[time-nuts] Re: Frequency counter recommendation wanted

[Lux, Jim]

On 3/13/23 4:38 PM, Bob Camp via time-nuts wrote:
> All of this still is based on a “I want to measure 10 MHz” sort of baseline. If 22.48 GHz is the typical
> target, that’s a whole different ballpark. Multiple channels (to measure delta’s on 1 pps signals )
> also change the target a bit.
>
> The next branch to this thing is the timebase. In a lot of labs, the counter had a TCXO (or XO) in
> it. It never / ever ran on that timebase. It always was locked to “something else”. In the modern
> day, that might well be a GPSDO. Indeed this does have an impact on your counter purchase.


And a trap I've been caught in is "just because it has an external 
reference input, that doesn't mean the output is in anyway locked to 
that reference"


Many modern replacements for the 8663 of the 1980s 1990s have this issue 
- JPL used tons of 8663s, and developed test setups using  the 10 MHz 
input to do things like modulate the phase of the output, because it's 
directly locked.  This is not the case with the modern E8663D, for 
instance.

The Keysight 336xx function generators have a 10 MHz input, but it feeds 
into a FLL of sorts with the internal oscillator, and all they do is 
make sure the internal oscillator is at the same frequency (with some 
TBD transfer function, much like a GPSDO). If you have two of them 
"locked" to the same source, there's no reason why they will maintain a 
fixed phase relationship.  This is very unlike the 3325 or similar 
function generators, which might have an indeterminate phase shift, but 
once you knew what it was, it stayed the same as long as they were 
powered up.  Instead, you need to use an explicit sync signal to all the 
generators.