Insulated Wire Velocity Factor: How to . . ??
 

I'm in the habit of using insulated stranded all-copper conductors for
HF wire antennas. Usually U.S. standard #14 Type THHN "house wire"
because of it's easy availability everywhere at low cost. However
insulated wires pose an annoying problem, their specific velocity
factors are not published and vary all over creation depending on a
whole collection of variables.

I've tried to nail down the Vf of my usual #14 THHN by cutting the
lengths of 20M dipoles to one or another of the usual equations. I put
together the antennas, hoist them to various operatng heights and nip
the coax feedline to it's minimum possible length so that I can find
the resonant point with an antenna analyzer while I'm on the ground
directly under the feedpoint. A process which I believe should lead me
to a "correction factor" from which the Vf can be determined.

Modeling a 20M dipole at 35 feet indicates a fairly sharp null at the
resonant point. But that's not what I get when I build the antenna and
try to measure the frequency of it's resonant point with an antenna
analyzer. I get a much flatter SWR curve from the analyzer than I do
from modeling probably because of feedline losses and because of the
low analyzer frequency resolution (MFJ-259B). To the point where the
antenna appears to be resonant over a range of maybe 200Khz. Which in
reality it can't be.

From a practical standpoint this scenario isn't any problem in the case
of a simple dipole. I "cut long", put the antenna up, sweep it with the
analyzer, find what seems to be the center of resonace, do the quickie
numbers, trim it and take it to the airwaves.

The problem comes when trying to accurately model complex, fussy wire
antennas like hex beams when the Vf of the wire is unknown. A one
percent error in conductor length at 14 Mhz is 140 Khz which makes
decent modeling just about useless.

So two questions in this regard: Is there a way to measure the Vf of a
wire without having to resort to using 2" Heliax to feed a dipole and
without a lab full of HP and GR test equipment? Second, assuming the
Vf becomes known how does one handle it during the modeling process?
Model the antenna wire lengths at an upward-shifted frequency based on
the Vf?

Thanks,

w3rv

On 28 Mar 2005 11:52:23 -0800, "Brian Kelly" wrote:

So two questions in this regard: Is there a way to measure the Vf of a
wire without having to resort to using 2" Heliax to feed a dipole and
without a lab full of HP and GR test equipment? Second, assuming the
Vf becomes known how does one handle it during the modeling process?
Model the antenna wire lengths at an upward-shifted frequency based on
the Vf?

Hi Brian,

1. Drive the design with power instead of low level excitation;

2. Remove half the transmission line muffling of results by using a
field strength meter to find resonance (another reason for power);

3. Find the Vf (as you put it) by derivation against a wire model
(through the difference in lengths of bare wire model resonance to
real wire resonance);

4. Use the new EZNEC which allows you to employ insulation over wire
and adjusting the thickness to conform with results found
experimentally with real wire at actual length;

5. Assign these insulation properties to all future designs in the
modeler.

73's
Richard Clark, KB7QHC

If you have a MFJ-259, you can get the VF of the wire easily
or
1. put up a 10-meter dipole using bare wire - note the resonant freq
2. duplicate exactly using 14 THHN Str - note the NEW resonant freq.

divide the resonant freq from step 1 by the resonant freq from step 2. (or
is it the other way around)

Anyway, I don't think you have to add your social security number.

or just use .95 as the VF and adjust as needed

On Mon, 28 Mar 2005 20:23:34 -0500, "Hal Rosser"
wrote:

or just use .95 as the VF and adjust as needed

Or buy a copy of Eznec v4

Richard Clark wrote:
On 28 Mar 2005 11:52:23 -0800, "Brian Kelly" wrote:

So two questions in this regard: Is there a way to measure the Vf of
a
wire without having to resort to using 2" Heliax to feed a dipole
and
without a lab full of HP and GR test equipment? Second, assuming
the
Vf becomes known how does one handle it during the modeling process?
Model the antenna wire lengths at an upward-shifted frequency based
on
the Vf?

Hi Brian,

1. Drive the design with power instead of low level excitation;

Sweep the dipole with a transmitter and an SWR bridge?

2. Remove half the transmission line muffling of results by using a
field strength meter to find resonance (another reason for power);

Same as above but with a field strength indicator? Just might work if I
use a 4-digit DVM and a diode.

3. Find the Vf (as you put it) by derivation against a wire model
(through the difference in lengths of bare wire model resonance to
real wire resonance);

That would seem to work but I'd expect to still have the flat curves
because of the coax losses. I'm starting to think I should go to a
lower frequency band like 40 or 80M to reduce the problems with the
coax. And to reduce the errors in cutting-to-length.

4. Use the new EZNEC which allows you to employ insulation over wire
and adjusting the thickness to conform with results found
experimentally with real wire at actual length;

.. . . all I gotta do is DO that! "The loop has been closed."

5. Assign these insulation properties to all future designs in the
modeler.

I've been using Nec Win Plus which is OK but it doesn't have the the
ability to handle velocity factors like EZNEC 4.0 can.

I don't have a big problem with scaling antenna dimensions to adjust
for the Vf because I physically model antennas with CAD first to get
the locations of the wire end points in 3D space. Which I can quickly
and easily load into NWP. The CAD program does all the tedious trig for
me. When I have a bare-wire model which "works" in NWP I can rescale
the physical model by 0.98 or 0.95 or whatever the Vf might be to get a
"close enough" fully dimensioned antenna design. But I still need to
find the Vf experimentally and we're back to square one. You fed me
some thinking fodder, I'll try a few things per above and get there one
way or another.

Tnx.

73's
Richard Clark, KB7QHC

w3rv

Hal Rosser wrote:
If you have a MFJ-259, you can get the VF of the wire easily
or
1. put up a 10-meter dipole using bare wire - note the resonant freq
2. duplicate exactly using 14 THHN Str - note the NEW resonant freq.

divide the resonant freq from step 1 by the resonant freq from step
2. (or
is it the other way around)

I wish. Finding the resoant frequecies is my fundamental problem.

Anyway, I don't think you have to add your social security number.

Heh.

w3rv

On 30 Mar 2005 08:08:27 -0800, "Brian Kelly" wrote:

2. Remove half the transmission line muffling of results by using a
field strength meter to find resonance (another reason for power);

Same as above but with a field strength indicator? Just might work if I
use a 4-digit DVM and a diode.

Excellant choice (add a filter cap too with resistive load for
averaging).

3. Find the Vf (as you put it) by derivation against a wire model
(through the difference in lengths of bare wire model resonance to
real wire resonance);

That would seem to work but I'd expect to still have the flat curves
because of the coax losses.

Hi Brian,

Actually, by using the FSM you entirely remove the transmission line
as disturbance to accurate response readings. Those come from the
external reading which interprets all power being applied AT the
antenna junction. However, it imposes upon you that you be scrupulous
about achieving the same drive levels at all the intermediate
frequencies across the swept band. If you do that, then the
transmission line characteristics for the drive going up to the
antenna junction fall out too.

Careful drive monitoring, and careful response monitoring render the
transmission line transparent to the measurement. Thus response/drive
is the antenna characteristic. Define one point's SWR, and you can
cast that into the suite of readings for a swept SWR curve. Take care
in that "one" SWR determination to anticipate the SWR lowering effect
of transmission line loss.

Then you do the same thing in software, and tailor the characteristic
insulation thickness to match your measurements. Having achieved
that, then you have your standard insulation. This does not give you
Vf until you then remove that virtual insulation and find the native,
bare wire resonance. This last step is satisfying (it answers your
question as to Vf), but the step before is more useful because you can
model other antennas from that standard.

73's
Richard Clark, KB7QHC

"Wes Stewart" wrote in message
...
On Mon, 28 Mar 2005 20:23:34 -0500, "Hal Rosser"
wrote:

or just use .95 as the VF and adjust as needed

Or buy a copy of Eznec v4


or just use .95 as the VF and adjust as needed

On Wed, 30 Mar 2005 18:16:04 -0500, "Hal Rosser"
wrote:


"Wes Stewart" wrote in message
.. .
On Mon, 28 Mar 2005 20:23:34 -0500, "Hal Rosser"
wrote:

or just use .95 as the VF and adjust as needed

Or buy a copy of Eznec v4


or just use .95 as the VF and adjust as needed

Or use .1 and adjust as needed.

Richard Clark wrote:
On 30 Mar 2005 08:08:27 -0800, "Brian Kelly" wrote:

2. Remove half the transmission line muffling of results by using
a
field strength meter to find resonance (another reason for power);

Same as above but with a field strength indicator? Just might work
if I
use a 4-digit DVM and a diode.

Excellant choice (add a filter cap too with resistive load for
averaging).

Yup.

3. Find the Vf (as you put it) by derivation against a wire model
(through the difference in lengths of bare wire model resonance to
real wire resonance);

That would seem to work but I'd expect to still have the flat curves
because of the coax losses.

Hi Brian,

Actually, by using the FSM you entirely remove the transmission line
as disturbance to accurate response readings. Those come from the
external reading which interprets all power being applied AT the
antenna junction. However, it imposes upon you that you be
scrupulous
about achieving the same drive levels at all the intermediate
frequencies across the swept band. If you do that, then the
transmission line characteristics for the drive going up to the
antenna junction fall out too.

"Eureka". You're right. This is the way to go. Or at least to try.

Careful drive monitoring, and careful response monitoring render the
transmission line transparent to the measurement. Thus
response/drive
is the antenna characteristic. Define one point's SWR, and you can
cast that into the suite of readings for a swept SWR curve. Take
care
in that "one" SWR determination to anticipate the SWR lowering effect
of transmission line loss.

Since coax losses don't vary much if at all over any of the individual
HF ham bands a decent inline wattmeter with maybe a 4 inch scale should
allow me to maintain a constant power output over the sweep.

Then you do the same thing in software, and tailor the characteristic
insulation thickness to match your measurements. Having achieved
that, then you have your standard insulation. This does not give you
Vf until you then remove that virtual insulation and find the native,
bare wire resonance.

Agreed.

This last step is satisfying (it answers your
question as to Vf), but the step before is more useful because you
can
model other antennas from that standard.

That's what I need. It'll make a worthwhile weekend project which, if
successful, should result in less futzing around with the cutters and
the soldering iron up the tower. I'll also compare the experimental
results of the bare wire sweeps to the predictions given by the modeler
and "calibrate" the modeler in this respect too. Might lead me to my
own real world ground condx vs. the generic "real ground" in the
modeler which is another big source of modeling non-truths.

73's
Richard Clark, KB7QHC

w3rv

"Wes Stewart" wrote in message
...
On Wed, 30 Mar 2005 18:16:04 -0500, "Hal Rosser"
wrote:


"Wes Stewart" wrote in message
.. .
On Mon, 28 Mar 2005 20:23:34 -0500, "Hal Rosser"
wrote:

or just use .95 as the VF and adjust as needed

Or buy a copy of Eznec v4


or just use .95 as the VF and adjust as needed

Or use .1 and adjust as needed.


NOW ur talkin' :-)

On 30 Mar 2005 19:16:44 -0800, "Brian Kelly" wrote:

Since coax losses don't vary much if at all over any of the individual
HF ham bands a decent inline wattmeter with maybe a 4 inch scale should
allow me to maintain a constant power output over the sweep.

Hi Brian,

I left that unsaid, expecting someone, if not you, would also come to
that conclusion. It does not work across all bands, but within a band
it will suffice. Also, even given the impression of accuracy that
most impart to their power meters, this method demands only "relative
accuracy" which can be exceptional when care is shown (try to maintain
a full scale indication or at least greater than 2/3rds at some
cardinal point on the scale).

At this point, one should reflect that if there is a mismatch, then
power at the feed point will vary somewhat. In other words, the
presumption of constant power (to subtract out the effects of the
transmission line) is violated. However, as a first pass estimation,
the method is still quite productive, and tightening up the method and
the numbers is an exercise left to the experimenter.

73's
Richard Clark, KB7QHC

"Brian Kelly" wrote in message roups.com...

I've been using Nec Win Plus which is OK but it doesn't have the the
ability to handle velocity factors like EZNEC 4.0 can.

You could also try 4nec2 in which a provision was added (as described
by L.B. Cebik) to model insulated wires.

The CAD program does all the tedious trig for
me. When I have a bare-wire model which "works" in NWP I can rescale
the physical model by 0.98 or 0.95 or whatever the Vf might be to get a
"close enough" fully dimensioned antenna design. But I still need to
find the Vf experimentally and we're back to square one. You fed me
some thinking fodder, I'll try a few things per above and get there one
way or another.

It also includes a drawing (drag and drop) style geometry editor and
you can rescale part of or the whole structure.

Arie.

build a scale model with insulation at 900 MHz and test it on a network
analyzer in the lab

remove insulation and retest?


Mark

The question that comes first to my mind is, "Why do you care?"
Certainly an antenna does not need to be resonant to work well. I can
imagine you'd like a reasonably low indicated SWR, just so your
transmitter has a reasonable load to drive.

If you really want to know what's going on at the antenna feedpoint,
you'll need to back the effects of the feedline out of your antenna
analyzer readings, or use an analyzer that does it for you. If you
have a reasonable estimate of the feedline loss and know its electrical
length (easy to find if you put a short at the end of the line and look
at the resulting impedances read on the analyzer), then you should be
able to translate your analyzer readings to actual feedpoint impedance.
Do you have the feedline properly decoupled from the antenna so it's
not a significant part of the radiating system? If not, there seems
little reason to bother making the measurements.

I'd expect half-wave dipole resonance to result in lowest SWR on a
50-ohm feedline, but it won't be a very sharp minimum. So is it worth
worrying about?

Another 'speriment to try: build a fairly wide-spaced two wire
transmission line from your wire. Short it at one end, open at the
other, and look for quarter-wave resonance; or short both ends and look
for half-wave resonance. Measure the resonant frequency, which will be
a pretty sharp resonance (much sharper than the dipole). Remove the
insulation and see how much the resonance changes. Try for various
spacings to see what effect the spacing has. (Expect that close
spacings will show more effect than wide.)

Cheers,
Tom

K7ITM wrote:
The question that comes first to my mind is, "Why do you care?"
Certainly an antenna does not need to be resonant to work well. I
can
imagine you'd like a reasonably low indicated SWR, just so your
transmitter has a reasonable load to drive.

If you really want to know what's going on at the antenna feedpoint,
you'll need to back the effects of the feedline out of your antenna
analyzer readings, or use an analyzer that does it for you. If you
have a reasonable estimate of the feedline loss and know its
electrical
length (easy to find if you put a short at the end of the line and
look
at the resulting impedances read on the analyzer), then you should be
able to translate your analyzer readings to actual feedpoint
impedance.
Do you have the feedline properly decoupled from the antenna so it's
not a significant part of the radiating system? If not, there seems
little reason to bother making the measurements.

I'd expect half-wave dipole resonance to result in lowest SWR on a
50-ohm feedline, but it won't be a very sharp minimum. So is it
worth
worrying about?

I agree 100%, it's not worth worrying about - if all I wanted is to get
a 20M dipole running on the air. But that's not my point. I'm trying to
use dipoles to determine the velocity factors of insulated wires used
for the radiators.

Another 'speriment to try: build a fairly wide-spaced two wire
transmission line from your wire. Short it at one end, open at the
other, and look for quarter-wave resonance; or short both ends and
look
for half-wave resonance. Measure the resonant frequency, which will
be
a pretty sharp resonance (much sharper than the dipole). Remove the
insulation and see how much the resonance changes. Try for various
spacings to see what effect the spacing has. (Expect that close
spacings will show more effect than wide.)

Now we're cookin', I like it, this approach has definite appeal and I
need to explore it for several reasons. First because of the sharp
nulls, it takes out the coax and it's a simple sort of "bench test" I
can do at ground level. I'll build three identical shorted 20M or 30M
close-spaced quarter wave lines. One with bare #14 stranded wire, one
with the insulated #14 THHN wire I usually use for quick & dirty
dipoles and one with The Wireman's very flexible #544 or #546 #14
insulated PVC jacketed wire I'd use to build a hex wire beam or a quad.


Then I'd use a grid dip meter to find the resonant frequencies of all
three. I'd use an HF rcvr with a digital freq display to listen to the
GDO rather than trust the GDO dial calibration and resolution. Yes?

Which leads into another head-scratcher I've had in the past. I've had
a bad time coupling a GDO to quad elements because it takes a couple
turns of wire near the GDO coil to get enough coupling between the quad
element and the GDO. Which in turn means that I've shortened the
element length and the result is wrong.

What's your suggestion on a method to accurately measure the resonant
frequencies of the quarter-wave lines in this exercise?

Thanks,

Cheers,
Tom

w3rv

Brian Kelly wrote:

Which leads into another head-scratcher I've had in the past. I've had
a bad time coupling a GDO to quad elements because it takes a couple
turns of wire near the GDO coil to get enough coupling between the
quad
element and the GDO. Which in turn means that I've shortened the
element length and the result is wrong.

What's your suggestion on a method to accurately measure the resonant
frequencies of the quarter-wave lines in this exercise?

OK, I admit to being a spoiled brat when it comes to making
measurements like this. The easiest for me would be to set up a
network analyzer, with the analyzer's source and receiver each coupled
lightly to the line. But there are other ways. You may not even need
a signal generator. You could loosely couple a receiver to the line,
and loosely couple an antenna to it on the other side, and as you tune
the receiver across the resonance, you should notice a sharp peak in
atmospheric noise. With a loaded Q of a few hundred, the peak at 10MHz
would be a very few tens of kHz wide. You could also build an
oscillator which uses the tuned line as the frequency-determining
element, and just count the frequency of the oscillator. A simple
version of a network analyzer could be done by lightly coupling an RF
generator into the line, and putting an RF detector across the line a
small distance up from the shorted end. I'd use either a simple diode
detector, which can have pretty high input impedance, or one of the
Linear Technology or Analog devices RF detectors, but since those are
lower impedance, tap them down very far on the line. -- I'd expect a
10MHz quarter-wave resonator made from 600 ohm line using AWG14 wires
to have an unloaded Q around 350.

You can achieve that loose coupling by calling the center of the short
across the end "ground" and tapping up just one or two percent of the
length of the line from that for the "hot" connection. Or you can
couple in with a loop, say of a diameter about equal to the line
spacing, held next to the shorted end of the line. "Reference Data for
Radio Engineers" shows various coupling schemes in the "Transmission
Lines" chapter. As long as you keep the coupling light (to keep the
loaded Q high), and are consistent in the way you arrange things, you
should be able to measure the resonance to within a fraction of a
percent repeatability, if not absolute accuracy. Relative measurements
should be all you need in this case. And I assume it's obvious that
though the resonant line will not have exactly the same VF as an
antenna, as you increase the wire spacing, it should approach the same
effect as you'll see in the antenna. Then compare with the formulas
Reg posted, and if you see significant differences, try to resolve
what's causing them.

By the way, one place where the "velocity factor" effect might be
noticable is in parasitic elements of an array in which you're trying
to achieve maximum gain. The element tuning will affect phasing among
the elements and therefore gain. If the design is narrow-band,
high-gain, you might actually notice some effect from the insulation.

Cheers,
Tom

Tom

Someone might refer to you as a "spoiled brat" regarding your interest in
the details. I consider you to be a carefull thinker.

Jerry


"K7ITM" wrote in message
oups.com...
Brian Kelly wrote:

Which leads into another head-scratcher I've had in the past. I've had
a bad time coupling a GDO to quad elements because it takes a couple
turns of wire near the GDO coil to get enough coupling between the
quad
element and the GDO. Which in turn means that I've shortened the
element length and the result is wrong.

What's your suggestion on a method to accurately measure the resonant
frequencies of the quarter-wave lines in this exercise?

OK, I admit to being a spoiled brat when it comes to making
measurements like this. The easiest for me would be to set up a
network analyzer, with the analyzer's source and receiver each coupled
lightly to the line. But there are other ways. You may not even need
a signal generator. You could loosely couple a receiver to the line,
and loosely couple an antenna to it on the other side, and as you tune
the receiver across the resonance, you should notice a sharp peak in
atmospheric noise. With a loaded Q of a few hundred, the peak at 10MHz
would be a very few tens of kHz wide. You could also build an
oscillator which uses the tuned line as the frequency-determining
element, and just count the frequency of the oscillator. A simple
version of a network analyzer could be done by lightly coupling an RF
generator into the line, and putting an RF detector across the line a
small distance up from the shorted end. I'd use either a simple diode
detector, which can have pretty high input impedance, or one of the
Linear Technology or Analog devices RF detectors, but since those are
lower impedance, tap them down very far on the line. -- I'd expect a
10MHz quarter-wave resonator made from 600 ohm line using AWG14 wires
to have an unloaded Q around 350.

You can achieve that loose coupling by calling the center of the short
across the end "ground" and tapping up just one or two percent of the
length of the line from that for the "hot" connection. Or you can
couple in with a loop, say of a diameter about equal to the line
spacing, held next to the shorted end of the line. "Reference Data for
Radio Engineers" shows various coupling schemes in the "Transmission
Lines" chapter. As long as you keep the coupling light (to keep the
loaded Q high), and are consistent in the way you arrange things, you
should be able to measure the resonance to within a fraction of a
percent repeatability, if not absolute accuracy. Relative measurements
should be all you need in this case. And I assume it's obvious that
though the resonant line will not have exactly the same VF as an
antenna, as you increase the wire spacing, it should approach the same
effect as you'll see in the antenna. Then compare with the formulas
Reg posted, and if you see significant differences, try to resolve
what's causing them.

By the way, one place where the "velocity factor" effect might be
noticable is in parasitic elements of an array in which you're trying
to achieve maximum gain. The element tuning will affect phasing among
the elements and therefore gain. If the design is narrow-band,
high-gain, you might actually notice some effect from the insulation.

Cheers,
Tom