Not sure what NVIS is? Please have a "google" to find out. There is a lot of excellent documentation available.
The only bit of info about NVIS I would like to point out in the post would be the fact that the 40m band is good during the average day and the 80m band is good for the average night. Running both in parallel could show some interesting daytime nighttime transitions.
Due to the harmonic nature of the 80m and 40m bands the easiest approach for a transmitter design would be to build an oscillator for either band and generate the other frequency by division or doubling. The downside here: shifts and offsets would also be divided or doubled.
A receiver also could make use of a single local oscillator. Here the most simple design would be a regular direct conversion mixer for 80m and a subharmonic direct conversion mixer for 40m. The respective audio frequencies could be fed into one single stereo sound card using left and right channels.
Frequency-wise, there are two obvious possibilities. Both have pros and cons:
- 3500400Hz & 7000800Hz
- 3579545Hz & 7159090Hz
The second option uses frequencies for which very inexpensive crystals are available, the big pro on the second option would be that is will enable many more hams to operate a transmitter legally (the ole novice story).
By now, you may have asked yourself why crystals still play a role here. Well, not so much for the transmitter, although they make nice filters for oscillators using digital gates. For a possible receiver those crystals would make ideal narrow front-end side-band filters, which are in particular important when operating in the middle of a busy band.
Want something more complicated?
What about a "superhet" design? With center (intermediate) frequency of 5.250800MHz and a 1.750MHz local oscillator the mixing products would be 3500800Hz and 7000800Hz. When shifting the intermediate frequency, both the 80m and the 40m frequencies will shift by the equal amount in the same direction, that's kinda cool!
Now to the tricky business how frequencies could be generated. Lets start with the easy one. 1.75MHz is subharmonic to 3.5, 7.0 and 14.0MHz. The first two call for trouble since those are too close to the final operating frequencies (*). But what about 14.0MHz? Crystals and even oscillators are available for this one! A division by 8 (ripple counter) will result in a very stable 1.75MHz local oscillator.
And here is the challenge: 5.250800MHz. There is a crystal for 5200kHz, but a 50kHz pull is too much and grinding is a tricky business. There may be a 10.5MHz crystal available, somewhere... As a last resort, a DDS would possibly do a superb job. This however would also be the most expensive solution.
(*) Problem for the TX, solution for the RX, subharmonic to 80 and 40 and the same time!
Want something even more complicated? No problem! That one is so overcomplicated, that is should rather be seen as experiment in thought. What about SDR? Take a 10m QRP crystal (28.060MHz). This frequency is perfect for a 40m SDR, center frequency: 7.015MHz. A quadrature local oscillator can be derive by a division by 2, resulting in a 14.030MHz local frequency and a 3.5075MHz SDR center frequency. For reception, 2 stereo channels are needed and to provide I and Q for both bands. TX in such a case could be done by either individual audio frequency generators w/ 90 phase shift networks or in a way similar to the LO, with a 56.9kHz generator.
As I said, the SDR is somewhat hypothetical, not practical in any way....
The superhet TX design presently appears to be favorable, together with a subharmonic direct conversion receiver for 80 (1.75x2) and 40 (1.75x4).