[time-nuts] Typical HP 5370B resolution ?

[Magnus Danielson]

On 03/04/2013 11:02 AM, Poul-Henning Kamp wrote:
> Content-Type: text/plain; charset=ISO-8859-1
> --------
> In message<4056222B698D4625893194A1E3AD4144 at athlon>, "Ulrich Bangert" writes:
>
>> given the fact that the measurement results of a 5370 are integer multiples
>> of 20 ps [...]
>
> Once you get to this level of detail, it is important to know that the
> actual number is not 20ps but 19.5763ps
>
> See page 8-19 in the manual.
>

Since we are in nit-picking mode, on page 8-29 it says the resolution is 
in 5/256 ns, which I get into 19,53125 ps (exact).

One way to understand this is that the gearbox of the 5370B 
interpolators creates a virtual clock of 51,2 GHz. Not bad for a core 
design in 200 MHz and some smart ECL logic from late 1970thies.

Sufficient trigger noise (~ 8 ps) will smooth out the quantization, just 
as Bruce mentions. This interpolation trick was used in the 5328A with 
suitable option in the TI mode. It's good for TI/freq/period averages, 
but not for their stability measures (unless compensated for).
Sufficiently low slew-rate can thus "improve" the result. :)

Cheers,
Magnus