Jeff Liebermann wrote:
. . .
It's an observation based upon my measurements with a field strength
meter on similar UHF colinear antennas (using 1/2 wave stubs for
phasing). Also on a center fed 2.4GHz Franklin sector antenna of
similar construction. Most of the voltage peaks were at the base of
the antenna, tapering off as the field strength meter was dragged to
the top of the antenna. Since the current through the antenna is
constant, I assumed that the bulk of the power came from the lower
elements of the antenna. My explanation was a geometric decrease in
radiatated power starting at the feed point.

There's quite a handful of potential problems with this:

1. You might have been in the near field. The relationship between field
strength in the near field and the radiated far field is very complex.
You can't determine the field in one based on measurements in the other.
2. If you're in the near field, the field strength you measure at a
given point depends on the type of antenna used. In the far field, the
field impedance (E/H) is a constant value, but not so in the near field.
In various places in the near field, an antenna which responds more
strongly to the E field (a "high impedance" antenna) will show higher
readings where the field impedance is high, and lower where it's lower.
In any case, the relationship between radiated field and local near
field strength isn't simple.
3. The power applied to the antenna is radiated in all directions,
although of course unequally. As I explained in my last posting, the
total field is the vector sum of the fields from the individual parts of
the antenna. Sampling near the antenna gives you no idea of how the
fields sum at a distant point.
4. It's very difficult to make even roughly accurate measurements even
at HF, let alone UHF or higher. One of several problems is that it's
extremely difficult to decouple the feedline when an electrically small
probe is used, so you end up not measuring what you think you are.

I've also seen a similar effect with relatively high gain (10dbi)
2.4GHz omni antennas in WISP applications. Any blockage of the lower
sections of the antenna, had a much bigger effect on the range and
measure signal strength than covering roughly an equal amount near the
top of the antenna.

That's interesting, and I'd like to get some more information about it.
Perhaps blocking the bottom had a greater effect on the pattern, moving
the maximum away from the direction of the other end of the path?

I can see that on some models. I never could successfully model an
antenna using coax cable sections as elements. Using a wire model,
the current distribution is constant along the length as you describe.
However, my field strength measurements show more RF towards the feed
point. It's difficult for me to tell exactly how much more RF because
my home made meter is not calibrated. I don't recall the exact
numbers but I can dig out the FSM and make some measurements on some
of the antennas I have hanging around on the roof this weekend.

Here's a model of a coax collinear, but using coax with unity velocity
factor. This "Franklin" array model was created by Linley Gumm, K7HFD.
Coaxial cable is modeled as a combination of transmission line model, to
represent the inside of the coax, and a wire to represent the outside.
The technique is described in the EZNEC manual. See "Coaxial Cable,
Modeling" in the index. I've posted the EZNEC equivalent to
http://eznec.com/misc/rraa/ as COAXVERT.EZ. The accompanying Antenna
Notes file is also there as COAXVERT.txt.

CM Coaxial Vertical Antenna
CM
CM ! Wire # 16 for I srcs, shorted/open TL, and/or loads.
CE
GW 1,1,0.,0.,6.76615,.02081892,0.,6.76615,.000127
GW 2,1,0.,0.,5.766841,.02081892,0.,5.725204,.000127
GW 3,1,0.,0.,4.684258,.02081892,0.,4.725896,.000127
GW 4,1,0.,0.,3.684949,.02081892,0.,3.643311,.000127
GW 5,1,0.,0.,2.602366,.02081892,0.,2.644002,.000127
GW 6,1,0.,0.,1.603057,.02081892,0.,1.561419,.000127
GW 7,1,0.,0.,.5204737,.02081892,0.,.5621104,.000127
GW 8,11,0.,0.,6.76615,0.,0.,5.766841,.00635
GW 9,11,.02081892,0.,5.725204,.02081892,0.,4.725896,. 00635
GW 10,11,0.,0.,4.684258,0.,0.,3.684949,.00635
GW 11,11,.02081892,0.,3.643311,.02081892,0.,2.644002, .00635
GW 12,11,0.,0.,2.602366,0.,0.,1.603057,.00635
GW 13,11,.02081892,0.,1.561419,.02081892,0.,.5621104, .00635
GW 14,6,0.,0.,.5204737,0.,0.,0.,.00635
GW 15,1,0.,0.,0.,.02081892,0.,.02081892,.000127
GW 16,1,208.1892,208.1892,208.1892,208.1913,208.1913, 208.1913,2.0819E-4
GE 1
FR 0,1,0,0,144.
GN 1
EX 0,16,1,0,0.,1.414214
NT 16,1,15,1,0.,0.,0.,1.,0.,0.
TL 1,1,2,1,50.,1.040946,0.,0.,0.,0.
TL 2,1,3,1,50.,1.040946,0.,0.,0.,0.
TL 3,1,4,1,50.,1.040946,0.,0.,0.,0.
TL 4,1,5,1,50.,1.040946,0.,0.,0.,0.
TL 5,1,6,1,50.,1.040946,0.,0.,0.,0.
TL 6,1,7,1,50.,1.040946,0.,0.,0.,0.
TL 7,1,15,1,-50.,1.040946,0.,0.,0.,0.
RP 0,181,1,1000,90.,0.,-1.,0.,0.
EN

I've seen models using coax with VF = 0.82 having a good pattern.

Nope. I'll give in easily on this one as it's highly likely I'm
wrong. However, I will double check my measurements on the roof
tomorrow and see if they're reproducible. I may have simply goofed
and/or drawn the wrong conclusion.

Incidentally, I've been offering this observation for several years
and you are the first to question it.

This isn't the first time that's happened.

Roy Lewallen, W7EL