When it comes to filtering the output of the ProgRock prior to multiplication, it's worth noting that the original crystal oscillator did not produce a nice sine wave: Remember that the crystal output is immediately used to produce harmonics in the following multiplier stages. In other words, for the purpose of harmonics alone, there is no reason why one would go through the trouble of doing so: For the receive crystal on the Mastr II, I *think* that you can simply program the ProgRock for the output frequency of the first multiplier, anyway, rather than crystal frequency, eliminating one such step - something that cannot be done on transmit due to the frequency-selective nature of the phase modulator.
The sole reason for some sort of filtering would be to attenuate energy that was far removed from the intended output frequency or its harmonics. While a low-pass filter would probably take care of 2x energy and higher, it will do nothing for energy that was much below this, near the desired output frequency or below it. Again, there may not be a need for such a filter in one's specific application - and if there is undesirable energy that is too close to the intended frequency (e.g. within a 100 kHz or closer) it may be challenging to construct suitably-tight filtering, anyway: In either case, even QRP-Labs' own band-pass filters might be too narrow for this - unless there is a redesign.
* * *
Somewhat OT, this reminds me of past discussions that I've seen (not in this group) where it was asked if one needs to low-pass a local oscillator prior to applying it to a mixer. The answer to this - assuming that the local oscillator is outputting ONLY integer-related harmonic energy - is simply "no, you don't." The reasons for this are:
- Doing the math, if one has appropriately used input bandpass filtering (on the "RF" port) and output filtering (on the "IF" port) you will find that harmonics won't contribute to spurious signals, anyway. (If one doesn't do the bandpass filtering, the use of a filtered local oscillator won't help you escape spurious signals, anyway...)
- Any mixer will produce harmonics during its switching, anyway - that's what it does.
- If you have ever used a switching mixer (e.g. QSD or "Tayloe") - such as that found in the QCX, many Elecraft radios, etc. it uses square waves, by definition.
A really good reference to this is Mini-Circuits Labs book "The care and feeding of diode ring mixers" (or something close to that) - an article from at least the 1980s - where they actually recommend square wave drive of mixers to lower insertion loss and improve power handling, citing the reasons above.
Again, if the local oscillator contains something *other* than integer harmonics, filtering is probably a good idea.