David wrote:
Can anyone post any informationon or web addresses on phase matching and
batch / set matching of coax cables? I am interested in principles and
procedures to phase match coax cables. Phase matching involves cutting coax
cables to the same electrical length. The phase matched cables can then be
analysed and put into batches or sets i.e. cables with similar
characteristics and frequency response are put into a set. What would be
analysed to form a set?

This practice is used on coax cables that feed antennas for satellite
systems. For example, if several antennas receive GPS signals, to
triangulate position, the GPS signals must all arrive at a particular point
at exactly the same time.

It also appears that going for the best quality low loss cable can actually
be worse, because using lower quality coax cable with some loss can dampen
the reflections from in-line connectors.

Stages are 1) phase matching 2) batch matching 3) assemble cables chosen by
batch matching process onto RF system and perform vector generation. A
vector gen file is created which is a calibration snapshot of the system
i.e. records the RF performance in the form of a signature or footprint.

Measuring instrument is a Network Analyser HP8753.

I suppose you left out some (rather important) stages...

The first one I'd put down is: select the cable. The physical design
of the cable is important. Some cable changes appreciably when it
flexes, and some changes appreciably as temperature changes. (If it
changes significantly when the humidity changes, you've REALLY picked
the wrong cable...) Cable with properly installed, high quality
connectors should not show significant reflections from the connectors,
and even if it does, if all the cables are assembled the same, the
effect should be the same on all of them.

Cable of a given physical size and general construction should have
close to the same loss. If you want MORE loss, pick SMALLER cable, and
especially pick cable with smaller inner conductor. Stranded
conductors also help increase the loss. But if you have two similarly
designed cables, and one shows appreciably more loss at GHz
frequencies, definitely avoid it. The key reason is that whatever is
causing the loss is likely the result of a poorly controlled
manufacturing process, and that's something you do NOT want to put up
with in precision matched cable assemblies.

What exactly do you want to match? Phase as a function of frequency,
over some particular range of frequencies? Attenuation, also over the
range of frequencies? Something else? Some combination of
characteristics? I would expect it would depend on the application.
If you assume that the cable assemblies are linear, and that they are
adequately represented for your application by a two-port model, then
measure them with your vector network analyzer. Especially since you
are going for MATCHING and apparently don't need absolute accuracy, let
the analyzer (powered on) and the cables stabilize in the stable
environment where you'll measure them, do a reasonable job calibrating
the analyzer, measure each of the cables (S11, S21, S12, S22, over your
range of frequencies), check the calibration of the VNA, remeasure the
cables (or at least spot-check them) to be sure you get the same
results the second time. If all is good, arrange the cables using
whatever measure of match you need for your application. If you need
matching over a range of temperatures, or where one cable is at a
different temperature than the others, measure under those conditions.
Note that if there are measurement errors, if your technique is
consistent, those errors should be very nearly the same for each cable
that measures the same: matching will be better than absolute
accuracy.

Note that if you are measuring completed assemblies over a range of
frequencies, you likely won't have the option of picking an "odd number
of 1/8 wavelengths," or any other particular number of degrees, for
your measurement. Let's say your cables are 10 nanoseconds long (a
couple meters). If you want to check the match from 1.6GHz to 1.7GHz,
a rather narrow range, the cable length varies by 360 degrees. But
that doesn't matter: for matching, it's the repeatability of the
measurement that does matter. If you get to caring about absolute
accuracy, check the accuracy specs of your instrument, and read ap
notes...

Have you checked for HP and/or Agilent and/or Rohde & Schwarz and/or
Anritsu (and perhaps some others) ap notes? I know for sure that there
are a lot of them from Agilent (which may have originally been written
at HP, before Agilent split off). Since you are using an HP8753, give
you local Agilent sales rep a call if you can't find the ap notes on
the web, and ask her/him for applications information.

Cheers,
Tom