[time-nuts] OT: xtal osc PN

[Magnus Danielson]

On 09/18/2010 09:48 AM, francesco messineo wrote:
> Hello all,
>
> sorry for the OT, but the electronic expertise of the group is too good :-)
>
> I'm looking for ideas and directions (articles and so on) to realize
> very good phase noise xtal oscillator, in the range 20-50 MHz for high
> performance frequency conversion. I would like to understand what
> circuits can be realized (not requiring too much professional and
> modern equipment, test eq. from the 70s-80s is ok) and what is the
> contribution of the active oscillator device, the xtal itself and the
> following buffers.
> Another idea that came on my mind was using digital oscillator (square
> wave, cmos) and then filtering for sine output, if this makes sense
> for a low PN point of view.
> Is there any way to measure the close-in PN of oscillators with an
> amateur setup?

First of all I think you need to quantify what you mean by "high 
performance frequency conversion" and what stability measures you are 
seeking as there are many degrees of excessiveness to attempt, and many 
of them may be well beyond what you need. Remember, we are time-nuts... :)

You can pick up a lot on the theory and measurement side if you look at 
Enrico Rubiolas material:
http://www.femto-st.fr/~rubiola/

I do recommend his book "Phase noise and frequency stability in 
oscillators" which is aimed in a theoretical way on oscillators, 
essentially driving in the point of what the Leeson theorem actually 
means. This establish the system parameter rules for oscillator design.

Rubiola has a bunch of measurement tricks in his sleeve and those is 
spread over various articles and presentations on the homepage.

Since Rubiolas focus isn't on actual oscillator-design but on theory and 
practical measurement setups on oscillators you will not be satisfied 
only with that.

Wenzel has a series of online articles which usually is a very good read:
http://www.wenzel.com/library1.htm

Further, fellow time-nut Bruce has collected a number of design ideas on 
his pages:
http://www.ko4bb.com/~bruce/

There is naturally more, but that should get you going in about the 
right direction.

Concept-wise, Leeson theorem explain how the amplifier noise in the 
oscillation loop wraps itself around the frequency corner shaped by the 
oscillators Q-value. Assuming that the Q-value is fixed, the phase noise 
is shaped by the amounts of white and flicker noise in the amplifier. 
Additional output buffer noise will then add ontop of that to form the 
complete phase-noise. So good oscillator performance is tied to 
achieving low white and flicker noise in amplifiers.

Additional contributions then comes from environmental and drift noise 
sources, so additional care is to be expected there if long-term 
stability is of interest.

For some applications isolational amplifiers is badly needed.

So, there is some initial pointers. There is much more, but should be of 
some use to you.

Cheers,
Magnus