John said:
There's a neat chapter in Jim Williams' first "Analog Design"
collection (1991 version, chapter 26) about the ultimately-failed
design of a frequency synthesizer system, with lots of groveling
around with nanovolt VCO pertubations, shot noise, microphonics, fun
things like that. If you haven't seen it yet, I suspect you'd find it
amusing. It's sort of like coming at DDS from the wrong direction.
I remember reading it--that was the Dana Labs Digiphase. They really
did make it work, eventually, and sold a few but then the company went
down the tubes....
My situation wasn't quite that bad, thank heavens--I did get it to work
with a funky decimal rate-multiplier driven pulse swallowing counter.
The nasty rate-multiplier jitter is much less of a problem when it's in
the lowest decades of the counter that runs the pulse-swallowing line.
In this case, two cascaded CD4527 rate multipliers drove the carry input
of a 74LS190 presettable down counter, which drove the modulus control
of a Motorola 10/11 ECL prescaler whose number I forget.
Worked great, eventually--you could set the 14 GHz oscillator frequency
directly on BCD dip switches and everything. I sure wasn't going to
make that 1/f noise spec with an 8.333 kHz comparison frequency--the FM
noise would have multiplied up by 124 dB.
I know a few stories like that Dana Labs one, which is one of the
reasons I wanted to work for IBM Research--they pay me to go play in my
laboratory, and have done for 18 years now. (Please, nobody here wise
them up.) ;-)
Management commitment that will survive failure is rare and precious.
Cheers,
Phil Hobbs
