[time-nuts] Square to sine wave symmetrical conversion

Charles Steinmetz csteinmetz at yandex.com
Tue Jul 28 11:26:58 EDT 2015


Jerry wrote:

>I don't understand why when there is already a clean crystal current in
>the circuit, someone wouldn't want to take advantage of that signal and
>use that rather than add filters and poor biasing to the mix to claim that
>something can be done when we are talking about ideal situations where
>the lowest phase noise is desired.

It was I who originally described the "high harmonic" oscillator, so 
I'll respond.  I didn't suggest that you'd *want* to do that.  You 
said you couldn't even imagine how one would build an oscillator with 
high harmonics, and I detailed one way.  And while it's not the most 
promising way to design a low-PN oscillator, it *is* a design that 
has been used in many commercially produced oscillators over the years.

And there are reasons for using at least some features of the design 
-- it is very robust over changes in component values and 
characteristics, so the designer can be confident that thousands can 
be made and they will all oscillate and have a predictable output 
amplitude.  This is not always the case with low-distortion designs 
that have very little excess gain.  Remember, while time-nuts and 
hams can select components and tweak every oscillator they make 
individually, manufacturers who are turning them out in quantity for 
sale at a reasonable price have more limited options.  Finally, it is 
not always convenient to tap the resonator current directly, and 
where it is possible the need to avoid loading the resonator may 
compromise noise performance.

So -- I was not suggesting that the architecture I described is the 
best way (or even a good way) to make a low-PN oscillator, but there 
are reasons why it has, nevertheless, been used by the designers of 
many commercial products.

Best regards,

Charles





More information about the time-nuts mailing list