AnsweredAssumed Answered

Questions about clock generators and DDS

Question asked by AtomicClock on Jan 17, 2012
Latest reply on Jan 19, 2012 by pkern

I have 10 MHz ultrastable sine wave ref. (13 dBm) from Rb atomic clock.

My targets:

1. freq synthesizing from few MHz to few hundred MHz with good phase coherence (low jitter);

2. Multi channels with different freqs.


Structure: I need a fast clock to genereate stable  clock signs. from 100sMHz up to 1 GHz and then use them as system clock  for various of DDS directly. This clock generator is reference to atomic  clock at 10 MHz eventually.



1. Connections between atomic clock and clock generator: I  noticed that AD9523-1 requires input slew rate no less than 400 V/us in  order to guarantee the nominal jitter performance of around 0.18 ps. But  13 dBm 10 MHz sine wave is one order of mag. lower than this  requirement. On the other hand, employing fast comparator (say ADCMP573) would bring  extra random jitter of 0.2 ps and deterministic jitter of 10 ps. Which  plan would be ultimately better: driving it directly or using  comparator? For the former it seems I could use a transfer to boost the  amp. and then clamp V by diodes to increase slew rate;


2. Connections between clock generator and DDS: all  clock generators' (from AD) output is digital, and normally DDS could  accept it. But why does the datasheet of AD9959 evaluation board say  sine wave is required? Should I ignore it and hook them up (EVAL board  of ADCMP573 -> EVAL board ofAD9959) directly anyway?


3. Last question is about buffer: for better isotlation, I'd like  to transformers to couple each channels. A 1 GHz digital sig. would be  greatly distorted due to the limit of the BW of transfors(so hard to  find one with a BW large enough to cover first few harmoics). How to  solve this problem?


Thank you very much!