[time-nuts] Frequency Distribution Design Basic Module

Charles Steinmetz csteinmetz at yandex.com
Sat Aug 29 17:02:00 EDT 2015

Luciano wrote:

>I didn't write that any instrument needs a reference signal of 
>more  than +13dBm (= 1Vrms) as you write, but a DA must be able to 
>handle the signal levels of the standard on the market.

If a standard puts out more than +13dBm, it can be padded down to 
that level.  There is no need to distribute the signal at the 
(higher) source level unless a client instruments need it (and to my 
knowledge, none does).

>Unfortunately,even in this case what you say does not conform to 
>reality. I am attaching to this mail a comparison I made by taking 
>from  the web some characteristics about DA on the market  or 
>prototypes built by people far more expert than me.
>As you will also see, what we can consider top, made by NIST, has 
>the same harmonic distortion of mine. With rare exceptions that 
>should be evaluated more carefully, harmonic distortion is about 30dBc.

DAs have traditionally had distortion specs in the -40 dBc range, and 
probably for historical reasons many current DA designs still have 
similar levels of distortion.  That does not mean this is the best 
that can be done, and now that we understand that distortion that 
high compromises the phase noise performance of client instruments, 
there is no reason to tolerate it if we can do better.  And we 
can.  Much better.

>To get as you claim 65dBc, you must insert an output band pass 
>filter or a low pass/notch filter

Not true.

>if you have found the web a DA admittedly broad band with features 
>you write (-65dBc)  can you give us the link?

I do not look for DAs on the web.  I design my own.

Best regards,


More information about the time-nuts mailing list