course, I found a Czech website with an 8051-based memory keyer complete with software which I could have copied in a heartbeat. (Look for the latest version of the Elbug ) Oh well. If I'd found that site a bit sooner, I guess you wouldn't be reading this page now!

Anyway, what I did was spend an enjoyable hour or two thumbing through my old files and magazines looking for a simple CMOS design. I found a really simple one in back copies of SPRAT, the UK-based QRP club newsletters. It featured just two CMOS chips, and, yes, I had those in my parts box. 

The design choice was made.

It turns out that this keyer design has a bit of a history. The first reference I can find to it is a TTL-based design from '73' magazine in the mid-60s. It seems this design was subsequently converted to CMOS by Pierre, FE1MOG. His design was published in SPRAT, the UK QRP magazine, back around 1971. This same circuit was republished in the recent RSGB book 'Low Power Scrapbook' (page 195), along with all of the mistakes of the original circuit(!) Coincidentally, and about the same time as I came across the information on the web about the German 8051-based keyer, I spotted a nearly identical keyer on the
website of Onno, PA2OHH. Onno's version used the /Q output of IC2a (Pin 2) while the original circuit used the Q (pin 1) output of IC2a, as mine does. The only impact of this is to reverse the dit and dah input connections. Everything else functions identically to the original circuit, at least as far as I can tell.

All of which goes to show that there is very little in this life that is truly new and original.

The DesignExample image - aligned to the right

The circuit diagram of the keyer is shown in the box to the right. Just "right click" on it to see it at full scale.

U1a and U1c form the dot clock while Q1 and Q2 invert the input signals from the paddles. The dit and dot paddles connect to the common ground (pin 2 on the input connector) to send the respective symbols. Q6 and Q7 form a simple audio oscillator with the tone frequency set by RV2. I actually used a fixed 33k resistor here. This oscillator is turned on by Q4 and Q5. SW3 allows the user to turn the sidetone off.

By the way, this was the first schematic I drew using Basic Schematic, a freeware program from Japan. I previously used CorelDraw exclusively. Since it's so easy to use, and the results are more than acceptable, I plan to use it for more circuits in future. I have to revise a few of the original symbols in the symbol library to get them closer to standard schematic symbols, but aside from a few minor issues, I'm very happy with it. Automatic placement of text around the component leaves a little to be desired too, but hey, it's free, right?

Solving a Few Problems

Nothing is simple in life.

It turned out that the original design I had located was shy of a few critical details. Some, of course, were obvious. The original circuit diagram in SPRAT (and in the latest RSGB-published 'Low Power Scrapbook') had omitted a few critical items, like ground connections on several pins on each of the ICs. Those were obvious, and fixed in a jiffy while building the unit.

Then I realised that the original design had another minor problem - The keyer inputs (for the 'dit' and the 'dah' paddles) connected the relevant 'dit' or 'dah' paddle inputs to the supply voltage rail in the keyer to send each symbol. The normal arrangement is to connect the relevant paddle input to ground whenever the appropriate paddle is pressed. The original approach left a potential risk of a short circuit if I was a bit careless with my metal-shrouded keyer paddles. (I use a cheap and cheerful 'Galbraith' paddle, by the way, bought many years ago from a local NZART branch. Sadly, they are no longer available)

Solving that problem was easy. I added two transistors to invert each input line, and these can be seen on the circuit diagram.

Life, of course, continued to pose little problems. Next up: The keyer had no sidetone oscillator. Most transceivers have built-in sidetone oscillators which produce a tone simultaneously with the keyer output signal. But I like to practice my sending a bit from time to time without turning on the transceiver and letting the world hear my lousy sending. So I needed to add a sidetone oscillator to the keyer.

Well, really, that was easily fixed too. Another couple of transistors in my favourite audio oscillator arrangement worked a treat. It drives a piezo speaker and draws only 1.5mA at full volume. You can find these little speakers inside many toys or inside musical greeting cards. I seem to recover two or three of these each year from broken items heading for the rubbish bin, and they are just great for this sort of project.

Building the Keyer

I was in a hurry, and it probably shows in the photographs.

I used a scrap of printed circuit board - It wasn't even cut square - and built it all in an afternoon. I built it 'Mahattan-style', which is to say, the chips were soldered upside down on the scrap of unetched circuit board, and other components soldered in around them, using the legs of the ICs like tagstrips of old. It actually looks a bit rough, and untidy, but it's surprisingly robust. Saves making a PCB too.

I used a pair of AA sized batteries to run it since the CMOS devices hardly draw any current. If I send continuously at top speed, the keyer draws 30mA. I know that's not ultra-low current, but it's about 10% of the current drawn by the old TTL design, so I think it's a low current keyer. Besides, sending at normal speed (for me), at around 12 to 15 WPM (words per minute), the current drain averages about 10mA with the sidetone turned on.

I glued the battery holder to the side of the box with hot glue. To tidy it all up, I loomed the wiring using a few nylon cable ties. I had some blue ones in my parts bin which added to the colour just a bit.

Wiring goes to the original connectors. There are a few holes left unused in the box. That box was originally made from some thin aluminium I recycled from old commercial VHF transceiver front panels. One unused hole that can be seen is the original mains cable entry. I left the rubber grommet on that hole just in case I ever want to build an AC power supply into the keyer. I suspect it will stay unfilled this way for at least another 30 years.

There is also a phono socket to connect a regular morse key. I didn't wire this back in because when I use a straight key, I plug it directly into the transceiver. Maybe I'll wire that connector in one day. It will just connect directly across the output socket, with a diode added to key the sidetone oscillator.

The switches on the front panel (from right to left) allow you to turn the power on, turn the sidetone on or off, and the last switch is wired directly across the output jack. This allows me to continuously key a transmitter for a few seconds. This is useful for tuning my older transmitters and antenna tuners.

The knob on the top left hand side of the front panel sets the sending speed of the keyer. It ranges from about 5 WPM to more than 50 WPM, as far as I can tell. It is much much faster than I can manage. If that range is too wide for you, fiddle around with the 470k fixed resistor, the 3M linear (not LOG!) variable resistor and the 47nF timing capacitor. If I was making it again, I'd use a 100nF capacitor, a 1M or 2M variable resistor, and a 470k fixed resistor.

The knob on the right was used to control the volume of the sidetone oscillator in the original unit. The piezo speaker didn't need a volume control. By the way, you can see the piezo speaker in the photo.
It's the large black object on the lower right hand edge of the PCB. It was recovered from an old cordless phone. The volume control currently just fills a hole in the front panel, and I left it there. But it does nothing.

While I run this keyer from 3V, it will work equally well from just above 2V to 12V. The piezo will get a bit loud at the higher voltages, so you may want to adjust the audio level. Try adding a 1k to 10k resistor in series with the piezo. Maybe that volume pot will come in handy one day after all!

Want to go back to the main page? Click here to return directly.