An optimal operation of a double-balanced diode ring mixer can be improved if the IF port is loaded with a non-induction impedance of 50 ohms. This can be achieved with a so-called duplexer (fig»), but also with an amplifier with broadband input impedance of approximately 50 ohms.

I have tested a number of circuits see below:

The installation (see photo) is not difficult and you need only to cut a track and solder the components. If one are not satisfied with the result or you would for any other reason generate the original circuit, it is sufficient to remove everything and recover the track.

In principle there is nothing to be adjusted after the modification, but optimal results can be achieved by trimming L103 on 15 m for maximum during transmit or receive.

Because I have PC-100A, 100B PC, PC-100C and PC-100D boards, I could not resist my urge to experiment further with PC-100D in my Atlas215x. In the past I replaced the absolete "P8000 and P8002 with multiple J310 in parallel. The goal was to load the output of L102 as closely as possible with 50 Ohm and not introduce too much amplification in the existing circuit.

It is reasonably well managed with the adjacent circuit and L102's output is matched over a wide range with 50 Ohms and VSWR = 1.3. The system was measured with two different antenna analyzers.

After much searching I purchased 1N5711 diodes, so the original 1N4148's of the first mixer were replaced.

Are the Altas transceivers got better with the mods? I have not (yet) measure, but the sets are not become worse.

Later in 2014 during experimentation revealed that J310 FET's of the same batch were rather similar. If they were switched in parallel instead working with their own resting current system (e.g. diagram above), it had no appreciable effect in the circuit.

Initially a low VSWR at the input was obtained with one J310, but with both (fig ») in parallel, the SWR was even better!

The coupling with the following tuned circuit (L103) was not done with a toroid wide band transformer, but with a 10 pF capacitor at the top of L103. As a result post-mixer amplifier and tuned circuit are less loaded and a sharper resonance is obtained. The intersection point of the voltage divider C104/C105 becomes useless. The tuned circuit must be re-adjusted after the application of the modification.

The 1 nF capacitor (C104) is in fact redundant and could be short-circuit. For resonance core L103 should be turned out, but unfortunately it does not go far enough. As an improvement, the combination of C104/C105 could be replaced by a high quality capacitor (in this case, 91 pF).

Note that Atlas hardly update their schematics. Multiple versions of print designs exist and slugs of tuned circuit have different colours. Also associated tuning capacitance's differ. For your information I have drawn all details of my six PC100(A…D) at the schematic (fig»).