Lets have a look at the PSK-20's design. We are dealing with a 9.000MHz i.f., with a 9.002MHz b.f.o. obviously. The l.o. operates at 5.070MHz. Both oscillators use crystals, a pulled up 9MHz one and a pulled up 5.0688MHz one.
Looking what could be done for the LSB to USB modification, I figure the 9MHz xtal could either pulled down by means of inductors, "penned" down or it can be replaced with another one.
Replacement would be provided by two different options. Since 9MHz is a very popular i.f., side-band crystals are available for not too much money. The second option possibly already sits waiting in the junk-box somewhere. The CB channel 3 transmit (overtone) crystal has got a nominal frequency of 26.985MHz. On the fundamental that would result in a 8.995MHz beat frequency. Still somewhat too low.... but.... in the original design, the 9MHz crystal is pulled up, remember? The CB crystal would therefore fit perfectly! Lets just assume the b.f.o. would be at 8.9985MHz.
Penning the crystal in place down, I would do "in situ". That way, everything will be set when the desired frequency is reached and only the crystal's housing would have to be soldered in place again.
Now lets have a look at the local oscillator. The JT65 centre frequency is 14.0775MHz. With an i.f. of 9.000MHz, this would be reached by a l.o. at 5.0775MHz obviously. Equally you could add up the b.f.o. and l.o. frequencies to end up at the "dial frequency". 5.0775MHz looks a an ambitious 8.7kHz up-wards pull for a 5.0688MHz crystal, and most likely it will be in the existing discrete oscillator.
However, all is not lost. Seen that 2x5.0775 results in 10.155, there would be a couple of great options for (pulled!) digital gate oscillators and Flip-Flop dividers, namely CB synthesizer crystals.
- 10.160 / 2 = 5.080 for JT65a: 5kHz downwards pull on 10MHz
- 10.140 / 2 = 5.070 for PSK31: 2kHz upwards pull on 10MHz
Alternatively, the 10.160MHz xtal could be "penned" down to 10.155MHz; equally, a 10.150MHz xtal could be penned down to 10.142MHz to allow for the 20m PSK31 frequency.
In this way, I hope I can create a decent dual frequency narrow-band data transceiver.