Sorry I wasn't clear - the Silabs parts seem to use a very different
mechanism than the devices I was warning about.

The Silabs parts use a MEMS resonator which is tuned somehow (probably by
adjusting the amount of electrostatically-generated tension on the silicon
structural beams which are part of the resonator). The tuning is done at
the factory and then the part contains temperature compensation and has
inherently low ageing. Since it is just an oscillator and not a
synthesizer, the jitter is low.

The least expensive of the programmable oscillators are PLL-based and can
have very significant jitter problems - which matter only for certain
applications. Even if their RMS jitter is not too bad, it may be
concentrated in very narrow frequency bands and still present a problem for
RF applications, for example.

Many of the newer PLL-based programmable oscillators use various fancy
techniques to reduce jitter, such as intentionally dithering the jitter so
that it spreads out more over frequency, or they cancel out the one or two
highest jitter spectral peaks.

The AD9850 which was mentioned is not a programmable oscillator but a DDS
IC. DDS also generates an output with deterministic jitter and has its own
set of problems. Many DDS devices also implement fancy corrections to
cancel out some of the jitter.

I am not casting aspersions on any of these technologies, just warning that
you need to pay attention to jitter if it matters to you. For typical
digital applications, it doesn't. For high-speed busses and telecoms
applications, it often does matter, but not in the same way that it matters
to RF applications.

I have built radio receivers using both cheap PLL-based programmable
oscillators and DDS chips and experienced these problems. They showed up as
"birdies" or spurious responses in the receiver. With the DDS chip I was
able to get around this by changing the DDS chip settings to shift the
spurious outputs for different frequency ranges of the receiver so that
they almost always fell outside of the present frequency band. But, the key
thing is that one must be aware of the need to consider this problem, which
does not arise with a simple oscillator.

Sean


On Sat, Oct 11, 2014 at 12:20 AM, Jean-Paul Louis <louijp@yahoo.com> wrote:

> Sean,
>
> This one (SI501) has a spec about jitter. See Page 6 of the data sheet.
> The device is specified as less than 26ps cycle to cycle jitter,
> 16ps period jitter, and less than 3.2 ps phase jitter.
>
> No sure about your requirements, but that part is what I use for a lot of
> my clock needs as the unit cost is also great, and the part is very small=
..
>
> Jean-Paul
> AC9GH
>
>
--=20
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.