[time-nuts] External Frequency Standard Inputs on Counters

Magnus Danielson magnus at rubidium.dyndns.org
Sat Apr 4 12:18:35 UTC 2009

Bruce Griffiths skrev:
> Hej Magnus
> I wasn't recommending this method for anything but a possible solution
> when no other method was feasible.
> Of course the other solution is to use a sufficiently short total cable
> length.
> The only questions are:
> How short ?
> How does the counter reference input impedance (R and C) affect the
> maximum total cable length?
> Daisy chaining via the reference output signal provided by some counters
> will increase the noise of the reference for counters further along the
> chain.
> The significance of this depends on the application and the performance
> of the reference frequency buffers inside the counters.

Naturally. But one can put the less sensitive or needing boxes down the 
line. It is worth mentioning that there is two principles for using 
external clock, one is to just select it in and distribute it inside 
where as the other is to phase-lock the internal crystal oscillator.

The later variant is better for those systems that has a low noise clock 
and frequency steering is needed.

> Using a high performance distribution amplifier reduces interaction
> between counters and the potential degradation due to each counter's
> reference output buffer.

Certainly. It also isolates various cables from each other. Thus, 
disconnecting a box does not become a problem. Also, there is no need to 
have all boxes turned on.

> Bruce
> Magnus Danielson wrote:
>> Bruce Griffiths skrev:
>>> Magnus
>>> Another option (useful when the source amplitude is such that
>>> terminating isn't an option and there isn't a suitable amplifier
>>> available to boost the source amplitude) is to trim the cable lengths to
>>> eliminate such standing wave nulls at the counter inputs.
>> Certainly, but consider the frequency of 10 MHz, 100 ns. 1 ns is 2 dm 
>> one-way, so a full-wave is 20 m, a half-wave is 10 m and a quarter-wave 
>> is 5 m. It is certainly possible, but a lot of cable.
> Thats true only for cable like RG58 with a solid PE dielectric.
> For cables having gas injected PE or PTFE foam dielectric the
> corresponding lengths are a little greater.

Certainly, but it does not help against the main point, it's alot of cable.


More information about the time-nuts mailing list