"BTW your digital AGC sounds neat.. "
Thanks :-) Yeah, it has two particular advantages from a user's point of view, one of which you spoke about below; noise: the only noise is from the analog chain to and from the actual treatment section, which would be there in an analog solution anyway. The fact that the processing is all digital also reduces the susceptibility to interference, p.s. noise, pcb layout, etc., but you know all that. The second is that it is fully programmable.
So yeah, digital AGC and a tight, programmable filter could go in the same box ... and now it seems I've committed to doing it :-)
As a bit of side-chat... I'm old enough to have no emotional bias toward digital vs analog; I'll use whatever gets the best results. In fact, the very project that brought me to the states in '82, and had me decide to stay here in '83 was a project to convert an analog machine tool control system to a microprocessor solution: a hundred or so 741's, various analog switches and multipliers replaced by a double stack of 8086/87 on Intel Multibus I. One of the options we considered seriously was to keep the math in analog form and only use digital processing for mode switching, signal routing, functional logic, and so on. We eventually went all digital, but it took assembly language programming to get the speed. The one thing analog does very well, though, is deal with continuous signals, and it isn't that long ago when we finally had access to sufficiently economical processing power that was able to do the job in discrete form for RF.
My second point is that I'm so late getting back into Amateur radio (I'd abandoned my UK VHF ticket in the mid 70's before I even really got started) and had I discovered that digital technology had been so enthusiastically and effectively welcomed into the hobby, I would have come back far sooner. I was absolutely blown away when I suddenly had the epiphanic realization that we have in our desktop computers more than sufficient processing power to do all the functions of an HF transceiver with barely a coil in sight.
Anyway, if somebody wants a programmable clock generator I can easily create one in the DAGLO box (the freestanding variant with have a two-line display like the QCX+).--