April 2014. After the successful conversion of the Icom 551 to 4m, I noticed that SSB mode - although working in a nice linear manner, seemed a bit reluctant to 'talk up', but it would reach full power on a good strong whistle. Upping the microphone gain made it much more responsive, but then it was too easy to run into distortion. This rig needed a processor.
I wanted it 'now', so I decided that it should be built using items from my dwindling stock of ancient and obsolete components.
My small stock of the famous Plessey SL6270 vogad device was long gone, and even the drawer marked 'RS1340 attenuator chip' contained only a piece of anti-static foam!
However, I still have a few NE602 (SA612) mixer chips kicking around, and I began to wonder if one of these could be persuaded to the job.
Here's the circuit of the finished processor.
The NE602 is a balanced modulator which can be used as an RF mixer, modulator and a gain-controlled amplifier. Pins 1&2 are inputs, pins 4&5 are outputs. I've applied the mic signal to pin 6, which would normally be fed from a local oscillator etc. In its quiescent state, pins 1&2 stand at 1.4 volts above 0v. In this condition, the mixer is balanced and virtually no audio from the mic emerges at pins 4 &5. If the voltage on pin 1 is raised or lowered away from that on pin 2, the mixer becomes progressively offset and audio starts to appear at the outputs.
In this application, a small offset is applied to pin 1 by the current flowing to 0v via R2 and R3 so pin 1 sits at a slightly lower voltage than pin 2, and the audio output is taken from pin 5 to the output level pot Vr1. The same output appears on pin 4 and is taken to the the base of Tr1 (any audio NPN device such as a BC547 will do) which amplifies the signal so it can drive the amplitude detector D1 and D2. At low microphone levels, the detector does nothing, but when enough audio arrives from the NE602, the detector develops a positive-going voltage which is also applied to pin 1 of the chip. This works against the standing offset and attempts to return the chip to its balanced state, reducing the audio output. From this point any further rise in audio level from the mic produces no rise in output from the chip and the overall level is therefore 'compressed'. We now have a much lower ratio between the loud bits and the quiet bits, which is ideal for SSB communication.
Have you spotted the flaw? It took me ages to work out what was happening when the circuit didn't behave quite as I imagined.....
The problem is that the mixer has a specific 'balance point', and moving pin 1 either up or down from this point will cause an increase in audio output. So if R2 and R3 make pin 1 sit at 1.3v, and the detector forces this to rise toward 1.4 volts (the balance point) all is well... until pin 1 goes through this point and the audio begins to rise again. At this point the compressor becomes an expander! As far as I can see, there is no easy answer to this other than limiting the gain of Tr1 so that it never drives the detector hard enough to cause this to happen. I did this by putting a 1k pot with adjustable bypass in the emitter. On a loud-as-possible whistle into the mic, Rv2 is adjusted so that this danger point is never approached. Despite this limitation, the compressor works well and has perked-up SSB performance considerably.
Another way of doing it may be to apply a suitable offset to pin 2, moving it further away from the flip-over region
Notes: The base bias resistor should be selected to settle the collector at about half supply.
The detector works best with germanium diodes diodes.
MAX supply voltage for the NE602 is 9 volts. I never go over 8v.