SSB transmitters using the phase shifting method
In the early days of SSB mode at amateur radio, attachments working with phase shifting technique were popular. They were simply connected to an existing CW transmitter. To this, high demands were then placed on frequency stability. In addition, its wave had to be as humming free as possible. Based on such an SSB adapter (Adapt-O-Citer by W6QLV) this SSB transmitter was designed. It can be powered by a crystal oscillator or an external VFO that likewise will connected to the crystal socket. It is recommended to let the VFO work on half the transmission frequency. Best results will be achieved at lower frequencies, e.g. in the 80m band. But in principle, the circuit is suitable for frequencies up to about 30 MHz. At higher frequencies, the simple crystal oscillator without buffer stage and stabilization can, however, there may be a slight frequency modulation, which leads to modulation distortions. With an externally connected VFO, this problem does not occur. In the 80m band, a PEP output of just under 10 watts can be achieved. Here, without rejusting the coils, a band segment of at most about 100 kHz can be used in each case.
As was found in my experiments with such a circuit, the suppression of the unwanted sideband was low. The achievable value of about 25 dB does not apply to all modulation frequencies because of the very simple held AF phase shifter. But the receiver-side intelligibility with a non-synchronized demodulator without sideband suppression (f.e. oscillating regenerative or direct-conversion receiver) was already much better than with a DSB or rather DSSC signal (Double Sideband Surpressed Carrier). Even smallest receiver frequency deviations here namely will lead to superpositions of upper and lower sideband and as a result to a very distorted reception. Also, the sideband suppression was already noticeable by the better energy balance: with good adjustment of RF and LF phase shifter, the power consumption of the PA was perceivable declined. Mathematically, a sideband suppression of 6dB would mean that the unwanted sideband would only be transmit with a quarter of the power.
For experiments with this circuit here the coil data for the 80m band:L1, L4, L5: 36 turns, coupling winding 7 turns
L2: 15 turns
L3: 18 turns with center tap, coupling winding 7 turns
Each copper wire 0.3 mm on shielded 8 mm bobbins with screw core
Based on my experiments with the Adapt-O-Citer, I replaced the PA and the microphone amplifier with transistorized circuits. In addition, I improved the audio phase shifter. Without further changes, a very usable sideband suppression could be achieved with the circuit. So gradually a complete simple SSB transistor transmitter was created.
As shown in the block diagram, the LF signal must be fed to the circuit with a phase difference of 90 ° at two points. The phase shift of the RF signal takes place in the same way as with the Adapt-O-Citer.
With the AF phase shifter, the signal is first split into two signals with a phase shift of 180 ° in a phase reversal stage. These feed a frequency-dependent phase shifter network, at the output of which in the speech frequency range two signals that are phase shifted by 90 ° are generated. These two signals are amplified in FET stages and then fed to the two DSB modulators. With the 2.5kΩ trimmer, the two signals are adjusted to the same level as possible. With this arrangement, a significantly better sideband suppression can be achieved, as with the simple phase network in the Adapt-O-Citer.
However, the resistors and capacitors built into the phase network must have an accuracy of at least 1%. If components with a correspondingly small tolerance are not available, they must be measured and selected accordingly. If necessary, suitable values can also be formed by combining two or more exemplars in parallel or in series. Switching from LSB to USB is very easy by swapping the 0 ° and 90 ° signals using a switch.
In order to be able to drive the arrangement with a dynamic microphone sufficiently, a microphone preamplifier is required. For a sufficiently narrow RF signal, it must be designed in such a way that only the voice frequency range is amplified (about 300 Hz to 2.7 kHz). The same is achieved with the somewhat outdated solution of connecting a carbon microphone under bias directly to the input of the phase shifter assembly. The better solution is then to use the later released telephone speech capsules with internal amplifier electronics, which are suitable as a replacement for carbon capsules and can therefore be connected in the same way. They do not generate the noises inherent in carbon microphones.
The circuit of the linear amplifier arised of the circuit proposal of a fox hunting transmitter. It was featured in Elektronisches Jahrbuch 1970 (Electronic Yearbook 1970, DMV-Verlag). What attracted me to the circuit was the unusual arrangement in which the base terminals of the transistors for RF were each connected to the output. Background were at that time hardly available RF power transistors. With this circuit, the transistors could be operated close to the cutoff frequency. In addition, the collectors were at zero potential so that the transistors could be mounted directly to the heatsink without isolation. When I built the circuit, were in Western Germany with the BD136 long ago cheap and much better suitable PNP transistors available. With several of this in parallel circuit, therefore, much greater power should can be achieved in such an arrangement. But I have not tested that yet.
I changed the circuit of the fox hunting transmitter by modifing the crystal oscillator into an RF amplifier and making the driver and power amp suitable for linear operation by inserting biases. To do this, I applied a bias voltage to the base terminals via RF chokes and inserted the emitter resistors (4.7 ohms for the driver, 0.47 ohms for the output stage). The now different DC potential of emitter and base had to be decoupled via capacitors. For the thermal stabilization of the final stage I used an arrangement known from the audio amplifier construction. The transistor used for this was to be mounted because of the thermal coupling with a clamp on the power amplifier heat sink. The resulting circuit worked very stably and gave an output power of about 5 watts at 80m and 40m depending on the dimensions of the resonant circuits.
The output transformer and later the driver circuit were wound on toroidal cores. The transfer from the preamp to the driver stage takes place via a capacitively loaded 10.7 MHz single-circuit filter (color code orange) from an FM radio. The parallel capacitor of 330 pF allowed it to be tuned to 3.65 MHz for resonance.
Phase transmitters can be directly operated at the final frequency, this means without fixed IF and without subsequent conversion to the operating frequency. In principle, also a variable-frequency operation with a VFO is possible. Well suited to this is the elsewhere shown Seiler oscillator with a following buffer stage. In the 3.6 MHz region it has excellent frequency stability.
The 80m band should be subdivided into segments of 100kHz or better still 50kHz for a sufficiently accurate RF phase difference and corresponding good sideband suppression. The adjustment elements are then matched separately for each area.