[time-nuts] Are the days of buying a crystal numbered?

Magnus Danielson magnus at rubidium.dyndns.org
Mon Feb 22 23:30:37 UTC 2010

Hal Murray wrote:
> The trick of including a PLL in an oscillator package has been around for a 
> while.
> The initial ones were programmed at the factory or distributor.  The idea was 
> to avoid the long wait while they polished the crystal to your specific 
> frequency.  They stocked them in a handful of frequencies and programmed the 
> right one when they got your order.
> Early ones had a bad reputation due to lots of jitter.

They where pretty bad. It got worse that the reps where denying that 
there was a problem. Single-cycle jitter (which was relevant for that 
app) was clearly vissible and didn't take much effort to measure and 
toss the samples over the shoulder.

But they where fairly simple PLL designs.

One case I saw would allow the PLL to oscillate badly sometimes when you 
started it. For that app it was totally useless.

> These look good to 
> me.  (But I don't work in this area so maybe their marketing stuff fooled me.)
> This is time-nuts.  What is good for one application will be useless for 
> another.
> Spurs might be a problem, it depends on your application and the exact 
> numbers of the target frequency and the crystal frequency.

I could not agree more. These are good contribution to the design-scope 
for some applications.


More information about the time-nuts mailing list