View Single Post
  #20   Report Post  
Old April 11th 05, 02:36 AM
Richard Clark
 
Posts: n/a
Default

On Sun, 10 Apr 2005 18:17:42 -0500, Tom Ring
wrote:

Hi Tom,

The number of variables in the description of your (Mark's) method is
rather considerable, so I will remark by the parts you offer:

The range has a source yagi for each band, that yagi has a low power AM
modulated signal on it (as I remeber, might be wrong on this).


So far, fine.

The other end has a reference yagi off to the side from the test position.


Commendable.

A yagi with "known gain" is run against the reference yagi, and the
relative signal strength is measured.


I might slyly point out how do you know the gain? It visits the age
old logical knot offered:
In a town of clean shaven men,
there is a barber who shaves everybody who does not shave himself;
who shaves the barber?

That gives us a known gain point
on the meter. A yagi is then placed in the test position and the
relative gain or loss is recorded. The "known gain" yagi can be put
back into the test position at intervals to check the calibration.


This method is called using a "transfer standard." As I offered, that
requires an absolute knowledge somewhere, and you have identified it
in this "known gain" yagi. However, the gain is actually immaterial
until you begin making claims of absolute gain. That is, most of this
correspondence is satisfied with relative gain comparisons as you
point out:

Obviously the absolute values may be suspect, but relative measurements
work well.


Quite true, however, you having once acknowledged suspicions you then
plunge back into the murky pool of absolutes:

The results also agree very well with YO predictions, with a
yagi in the 18.4 dBd predicted range being low by .3 as measured, which
is roughly what he expected to happen. Most more normal gain, 14 to 15
dBd for 432, were within .1 of predicted. Bands tested on this range
were 144, 220, and 432.


Well, here we run counter to my experience with real life components.
They varied by several times your 0.1dB, and this was often times for
the same item tested repeatedly (I never measured any item less than
five times and never five times repeatedly, in a row).

Multiple prototype 2M and 70cm EME antennas that my partner and I built,
stored safe from corrosion, tested the same +- .1dB with a several year
gap between the tests.


OK, the method is good and robust, but your sudden departure from
expected results are on the scale of 5 to 6 times the range of your
typical error.

If this is to be attributed to oxidation on the elements, that still
seems suspect. The oxidation is not lossy, and certainly is not
sufficiently thick enough to shift the resonance. Oxidation is one of
the charms of aluminum, it is self sealing.

I would offer that if the elements oxidized, so did the connectors (or
connections). Simple, repeated connector matings (like swapping in
and out for the range test) were sufficient to break bad contacts and
make the difference which was attributed to scrubbing the elements.
In the normal course of my calibration of various items with
connectors, I always inspected and cleaned them first. N connectors
have erosion problems that will give rise to variations outside of
0.1dB - comes from those threads. The "standard gain" antenna should
be suffering from this erosion by now, but you don't report it.

This raises suspicions for me - you have too much fulfillment of
expectations which is truly extraordinary. I have made thousands of
calibrations of isolators, pads, couplers, meters and so on that have
shown a gaussian distribution of results for premium equipment. Your
range experience shows very little variation - much too little when we
are talking about being within 0.1dB.

What equipment he uses for the ratio measurement, and precisely how it
is done, I don't know. I will attempt to contact him and find out if I
can get this damn sinus infection under control in the next few days.


Well I hope you shake the infection off. Further details are unlikely
to resolve this corrosion as it is too much a matter of "you had to
have been there" kind of thing.

73's
Richard Clark, KB7QHC