Congratulations, Errol! Great news and great relief! I had a glass of wine, lacking the champagne, to celebrate while supping on the deck of a late summer's evening. This has been an epic, starting with your posting on July 29. You have demonstrated that you are no slouch when it comes to rolling up your sleeves and digging into the electronics, that you are doggedly determined and that you have good insight into what's going on.
If I have helped you find the problem, I'm pleased to have done so, but you've done all the work. My big fear was that I was misleading you and that all the poking and prodding of the patient was going to cause worse problems. Well, we did kill Q7 in the process; in hindsight, we should have first tried the less invasive test of the less likely possibility rather than cutting it out. Sorry about that - this has been a learning experience for me, too.
If anybody has any advice on how we should have tackled the diagnosis in a better way, I'd love to hear it - for the next time! And, Errol, your diagrams may be helpful to others. We're getting close to our storage limit on the website so send them to me and I'll see if I can fit them in.
As for further improvements, you indicate an interest in AM broadcast on MW and USB voice reception on SW. If you are using an external antenna for SW, a high pass filter with a cutoff frequency below 4MHz should provide reduction of hash interference above 4MHz due to front end overload by signals below. This is especially true if you are near strong MW stations. If you are trying to receive distant MW stations in the midst of strong local stations, I would suggest a tuned magnetic loop antenna and shielding of the DX-394 case - the shielding will reduce pickup on the internal antenna and the loop will allow you to steer its null onto an adjacent frequency interferor and provide additional front end selectivity to reduce overload.
If you are fussy about getting the SSB voice pitch just right, you could do my SSB Clarifier mod. Tightening the selectivity of the DX-394 for SSB is rather tricky - replacing the 6-7kHz wide filter CF2 with a proper 1.8-3kHz SSB speech filter requires modifying the BFO offsets from +/-3.5kHz from 455kHz to +/-1.5kHz or thereabouts and throws off the accuracy of the frequency display by the difference. Proper alignment of all oscillators is the one thing you could do that requires no mods - I wrote some procedures that used minimal equipment a year or so ago in the messages.
If you have the luxury of being able to use a computer for demodulation, you could do the Bill Beamon mod to tap the 2nd IF, feed it out through the Q29 buffer to the Tape Out jack and on to an outboard downconverter so that 455 kHz is shifted to somewhere in the 5-15kHz range, into the soundcard and run demodulator software in the computer. SDRadio, DReaM and IFDSP all do a remarkable job of SSB demodulation, incorporating DSP filters of selectable bandwidth.
You mentioned the one cap for the Crosstalk Mod - the other should bypass R64 - although Frank Cathell thought that the one you did may have been the more beneficial of the two.
So onwards and upwards!
73, Tom VE3MEO