[time-nuts] Question about SA.33 Rb clock
kb8tq at n1k.org
Tue Oct 17 10:50:52 EDT 2017
> On Oct 17, 2017, at 12:37 AM, Hal Murray <hmurray at megapathdsl.net> wrote:
> kb8tq at n1k.org said:
>> The gotcha is that the spur spec’s are not always met. As you might guess,
>> doing testing over really wide DDS ranges is impractical. Some designs use
>> cleanup loops. The gotcha then becomes a spur (say at 0.053 Hz) that is
>> inside the cleanup loop bandwidth ….
> How narrow can I make the loop bandwidth with a traditional analog filter
> using reasonable parts?
Given that this is Time Nuts …. If you analog filter can be the size of a 6’ rack,
it can be pretty narrow :) The bottom line is that it also gets *very* expensive.
> Do people build GPSDO style digital filters?
If they are smart they do. It’s a much lower cost way to do it. Today they can
yield results that are just as low noise as an analog approach.
> Is there a reasonable formula for the spurs from a DDS? That sounds like a
> place where guaranteed-by-design would be appropriate.
You can predict some of the spurs, but not all of them. Full prediction would
require a *lot* of data about the exact ADC you have on your chip. What gets
guaranteed is some sort of “worst case -80 dbc” sort of stuff. The gotcha is that
a spur at 80 db down is a really big deal if it’s 0.01 Hz off carrier and you are
looking at ADEV.
In a Telecom Rb sort of application (or in a lot of other applications) the performance
you get is “plenty good enough” to build a system that works. If that’s the case,
why spend more money?
> These are my opinions. I hate spam.
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
More information about the time-nuts