Carrying on the Practical Way
Tony Jones G7ETW presents several solutions to the need for a Morse practice oscillator.
Tony Jones G7ETW presents several solutions to the need for a Morse practice oscillator.
Simple, useful construction projects taking an evening or less to complete have always been popular in Practical Wireless, and one of our ‘hardy perennials’ is a CW practice oscillator. I expect most radio amateurs have made one at some time, but a request in the Letters page in the November 2019 issue gave me a good excuse to revisit this.
I offer you six options. I can take no credit for the circuits by the way; I found them all on the internet.
NE555 Astable Oscillator
This is by far the commonest circuit to be found when Googling ‘morse practice oscillator’. Fig. 1 shows a common circuit.
An NE555 oscillator requires few components and is easy to configure. It produces a loud tone, easily made variable in pitch and volume, but the output is a square wave and the NE555 cannot drive a low-impedance loudspeaker.
In this case the whole oscillator is switched but sometimes the oscillator runs constantly and the key switches in the loudspeaker. This makes no difference; this is a reliable circuit that works well.
(If it doesn’t work, check that the capacitor to ground from pin 5 has not been omitted.)
R2 and R3 (call this Rv) in series vary the charging current for C1, and the chargedischarge cycle controls the frequency. Frequency = 1.44 ÷ (C1 × (R1 + 2×Rv)) These values give a frequency range from 300Hz to 3kHz depending on where R3’s wiper is set.
Using an 80Ω loudspeaker would limit the current draw to something the NE555 can handle, but these are hard to come by. Hence the 47Ω resistor in series with the 8Ω loudspeaker. A pair of modern headphones wired in series would give a 64Ω load, but I think it would be uncomfortably loud.
The only real downside is the sound, which is a square wave and harsh to the ear. A square wave consists of an infinite number of sine waves, summed. Adding a low pass filter to suppress all frequencies above the fundamental would give a sinewave output.
Fig. 2 shows an RC filter with a cut off of 1kHz – this should give a nice sound at frequencies up to around 600Hz. Cut-off frequency = 1 ÷ (2 × π × R × C) For 1kHz, 3.3kΩ and 47nF are very convenient values.
Two-transistor Astable Oscillator
I first used this circuit to make an LED flasher for my wife’s choir’s Christmas concert, and I offer it as a lovely example of what astability really is. I had two DMMs monitoring voltages on the capacitors and transistor bases and this entertained me for hours!
Fig. 3 shows one of these oscillators, also called a relaxation oscillator, set up for audio frequencies. There are two outputs, opposite in state and constantly changing, controlled by an RC network. The output is a square wave.
In the femtoseconds after switch-on all bets are off, but a few milliseconds later let’s say TR1 is switched on and TR2 is off. Tr1’s collector will be nearly at ground, and Tr2’s will be at 9V. C1 starts charging via R2 and after a certain time 0.6V appears on the base of Tr2, switching that transistor on.
Tr1’s base voltage drops, and the transistor is switched off. C2 now starts to charge, and in time the other side of the circuit does the same thing.
This repeats. The loudspeaker experiences a pulsing waveform, at a fixed frequency.
For this circuit to work, R1 and R2 must be of equal value (call this R) and C1 and C2 must also be of equal value (call this C).
Frequency = 1 ÷ (1.38 × R × C) With the suggested values, that is about 800Hz.
To make this a variable oscillator, the
capacitor or resistor pairs need to be made variable. Variable capacitors are a fearsome price nowadays − use a twingang stereo potentiometer for R1 and R2.
Twin-Tee Oscillators
We have to start with a band-stop notch filter, Fig. 4. There is an RCR circuit and a CRC circuit, wired in parallel. The values are related.
The RCR circuit is R1 and R2 (the same value; call this R) and C3. This acts as a low-pass filter.
The CRC circuit is R3 and C1 and C2 (the same value; call this C). This acts a highpass filter.
C3 in the RCR circuit = 2 × C in the CRC circuit.
R3 in the CRC circuit = R ÷ 2 in the RCR circuit.
To properly analyse how a Twin
Tee notch filter works would require a determination of each section’s frequency response, then combining the results, and can’t that be done without some pretty gruesome maths. Fortunately, it all boils down to something quite simple: Frequency = 1 ÷ (2 × π × R × C) When a Twin-Tee filter is incorporated into an amplifier, the amplifier becomes an oscillator. This is because the filter is in the feedback circuit, and only one frequency (well, a narrow range of them) can get through.
Twin-Tee oscillators can obviously be implemented with discrete components, and only one transistor is needed, but see Fig. 5 for a Twin-Tee, on about 600Hz, using a quad-amp LM324.
Build a Function Generator
The AD9850 IC is a complete DDS Synthesiser that offers square and sine wave outputs (two of each) at frequencies varying from 1Hz to 60MHz. For a Morse practice oscillator this is total overkill, obviously, but AD9850 chips are cheap and experience gained on a simple project like this is directly transferable to RFbased oscillators as used in SDR designs. Google ‘Arduino AD9850’ to find some nice examples of a complete DDS with a nice display.
But for a simpler implementation, there is no need to buy a ‘proper’ function generator chip.
A quad op-amp, using all four amplifiers, can do the same job. See Fig. 6 for one based on the LM324.
Stage 1 is a comparator circuit in which the capacitor C1 charges and discharges alternately with a frequency given by 1 ÷ 2 RC. This outputs a square wave.
Stage 2 is an integrator circuit, which converts the square wave into a triangular wave of the same frequency.
Stages 3 and 4 remove the triangular wave’s sharp edges and produce a sinewave, still at the same frequency. Add a trusty LM386 amplifier via a DC blocking AC-coupling capacitor and a key and we get a Morse oscillator.
Other ideas
LM386 chips also oscillate – like all highgain amplifiers, sometimes when you don’t want them to! But this chip can be used as an oscillator intentionally. Fig. 7 is taken straight off the LM386 data sheet.
Again, the frequency is given by f = 1 ÷ (2 × π × R × C)
Arduino DDS
I don’t mean an Arduino driving a DDS shield, I mean why not get the Arduino do the whole job? True, it’s not the fastest computer in the world – it’s not even technically a computer, lacking an Operating system – but there is a pleasure in using these.
An Arduino has a 5V 10-bit D-to-A, and can execute loops. So, it should be possible to store some values for a sine wave (or indeed any other kind of wave) and realise these as analogue values using a timed loop. Like the NE555, an Arduino can’t drive an 8Ω loudspeaker, but an external amplifier is the easy part.
And the values that need to be stored – there’s no need to store a whole cycle of course.
Each quarter cycle has the same values, and all that changes is the direction of passing through them and which side of a nominal zero line we’re at. The code gets more complex, but I can almost see it in my head now…
This is well worth a try. Aside from the audio possibilities, people have made RF DDSs as well. That’s taken me a long way from an NE555 oscillator, but then my articles are apt to do that!