[time-nuts] Square to sine wave symmetrical conversion (part 2)

[Gerhard Hoffmann]

Am 28.07.2015 um 03:58 schrieb Bill Byrom N5BB:

[ nice and understandable summary of the Lee / Hajimiri ideas ]


OTOH with this Dirac-style resonator feeding one collects the
noise sidebands from order 1 to MAXINT and one needs to align
all these harmonics nicely.

It also does not help against low frequency effects that require
stability over many cycles, such as the 1/f region.

And it is only half the work. We not only need to feed the resonator,
we also want RF from it. Thus, to stay in this principle, the sustaining
amplifier would have to fetch its input also in Dirac style.

Now we are pretty close to an  ultra wideband  nonlinear loop gain.
Any low pass filtering would have to be done in front of the feeding
pulse former, or the harmonics will not align. And apart from
BW limiting we need something to set the net loop phase to 0.

The pulse output of the sustaining amplifier is also not nice to use
unless we want to feed a sampler or 1:2-FlipFlop.

Output filtering via the resonator, say, in the elegant Burgeon(?) style
would also be forbidden.

As I see it, L/H creates a lot of problems, and worse, it does not
provide a design algorithm, even if we accept the complications.


regards, Gerhard,  DK4XP


(Arghh, I'm writing this on a 4*2 thread machine at 3.8 GHz, and
Thunderbird produces typing delays...  Turbo-C on Z80 with a
Wyse-50 terminal felt better.)