[geeks] KVMs, high-res video, and ghosting

[Dave McGuire]

On December 7, Bill Bradford wrote:
> Anybody know of a company that makes GOOD KVMs, that dont result in
> ghosting/shadows when using resolutions of 1024x768 and higher?  
> 
> I have yet to find one that didnt cost $500+.

  The problem with ghosting in high-resolution video is primarily an
issue of quality of design and quality of materials.

  It's important for video cabling to have a constant impedance.
Incorrect cables, connectors, and other components in the line can
create "bumps", or small mismatches, in line impedance.

  Impedance mismatches cause signal reflections of an amplitude that
corresponds to the degree of mismatch.  These reflections cause
out-of-phase "images" (at the waveform level) of the original signal
to appear on the cable, resulting in ghosting.

  Higher resolution video means higher frequencies present on the video
cable...and since impedance mismatches become much more of an issue at
higher frequencies, higher-resolution video is more susceptible to
ghosting problems with cheap cabling, connectors, and switches than
video at lower resolutions.

  KVM manufacturers can cut costs by using ordinary PC board traces to
move the video around the box between the connectors and the switching
elements.  A "proper" one will either use coax for all the video lines
and terminate it very close to the switching elements with
impedance-matching networks, or use regular PC board traces with
matching networks to correct for any impedance mismatch caused by the
traces.  Those matching networks consist of capacitors and either
inductors or resistors, which do cost money, and are often omitted
because KVM switches are generally associated with cheap commodity
PeeCee hardware.

  Another problem is that the commonly-used HD15 is about as far from a
constant-impedance connector as one can get.  PeeCee People complained
about the "proprietary" DB13W3 (and similar) video connectors used in
the workstation world, but they carry the video signals over coax,
and, if properly designed and installed, provide constant impedance
along the line.  There's no real way around this, due to the
unfortunate proliferation of the terribly ill-suited HD15 video
connector.

  There's no easy way to judge the quality of manufacture of a
particular KVM switch or video cable without cracking it open and
looking at the construction or putting it on a network analyzer (not
the computer network kind, the RF kind) and looking at the impedance
curves.  I suppose an alternative method might be to put it on a
time-domain reflectometer and look for flatness, but I've not tried
that.

      -Dave

-- 
Dave McGuire
St. Petersburg, FL