[time-nuts] DDS module

[Magnus Danielson]

On 22/07/13 04:42, Didier Juges wrote:
> I am familiar with the Epson Seiko SG8002 series and nowhere does it say they are programmable. They are not user programmable for sure, only the distributor has the tools to do it. The data sheet does not have any information that would be a hint that they are PLL's.
>
> Http://WWW.KO4BB.com/Timing/MicroprocessorCrystals.php

Oh, those. I remember those. About 10 years ago we had a lot of issues 
with these programable clocks. By measuring the 1 cycle jitter profile 
(use what you got, had a good sampling scope) I could fairly quickly see 
if it was up to no good or a possible candidate. You could clearly see 
the separation into two lines for the clock and that gave us a hint, 
which correlated well with BER estimates on the high-speed links.

If you look around, you will find several companies patenting their 
field programmable device, and it's infrastructure more on the concept 
level than the actual technology in the crystal. It's a nice complement 
for many odd frequencies, but when you want a *clean* signal it doesn't 
help. We had the distributors saying things like "you can't see that!" 
and we replied "Sure we can, look!". It was at that time that 
oscillator-selection no longer was a datasheet and price issue for us, 
but testing became a routine issue for the critical stuff. It was a good 
learning experience for a then young company and yours truly.

Cheers,
Magnus