On Apr 10, 2:24 pm, "Antonio Vernucci" wrote:
I am also surprised by the high resistance you are seeing with the
hairpin wires removed. From what you posted before, and from the 3-
element NBS and the 6-element designs I ran in EZNEC, I would expect a
lower feedpoint resistance than that. If I put a pure inductance in
parallel with your 42.4-j39 ohms, I get the same answer as you did,
that it moves to 78 ohms resistive.

My offer to look at your EZNEC file is still open, of course. Maybe I
could see something in it--I know that I can look and look at
something and not see an obvious problem, even when I know perfectly
well what to look for.

If your personal mail on the newsgroup is correct, I could send you the EZNEC
file of my antenna.

Also--in the model before you split the D.E. into three wires, did you
get a more reasonable feedpoint impedance (with no hairpin, just
looking at the D.E. feedpoint)? I think we were expecting something
like 19-j58 ohms. If you get that, then I would look for the reason
things change (so very much!) when you split the D.E. into three
separate (but connected, end-to-end) wires. I can understand a small
change but not so large a change. Is the total number of segments for
the three wires making up the D.E. still about the same as it was in
the original version of the model, with just one wire for the D.E.?

I am also surprised as I would have expected a much lower resistance (as you
suggested)

I did the test you have suggested and I noted some change, but not much. With a
"solid" D.E., i.e. not broken in three pieces (and without the hairpin of
course), impedance is 43.27 - j 40.97 ohm, not terribly away from 42.4 - j39
ohm. In that test I configured EZNEC for 17 segments, whilst when the D.E. was
broken in three parts I had used 17 segments for the two outer arms and only 5
segments for the 10-cm center section, due to its much shorter length (I see
that result depend somewhat on the number of segments).

73

Tony I0JX

Yes, the email should be right. I do have Eznec5+, so I can check
against Roy's suggestion. I'm sure he'll be happy to help with things
that may be related to how NEC/EZNEC behaves, if we can identify them.

Cheers,
Tom

What you're probably seeing is a numerical problem in the NEC calculating
engine. It's very fussy about the region near a source, and doesn't like small
loops which include a source. You should run an Average Gain check (see
"Average Gain" in the EZNEC manual index), which will reveal whether this is
the problem.

The double precision calculating engine in EZNEC+ is considerably more
tolerant of small loops, but can still have problems with average gain for
other reasons.

Roy Lewallen, W7EL

Hi Roy,

thanks for the tip.

As a matter of fact without the hairpin the average gain is almost zero, whilst
with the hairpin is about -0.8 dB, that correponds to the gain drop I notice. So
the problem you had anticipated actually occurs.

I then tried to simulate the hairpin with a proper-length shorted trasmission
line and, doing so, the average gain is almost 0. Evidently the program does not
like the short hairpin loop.

73

Tony I0JX

Yes, the email should be right. I do have Eznec5+, so I can check
against Roy's suggestion. I'm sure he'll be happy to help with things
that may be related to how NEC/EZNEC behaves, if we can identify them.

Cheers,
Tom

Tom,

the mistery seems to be solved:

- with the "real" hairpin the program does not yield correct answers, neither in
terms of gain nor in terms of impedance. Roy W7EL anticipated that the problem
is due to the feedpoint being in a short loop. I did the test he suggested (i.e.
calculating the Average Gain) and the result I got clearly showed that there was
a problem.

- instead, with the hairpin simulated by means of a shorted piece of
transmission, the results are fully consistent. Firstly the gain does not vary
with and without the hairpin. Secondly the impedance I read removing the hairpin
is 24.5 +j65.37 ohm, a value fully consintent with the pure 200 ohm I get with
the simulated hairpin.

At least I learned one should not try to model "real" hairpins.

At this point I would not further bother you, by sending you files that would
not tell you much. It was a real pleasure for me to have such an interesting
discussion with you and, if I may, I will write you should some other strange
situation occur in the future.

73

Tony I0JX

Antonio Vernucci wrote:
I then tried to simulate the hairpin with a proper-length shorted
trasmission line and, doing so, the average gain is almost 0.

Is that shorted transmission line lossless?
--
73, Cecil http://www.w5dxp.com

I then tried to simulate the hairpin with a proper-length shorted trasmission
line and, doing so, the average gain is almost 0.

Is that shorted transmission line lossless?
--
73, Cecil http://www.w5dxp.com

I presume so. It would seem to me that in EZNEC there is no way of simulating a
lossy transmission line.

73

Tony I0JX

On Apr 11, 11:54 am, "Antonio Vernucci" wrote:
Yes, the email should be right. I do have Eznec5+, so I can check
against Roy's suggestion. I'm sure he'll be happy to help with things
that may be related to how NEC/EZNEC behaves, if we can identify them.

Cheers,
Tom

Tom,

the mistery seems to be solved:

- with the "real" hairpin the program does not yield correct answers, neither in
terms of gain nor in terms of impedance. Roy W7EL anticipated that the problem
is due to the feedpoint being in a short loop. I did the test he suggested (i.e.
calculating the Average Gain) and the result I got clearly showed that there was
a problem.

- instead, with the hairpin simulated by means of a shorted piece of
transmission, the results are fully consistent. Firstly the gain does not vary
with and without the hairpin. Secondly the impedance I read removing the hairpin
is 24.5 +j65.37 ohm, a value fully consintent with the pure 200 ohm I get with
the simulated hairpin.

At least I learned one should not try to model "real" hairpins.

At this point I would not further bother you, by sending you files that would
not tell you much. It was a real pleasure for me to have such an interesting
discussion with you and, if I may, I will write you should some other strange
situation occur in the future.

73

Tony I0JX

Hi Tony,

I suppose you are not using EZNEC 5+. If that is the case, I would be
happy to run your model on the 5+ version with double-precision
calculations, and see if that makes a difference. It should be easy
enough to run it. It it no problem to me, just a fun diversion. And
of course I would be happy to have discussions with you in the future.

Cheers,
Tom

Antonio Vernucci wrote:
Is that shorted transmission line lossless?

I presume so. It would seem to me that in EZNEC there is no way of
simulating a lossy transmission line.

So what would be the results if the real-world
transmission line losses were accounted for?
--
73, Cecil http://www.w5dxp.com

On Apr 11, 12:49 pm, "Antonio Vernucci" wrote:
I then tried to simulate the hairpin with a proper-length shorted trasmission
line and, doing so, the average gain is almost 0.

Is that shorted transmission line lossless?
--
73, Cecil http://www.w5dxp.com

I presume so. It would seem to me that in EZNEC there is no way of simulating a
lossy transmission line.

73

Tony I0JX

At least in EZNEC 5+, the Transmission Lines window has an entry for
line loss, in dB/100 feet at a frequency you also enter. But the loss
for your 0.8cm diameter aluminum tubing spaced 10cm is only about
0.06dB/100 feet at 50MHz. It is certainly safe to ignore that,
considering that the loaded Q of the matching system is vastly lower
than the Q of the hairpin line as a resonator.

Cheers,
Tom

Antonio Vernucci wrote:

I presume so. It would seem to me that in EZNEC there is no way of
simulating a lossy transmission line.

EZNEC v. 5.0 has that ability. Earlier versions do not.

Roy Lewallen, W7EL

I presume so. It would seem to me that in EZNEC there is no way of
simulating a lossy transmission line.

EZNEC v. 5.0 has that ability. Earlier versions do not.

Sorry Roy,

I still have the old 3.0 version and I shall certainly get the newer version.

On this occasion, may I ask a question privately?

73

Tony I0JX