[time-nuts] cable delay variation (was: Question about the PLL of Trimble Thunderbold)

[Attila Kinali]

On Mon, 29 Oct 2018 12:03:25 -0700
Hal Murray <hmurray at megapathdsl.net> wrote:

> If I go looking for good cables, do they specify temperature coefficient?  I 
> don't remember ever seeing it when scanning specs, but I probably wasn't 
> looking for it so I could have skimmed over something.

Yes. The key word here is "phase stable coax." If you search for that,
several cables will pop up. They wont be cheap, though.


> What determines the dispersion of a cable?  Is it as simple as bigger wire 
> (more copper) is better?

For us, it's mostly the frequency dependence of the dielectric.
If you go for higher frequencies, it also becomes a matter of
the different modes a cable supports, which all have different
velocities. The latter is the reason, why GHz cables and connectors
are becomming thinner and thinner as we go up in frequency.


> Why do I care about the dispersion as long as all cables match?

Because it degrades the slew rate of the pulse. If you have
a very bad case of dispersion (combined with long cables),
the rising slope will look more like a jagged mountain range
than a step (though it's unlikely you hit that with modern
cables, unless you go for several km of cable). Attached is
a picture of what it might look like.

If you remember the old telegraph and telephone lines, they
all used to have inductors placed on them every few km, to
compensate the dispersion. As most of our communication these
days is digital and we have relative short lengths before a
regeneration step happens, there isn't much need for dispersion
compensation anymore. Unless you go for submarine cables or
optical fibers, both of which have elements with negative
dispersion inserted.

			Attila Kinali

-- 
<JaberWorky>	The bad part of Zurich is where the degenerates
                throw DARK chocolate at you.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: dispersion.png
Type: image/png
Size: 672 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20181029/46b9e344/attachment.png>