[time-nuts] OT: xtal osc PN

Magnus Danielson magnus at rubidium.dyndns.org
Sat Sep 18 14:52:42 UTC 2010

On 09/18/2010 04:12 PM, francesco messineo wrote:
>> One solution would use a stable standard oscillator, say 10 MHz, and
>> then use a bandpass filter to select suitable overtones for first
>> mixdown. You can select several options for selection of overtones, but
>> fixed LC-resonators comes to mind.
> This is a neat idea, but works only for overtones of the standard,
> some of my needed frequencies aren't overtone of 10 (or 5) MHz.

What frequencies you actually needed was not clear to me, so I assumed 
that 30, 40, 50 and 60 MHz would be useful frequencies, in which case it 
would be a simple and not very complex approach. It has been used in 
frequency counters for ages to mix-down the signal and then count the 
beat frequency.

>> Another variant is to use a fairly low-noise VCO and then PLL lock it
>> with wide bandwidth to a stable fixed reference (such as a 5 or 10 MHz
>> TCXO or OCXO of your choice, possibly divided down to suitable
>> step-frequency) as the PLL does some interesting things with phase
>> noise... within the PLL bandwidth the reference phase noise will
>> dominate where as outside of the PLL bandwidth the VCO phase noise will
>> dominate. This comes in handy, and for such PLL applications you want
>> the PLL to be wideband.
> this is also interesting, but again, isn't a PLL overkill for just 4
> fixed frequencies? I don't mind building separate oscillators. However
> the PLL approach could be interesting for other reasons (stability),
> any pointer? :-)

A simple PLL is not that complex these days. As long as you have fairly 
high comparator frequency after dividing down the VCO and reference you 
could get away fairly easilly. Standard programmable dividers in the TTL 
family and a single chip for phase-comparator will work fairly well.
There is gazillions of examples among hams for this approach.

>> A third alternative is to again let a stable reference of choice drive a
>> modern DDS chip, for instance AD9971 or so.
>> I am not a radio amateur, so I won't be able to say which is the best
>> solution for your needs, but that is at least what I would be looking at
>> if I where to build something like this.
> the best solution depends on many factors:
> -) if there were many others with my same needs, we'd probably find
> better to use a modern DDS (share the pcb making and someone who can
> do the soldering of such packages), instead, so far the low-VHF people
> aren't generally aware that better frontends are possible, you find
> lot of work in the HF receivers and next to nothing in the lower VHF
> where big signals and intermodulations come from broadcasts and not
> from other amateur stations!

Maybe the DDS board from the DMTD project would fit your needs?

> -) if much more than 4 different frequencies would be needed, then PLL
> would be the best choice anyway;

Actually, a DDS beats PLL on number of frequency any day of the week.

> -) if cost was not an issue, probably the best thing would be ordering
> 4 ready made OCXO from a respectable company :-)

Certainly, it would be easy, but you would not learn anything.

>> The link to Enrico I sent you is more the knowledge of the field, but if
>> you follow the links to Wenzel and Bruce stuff you have some designs to
>> look at. I wonder if you really need to go deep into the field to get
>> satisfied.
> Probably not that deep, I'm convinced that a well studied and known
> good xtal oscillator circuit could already do the job, I'm just not
> able to judge the circuit myself, so I must ask for other's advice
> (while I try to setup my own PN test bed).

John's and Bruce's pages alongside Wenzel should be the place to start then.


More information about the time-nuts mailing list