On Apr 8, 4:40 pm, Wimpie wrote:
On 9 abr, 01:26, K7ITM wrote:



Hi Wim (and lurkers),

On Apr 8, 1:56 pm, Wimpie wrote:

On 8 abr, 18:52, K7ITM wrote:

...

I just ran EZNEC on a frequency-scaled version of the 14MHz 5 element
Yagi included in the sample files, with the D.E. slightly shortened to
allow a decent hairpin match to 200 ohms at the design center
frequency.

Did you also scale the thickness of the elements?

Yes, EZNEC lets you scale everything in the same proportion in one
quick operation.

I did a frequency sweep, 49 to 51 MHz, in 0.25MHz steps.

...

I don't know whether this will give sufficient BW improvement for the
Yagi as the Q is also determined by the reflector en directors. I am
looking forward to your simulation results.

Yes, of course. If the antenna itself is not wide-band, there is
rather little you can do about it. It may be possible with a more-or-
less complicated network to make the SWR bandwidth somewhat greater
(assuming still low loss--if you have a lossy network, you can make
the SWR bandwidth very large -- ;-) But making the SWR bandwidth
large does not make the antenna's gain bandwidth large. That is, for
a given set of directors and reflectors, the pattern will change with
frequency without much regard for what you do to get power into the
antenna.

Thanks for the encouragement about running some simulations. I'll
report results as I have a chance, probably this evening (US Pacific
coast time).

Cheers,
Tom

Hello Tom,

I do not have experience with EZNEC, so I didn't know that it also
scales element diameter.

Fully agree with regards to VSWR BW, gain BW and wide band dummy
loads....

Here it is UTC+2, so it is time for me te visit my bed for some hours.
Tommorow VAT administration is waiting.

Best regards,

Wim

Hi Wim and lurkers,

Yes, EZNEC has some nice features. I exercised a couple of them
tonight, doing frequency sweeps with inductive and capacitive
matching. I did cheat: since I do not know the dimensions of the
hairpin match, I elected to just use a pure lumped inductance. I
suppose the error compared with a transmission line stub (hairpin)
won't be great. I also used for the first time ASCII file import for
the wire description, making use of Excel to generate the wires. (Why
did it take me so long?? ;-)

Somewhat to my surprise, I found that the SWR bandwidth of the
shortened D.E. (driven element) loaded with a shunt inductance and
that of the lengthened D.E. loaded with a shunt capacitance was not so
different.

The antenna I picked to model was one in the Orr and Cowan book, "The
Beam Antenna Book." It's a 6-element 1.20 wave long design, with 0.2
waves D.E. to reflector, and 0.25 waves D.E. to first director and
also between each adjacent pair of directors. For 6 meters it is a
little shorter than Tony's antenna, but not very much shorter. But
the elements (per Orr and Cowan's numbers) are 2 inches (about 50mm)
diameter. The large diameter almost certainly contributes to the
fairly broad bandwidth. All models below are done in freespace.

For the shortened D.E.: D.E. length = 95.4 inches. Feedpoint
impedance at 50.1MHz is 16.32-j54.63 ohms. Gain varies from 11.78dBi
at 49.5MHz to 11.98dBi at 50.1 to 11.68 at 50.5. f/b varies uniformly
from 19.43dB at 49.5 to 8.21dB at 50.5. Matched at 50.1MHz to 200
ohms with 189.4nH shunt at the feedpoint yields a 1.4:1 SWR bandwidth
of about 49.89MHz to 50.28MHz.

For the lengthened D.E.: length = 115.8 inches; impedance at 50.1 =
29.28+j70.7 ohms. Gain: , 12.05max, . f/b shows
same uniform degradation with increased freq; to
. 38.35pF shunt across feedpoint yields 1.4:1 SWR BW of
about 49.89 to 50.34MHz, just slightly greater than the shortened one.

(Neither of these includes frequency effects of the 4:1 balun to get
back to 50 ohms...)

I also tried a much-lengthened D.E., to get it up to 50+j204.1 ohms.
Then a series 15.57pF gives a 50 ohm match. Gain is similar to above,
11.83dBi -- 12.11dBi -- 11.93dBi, and f/b is also similar, 22.82 --
8.83. I notice both the gain and the f/b improve slightly as the D.E.
is lengthened, though it's small enough that you'd never notice it--it
would be swamped out by other effects. The 1.4:1 SWR BW for this
configuration is about 49.85MHz to 50.32MHz; this of course against a
50 ohm reference which is different from the first two examples above
which are against a 200 ohm reference.

OK--I think I got all the critical stuff in there...

Cheers,
Tom

K7ITM wrote in
:

....
Yes, EZNEC has some nice features. I exercised a couple of them
tonight, doing frequency sweeps with inductive and capacitive
matching. I did cheat: since I do not know the dimensions of the
hairpin match, I elected to just use a pure lumped inductance. I
suppose the error compared with a transmission line stub (hairpin)
won't be great. I also used for the first time ASCII file import for

It probably isn't.

However, you could get an idea from my Two Wire Line Loss Calculator
(http://www.vk1od.net/tl/twllc.htm). It looks like a hairpin made of 4mm
dia aluminium 50mm spacing and 150mm in length give an impedance of 0.02
+j61... so the Q is probably mainly determined by the end connections.

a o/c stub made of 4mm dia aluminium 50mm spacing and 1300mm in length
give an impedance of 0.07-j80... so, it is not quite as good
electrically, it is unweildly and end connection resistance will probably
still be significant. This is no doubt why people use a hairpin in
preference to an o/c stub!


Owen

On Apr 9, 2:08 am, Owen Duffy wrote:
K7ITM wrote :

...

Yes, EZNEC has some nice features. I exercised a couple of them
tonight, doing frequency sweeps with inductive and capacitive
matching. I did cheat: since I do not know the dimensions of the
hairpin match, I elected to just use a pure lumped inductance. I
suppose the error compared with a transmission line stub (hairpin)
won't be great. I also used for the first time ASCII file import for

It probably isn't.

However, you could get an idea from my Two Wire Line Loss Calculator
(http://www.vk1od.net/tl/twllc.htm). It looks like a hairpin made of 4mm
dia aluminium 50mm spacing and 150mm in length give an impedance of 0.02
+j61... so the Q is probably mainly determined by the end connections.

a o/c stub made of 4mm dia aluminium 50mm spacing and 1300mm in length
give an impedance of 0.07-j80... so, it is not quite as good
electrically, it is unweildly and end connection resistance will probably
still be significant. This is no doubt why people use a hairpin in
preference to an o/c stub!

Owen

Thanks, Owen. Actually, the error I was thinking of was not the Q,
since both a good hairpin and a good helical coil will have Qu's very
much higher than the loaded Q of the matching network -- but rather of
the slope of reactance versus frequency, since the hairpin is a
transmission line and will presumably show a sharp resonance at a
different frequency from the coil's self resonance. But that error
should also be small--negligible as Wim notes.

Cheers,
Tom

K7ITM wrote in
:

On Apr 9, 2:08 am, Owen Duffy wrote:
K7ITM wrote

om:

...

Yes, EZNEC has some nice features. I exercised a couple of them
tonight, doing frequency sweeps with inductive and capacitive
matching. I did cheat: since I do not know the dimensions of the
hairpin match, I elected to just use a pure lumped inductance. I
suppose the error compared with a transmission line stub (hairpin)
won't be great. I also used for the first time ASCII file import
for

It probably isn't.

However, you could get an idea from my Two Wire Line Loss Calculator
(http://www.vk1od.net/tl/twllc.htm). It looks like a hairpin made of
4mm dia aluminium 50mm spacing and 150mm in length give an impedance
of 0.02 +j61... so the Q is probably mainly determined by the end
connections.

a o/c stub made of 4mm dia aluminium 50mm spacing and 1300mm in
length give an impedance of 0.07-j80... so, it is not quite as good
electrically, it is unweildly and end connection resistance will
probably still be significant. This is no doubt why people use a
hairpin in preference to an o/c stub!

Owen

Thanks, Owen. Actually, the error I was thinking of was not the Q,
since both a good hairpin and a good helical coil will have Qu's very
much higher than the loaded Q of the matching network -- but rather of
the slope of reactance versus frequency, since the hairpin is a
transmission line and will presumably show a sharp resonance at a
different frequency from the coil's self resonance. But that error
should also be small--negligible as Wim notes.

Hi Tom,

Warning bells sound to me when applications call for reactors fabricated
from TL sections. Not to say that are always bad, but they aren't always
good, and they bear examination.

I think that may have been in Tony's mind over the matching network.

In the case of the hairpin, fabricated from substantial material, it
looks good in this application. Because the line section is so short,
inductive reactance is almost linearly proportional to length, and self
resonance of the line section itself is nearly a decade higher.

My experience with one of the HyGain (now MFJ) 2m Yagis was that VSWR was
stable with weather and over time. I had taken some care to exclude air +
water from key connections, in fact I replaced all the fastners used for
electrical connection with SS when the beam was less than a year old
because of corrosion. Tweny years later the remaining fastners used only
for mechanical retention required replacement due to corrosion. But,
after the first rework, the matching system seemed robust (bird proof)
and reliable.

I have used them since on home made antennas with good success.

Owen

On Apr 9, 2:59 pm, Owen Duffy wrote:
...
Hi Tom,

Warning bells sound to me when applications call for reactors fabricated
from TL sections. Not to say that are always bad, but they aren't always
good, and they bear examination.

....

:-) Yes, indeed. I just this afternoon fabricated a 500MHz LPF that
I wanted to give about 40dB attenuation up to several GHz. My design
was a 5th order elliptical, so I had two parallel tanks for the series
paths and three shunt capacitors. Because the coils (10nH and 15nH)
are so small, I though about using a shorted stub instead, but then
realized that my shorted stub would look like a short when it was 1/2
wave long, not a good thing to have as a series element at a frequency
you want to block. Probably would have worked OK since the two stubs
were different and wouldn't have the same resonance frequency; and by
the time you reached that freq., the shunt caps would be pretty
effective on their own. Anyway, it works fine with tiny coils. Thank
goodness for good microscopes.

Cheers,
Tom

On 9 abr, 08:40, K7ITM wrote:
On Apr 8, 4:40 pm, Wimpie wrote:



On 9 abr, 01:26, K7ITM wrote:

Hi Wim (and lurkers),

On Apr 8, 1:56 pm, Wimpie wrote:

On 8 abr, 18:52, K7ITM wrote:

...

I just ran EZNEC on a frequency-scaled version of the 14MHz 5 element
Yagi included in the sample files, with the D.E. slightly shortened to
allow a decent hairpin match to 200 ohms at the design center
frequency.

Did you also scale the thickness of the elements?

Yes, EZNEC lets you scale everything in the same proportion in one
quick operation.

I did a frequency sweep, 49 to 51 MHz, in 0.25MHz steps.

...

I don't know whether this will give sufficient BW improvement for the
Yagi as the Q is also determined by the reflector en directors. I am
looking forward to your simulation results.

Yes, of course. If the antenna itself is not wide-band, there is
rather little you can do about it. It may be possible with a more-or-
less complicated network to make the SWR bandwidth somewhat greater
(assuming still low loss--if you have a lossy network, you can make
the SWR bandwidth very large -- ;-) But making the SWR bandwidth
large does not make the antenna's gain bandwidth large. That is, for
a given set of directors and reflectors, the pattern will change with
frequency without much regard for what you do to get power into the
antenna.

Thanks for the encouragement about running some simulations. I'll
report results as I have a chance, probably this evening (US Pacific
coast time).

Cheers,
Tom

Hello Tom,

I do not have experience with EZNEC, so I didn't know that it also
scales element diameter.

Fully agree with regards to VSWR BW, gain BW and wide band dummy
loads....

Here it is UTC+2, so it is time for me te visit my bed for some hours.
Tommorow VAT administration is waiting.

Best regards,

Wim

Hi Wim and lurkers,

Yes, EZNEC has some nice features. I exercised a couple of them
tonight, doing frequency sweeps with inductive and capacitive
matching. I did cheat: since I do not know the dimensions of the
hairpin match, I elected to just use a pure lumped inductance. I
suppose the error compared with a transmission line stub (hairpin)
won't be great. I also used for the first time ASCII file import for
the wire description, making use of Excel to generate the wires. (Why
did it take me so long?? ;-)

Somewhat to my surprise, I found that the SWR bandwidth of the
shortened D.E. (driven element) loaded with a shunt inductance and
that of the lengthened D.E. loaded with a shunt capacitance was not so
different.

The antenna I picked to model was one in the Orr and Cowan book, "The
Beam Antenna Book." It's a 6-element 1.20 wave long design, with 0.2
waves D.E. to reflector, and 0.25 waves D.E. to first director and
also between each adjacent pair of directors. For 6 meters it is a
little shorter than Tony's antenna, but not very much shorter. But
the elements (per Orr and Cowan's numbers) are 2 inches (about 50mm)
diameter. The large diameter almost certainly contributes to the
fairly broad bandwidth. All models below are done in freespace.

For the shortened D.E.: D.E. length = 95.4 inches. Feedpoint
impedance at 50.1MHz is 16.32-j54.63 ohms. Gain varies from 11.78dBi
at 49.5MHz to 11.98dBi at 50.1 to 11.68 at 50.5. f/b varies uniformly
from 19.43dB at 49.5 to 8.21dB at 50.5. Matched at 50.1MHz to 200
ohms with 189.4nH shunt at the feedpoint yields a 1.4:1 SWR bandwidth
of about 49.89MHz to 50.28MHz.

For the lengthened D.E.: length = 115.8 inches; impedance at 50.1 =
29.28+j70.7 ohms. Gain: , 12.05max, . f/b shows
same uniform degradation with increased freq; to
. 38.35pF shunt across feedpoint yields 1.4:1 SWR BW of
about 49.89 to 50.34MHz, just slightly greater than the shortened one.

(Neither of these includes frequency effects of the 4:1 balun to get
back to 50 ohms...)

I also tried a much-lengthened D.E., to get it up to 50+j204.1 ohms.
Then a series 15.57pF gives a 50 ohm match. Gain is similar to above,
11.83dBi -- 12.11dBi -- 11.93dBi, and f/b is also similar, 22.82 --
8.83. I notice both the gain and the f/b improve slightly as the D.E.
is lengthened, though it's small enough that you'd never notice it--it
would be swamped out by other effects. The 1.4:1 SWR BW for this
configuration is about 49.85MHz to 50.32MHz; this of course against a
50 ohm reference which is different from the first two examples above
which are against a 200 ohm reference.

OK--I think I got all the critical stuff in there...

Cheers,
Tom

Hello Tom,

Regarding the "cheating", no problem with that. When length of hairpin
is far below 1/8 lambda, difference with lumped component is
negligible (for this situation).

Owen mentioned that a 150mm hairpin has the required reactance. The
0.02 Ohms loss resistance in the hairpin does not dissipate
significant power (not measurable in practice within the antenna
setup). This also shows that Tony does not have to worry about hairpin
loss, even in his 200 Ohm intermediate impedance.

Your simulation made fully clear that Tony's problem is in the antenna
and not in the matching.

Thanks for sharing your results,

Wim
PA3DJS
www.tetech.nl
please remove abc from the address.

Tom and Wim,

I agree with your concluion that the bandwidth narrowness is due to the antenna
design and not to the matching system.

I have simulated the antenna on EZNEC and the narrow band property of the
antenna is very evident.

I would like to also put the hairpin in the EZNEC model, but I am not very good
at EZNEC and I would not now know to do it.

If you are interested I can send you the EZNEC file via e-mail.

73

Tony I0JX

On Apr 9, 9:02 am, "Antonio Vernucci" wrote:
Tom and Wim,

I agree with your concluion that the bandwidth narrowness is due to the antenna
design and not to the matching system.

I have simulated the antenna on EZNEC and the narrow band property of the
antenna is very evident.

I would like to also put the hairpin in the EZNEC model, but I am not very good
at EZNEC and I would not now know to do it.

If you are interested I can send you the EZNEC file via e-mail.

73

Tony I0JX

Hi Tony,

I would be happy to have a look at the file and add the hairpin if you
wish. But I can also guide you through doing it yourself. It should
be quite easy. Simply click on the " Trans Lines" button, the second
one below Sources in the main window. Specify the first end the
same as the source--perhaps 50% along wire 2, assuming the D.E. is
wire 2. That end will then be in parallel with the source. You can
type "S" in for the other end's wire #, or if you click in the End 2
Wire # box, you should see a list of "open" and "short" and you can
select "short" there. Then enter the length, Z0, velocity factor (1)
and loss. Reverse or normal doesn't matter since only one end is
connected.

In as much as the hairpin actually IS in the field of the antenna, and
couples to it to some degree, there may be a small error in using the
EZNEC transmission line, instead of entering it as wires, but for a
short stub I would not expect that to be a significant issue. I do
know that the 1/4 wave stub used to separate elements of a collinear
does couple to the antenna fields and that coupling is a significant
factor in the performance of the antenna--or at least that is what
R.W.P. King tells us.

And...thanks much for the interesting discussion topic. I learned a
few things from it.

Cheers,
Tom

Hi Tony,

I would be happy to have a look at the file and add the hairpin if you
wish. But I can also guide you through doing it yourself. It should
be quite easy. Simply click on the " Trans Lines" button, the second
one below Sources in the main window. Specify the first end the
same as the source--perhaps 50% along wire 2, assuming the D.E. is
wire 2. That end will then be in parallel with the source. You can
type "S" in for the other end's wire #, or if you click in the End 2
Wire # box, you should see a list of "open" and "short" and you can
select "short" there. Then enter the length, Z0, velocity factor (1)
and loss. Reverse or normal doesn't matter since only one end is
connected.

Thanks on you guidance. My hairpin is a U of aluminum tube connected directly to
the antenna feedpoints. The U center is connected to the boom.

The hairpin tube diameter is about 0.8 cm. The three sides of the U are
approximately 35 cm, 10 cm and 35 cm. I presume that I shall enter 35 cm as the
transmission line length, but how to calculate Z0? Moreover what about the
velocity factor, perhaps 0.98 for a tube in open air?

Would you have a formula at hand?

Thanks and 73

Tony I0JX

On Apr 9, 11:36 am, "Antonio Vernucci" wrote:
Hi Tony,

I would be happy to have a look at the file and add the hairpin if you
wish. But I can also guide you through doing it yourself. It should
be quite easy. Simply click on the " Trans Lines" button, the second
one below Sources in the main window. Specify the first end the
same as the source--perhaps 50% along wire 2, assuming the D.E. is
wire 2. That end will then be in parallel with the source. You can
type "S" in for the other end's wire #, or if you click in the End 2
Wire # box, you should see a list of "open" and "short" and you can
select "short" there. Then enter the length, Z0, velocity factor (1)
and loss. Reverse or normal doesn't matter since only one end is
connected.

Thanks on you guidance. My hairpin is a U of aluminum tube connected directly to
the antenna feedpoints. The U center is connected to the boom.

The hairpin tube diameter is about 0.8 cm. The three sides of the U are
approximately 35 cm, 10 cm and 35 cm. I presume that I shall enter 35 cm as the
transmission line length, but how to calculate Z0? Moreover what about the
velocity factor, perhaps 0.98 for a tube in open air?

Would you have a formula at hand?

Thanks and 73

Tony I0JX

Hi Tony,

Since the spacing is so large (10cm), it may be better to enter the
hairpin as wires in the model. It might be interesting to compare the
results between wires and a transmission line. First the wires:

I assume that the D.E. is one wire in the model you have now. I
suppose it has an odd number of segments, so that the source can go
just in the middle. I'm assuming the hairpin parallel tube segments
are center-to-center 10cm apart. I will assume the D.E. is currently
250cm long, all at some x value, and y extending from -1.25m to
+1.25m, as one "wire" of some diameter. To be able to connect the
hairpin at the right places, we need wire ends at -.05m and +.05m, so
we can change the D.E. wire to be one segment between y=-.05m and +.
05m, and then add 2 wires the same diameter as the D.E., one from
y=-1.25m to -.05m, and one from y=+.05m to +1.25m, and each with about
half as many segments as the original D.E., all at the original x
value. (Or exchange x and y if I have guessed wrong.) So if the
original D.E. had 11 segments, put 5 segments on each of the two new
wires. Now add three new wires to represent the hairpin U. Assuming
the hairpin is oriented perpendicular to the plane of the antenna, it
will be all at the same x value as the D.E. I'll assume the antenna
is at z=0. So the wires will have ends at x,-.05,0;x,-.05,.35 and x,-.
05,.35;x,+.05,.35 and x,+.05,.35;x,+.05,0 -- and all with diameter 8mm
or .008m.

For the transmission line model: the diameter (8mm) and spacing
(100mm center to center) tells me the line impedance is about 380
ohms. It should be OK to assume VF=1.00, if there is no insulation
around the tube. One problem is that the short across the end is not
really a short--it is more an inductance. We could go to the trouble
of calculating the inductance and putting that as a load on the end of
the transmission line instead of just a short, but it should be a good
approximation to just lengthen the line by half the length of the
"short"; I would enter the line length just as 40cm instead of 35. (I
won't be surprised if Owen pops in here with either a confirmation or
a better suggestion!)

Cheers,
Tom