16 el coaxial colinear (2.4ghz) question

I built this antenna expecting better performance from my WiFi
setup. Stock antennas are 1/2 wave verticals at the back of the
router. New antenna is a 16 element coaxial collinear @32' agl.
Total transmission system losses work out to 5.299db @2.412
ghz (includes feedline + N connectors).

So, why does this antenna only have about (rough measurement)
1-3 db gain over the stock antenna? I was meticulous in my work,
measuring the antenna parts with a digital caliper and checking
the connections with a DVM along the way.

Maybe my design is a fault? I built the antenna just like this:

http://wireless.gumph.org/content/4/...s-antenna.html

Except in my version, I've got 16 elements made from RG58. I also
made 2 four element antennas for the back of my thinkpad and they
seem to be working about twice as far as the antennas they replaced
(inverted V's).

Any ideas?

73's de Ken KG0WX - Kadiddlehopper #11808,
Flying Pigs #-1055, Grid EM17io,
Elecraft K2 #4913, XG2, 4SQRP Tenna Dipper,
Heath GD-1B, MP-1(X)antenna

I salute anyone who has the patience to build a 16 element coaxial
collinear.

Tr to visualize the radiation pattern of the antenna as a donut. The
more elements you add to a collinear, the more the donut gets squished.
Because your antenna has altitude, you may be sitting beneath the idea
location of the donut. There are tricks to get downtilt in verticals.

You also need to consider the loss in the feedline.

BTW, I've seen coax colinear designs that aren't totally made out of
1/2 sections. The variation may be to get a better impedance match.


Ken Bessler wrote:
> I built this antenna expecting better performance from my WiFi
> setup. Stock antennas are 1/2 wave verticals at the back of the
> router. New antenna is a 16 element coaxial collinear @32' agl.
> Total transmission system losses work out to 5.299db @2.412
> ghz (includes feedline + N connectors).
>
> So, why does this antenna only have about (rough measurement)
> 1-3 db gain over the stock antenna? I was meticulous in my work,
> measuring the antenna parts with a digital caliper and checking
> the connections with a DVM along the way.
>
> Maybe my design is a fault? I built the antenna just like this:
>
> http://wireless.gumph.org/content/4/...s-antenna.html
>
> Except in my version, I've got 16 elements made from RG58. I also
> made 2 four element antennas for the back of my thinkpad and they
> seem to be working about twice as far as the antennas they replaced
> (inverted V's).
>
> Any ideas?
>
> 73's de Ken KG0WX - Kadiddlehopper #11808,
> Flying Pigs #-1055, Grid EM17io,
> Elecraft K2 #4913, XG2, 4SQRP Tenna Dipper,
> Heath GD-1B, MP-1(X)antenna

"Ken Bessler" <(E-Mail Removed)> hath wroth:

>I built this antenna expecting better performance from my WiFi
>setup. Stock antennas are 1/2 wave verticals at the back of the
>router. New antenna is a 16 element coaxial collinear @32' agl.
>Total transmission system losses work out to 5.299db @2.412
>ghz (includes feedline + N connectors).

I wish my measurements were accurated to so many signifigant figures.

>So, why does this antenna only have about (rough measurement)
>1-3 db gain over the stock antenna? I was meticulous in my work,
>measuring the antenna parts with a digital caliper and checking
>the connections with a DVM along the way.

Show us a photo of your construction.

>Maybe my design is a fault? I built the antenna just like this:
>http://wireless.gumph.org/content/4/...s-antenna.html

Yep. Just about everything about that design is wrong. Compare it
with a proper construction article:
http://www.nodomainname.co.uk/Omnico...4collinear.htm
Some differences are:
1. Each sections is 0.5 wavelength (electrical), not 0.25 wavelength.
Only the very top element is 0.25 long.
2. Low loss LMR-400 is used, not high loss RG-58 junk coax.
3. The base section includes a dual 1/4 wave sleeve balun to prevent
the coax from radiating and to keep the radiation angle from going up.

>Except in my version, I've got 16 elements made from RG58. I also
>made 2 four element antennas for the back of my thinkpad and they
>seem to be working about twice as far as the antennas they replaced
>(inverted V's).
>
>Any ideas?

However, even if you build such an antenna, there are problems.
1. The alternating coax vertical collinear is twice as long as
necessary for a given gain. Every other 0.5 wave section does not
radiate and only acts as a phasing section. Since only every other
element radiates, it ends up twice as long. A better design would be
to use inductors or stubs for phasing, resulting in a smaller antenna.
Something like:
http://www.nodomainname.co.uk/Omni_lowpower/
http://mobileaccess.de/wlan/download...0-all-inst.pdf
http://martybugs.net/wireless/collinear.cgi

Incidentally, the 0.5 wave section design is not the only way to build
a vertical collinear. See:
http://www.cebik.com/vhf/cc.html

2. High gain vertical collinear suck because most of the RF comes out
from the first (bottom) section. The way it *ROUGHLY* works is half
the power radiates from the first 0.5 wave section. Half of what's
left radiates from the next section. Half of what's left radiates
from the next section, ad nausium. By the time you get to the top,
there's very little radiation. In addition, the use of coax cable
adds some attenuation. For fixed tower use, the addition of ground
radials often raise the radiation angle above the horizon.


--
Jeff Liebermann (E-Mail Removed)
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

Jeff Liebermann <(E-Mail Removed)> hath wroth:

>Some differences are:
>1. Each sections is 0.5 wavelength (electrical), not 0.25 wavelength.
>Only the very top element is 0.25 long.

Oops, I goofed. The RG-58 design is 0.5 wavelength (electrical)
sections.

This version is a bit better:
http://wireless.gumph.org/articles/homemadeomni.html
--
Jeff Liebermann (E-Mail Removed)
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

Jeff Liebermann wrote:
> "Ken Bessler" <(E-Mail Removed)> hath wroth:
>
> >I built this antenna expecting better performance from my WiFi
> >setup. Stock antennas are 1/2 wave verticals at the back of the
> >router. New antenna is a 16 element coaxial collinear @32' agl.
> >Total transmission system losses work out to 5.299db @2.412
> >ghz (includes feedline + N connectors).
>
> I wish my measurements were accurated to so many signifigant figures.
>
> >So, why does this antenna only have about (rough measurement)
> >1-3 db gain over the stock antenna? I was meticulous in my work,
> >measuring the antenna parts with a digital caliper and checking
> >the connections with a DVM along the way.
>
> Show us a photo of your construction.
>
> >Maybe my design is a fault? I built the antenna just like this:
> >http://wireless.gumph.org/content/4/...s-antenna.html
>
> Yep. Just about everything about that design is wrong. Compare it
> with a proper construction article:
> http://www.nodomainname.co.uk/Omnico...4collinear.htm
> Some differences are:
> 1. Each sections is 0.5 wavelength (electrical), not 0.25 wavelength.
> Only the very top element is 0.25 long.
> 2. Low loss LMR-400 is used, not high loss RG-58 junk coax.
> 3. The base section includes a dual 1/4 wave sleeve balun to prevent
> the coax from radiating and to keep the radiation angle from going up.
>
> >Except in my version, I've got 16 elements made from RG58. I also
> >made 2 four element antennas for the back of my thinkpad and they
> >seem to be working about twice as far as the antennas they replaced
> >(inverted V's).
> >
> >Any ideas?
>
> However, even if you build such an antenna, there are problems.
> 1. The alternating coax vertical collinear is twice as long as
> necessary for a given gain. Every other 0.5 wave section does not
> radiate and only acts as a phasing section. Since only every other
> element radiates, it ends up twice as long. A better design would be
> to use inductors or stubs for phasing, resulting in a smaller antenna.
> Something like:
> http://www.nodomainname.co.uk/Omni_lowpower/
> http://mobileaccess.de/wlan/download...0-all-inst.pdf
> http://martybugs.net/wireless/collinear.cgi
>
> Incidentally, the 0.5 wave section design is not the only way to build
> a vertical collinear. See:
> http://www.cebik.com/vhf/cc.html
>
> 2. High gain vertical collinear suck because most of the RF comes out
> from the first (bottom) section. The way it *ROUGHLY* works is half
> the power radiates from the first 0.5 wave section. Half of what's
> left radiates from the next section. Half of what's left radiates
> from the next section, ad nausium. By the time you get to the top,
> there's very little radiation. In addition, the use of coax cable
> adds some attenuation. For fixed tower use, the addition of ground
> radials often raise the radiation angle above the horizon.
>
>
> --
> Jeff Liebermann (E-Mail Removed)
> 150 Felker St #D http://www.LearnByDestroying.com
> Santa Cruz CA 95060 http://802.11junk.com
> Skype: JeffLiebermann AE6KS 831-336-2558

http://www.nodomainname.co.uk/Omnico...4collinear.htm
was the design I had in my head but couldn't find. However, I still
think it needs to be examined if he is "sitting" under the donut. I
looked at the amount of work that went into this design and figured it
would be better to buy a commercial antenna. Pretty much the biquad is
all I would suggest for homebrew after looking at most of the antennas
on the net.

I've built plenty of VHF and UHF super J-poles and wondered two things.
One, why can't the super J-pole be adapted to wifi. Is it the lack of
small diameter copper pipe? Two, the coaxial collinear, being something
like a half wave, still would benefit from some sort of ground plane,
yet this is never mentioned in the designs.

(E-Mail Removed) hath wroth:

>http://www.nodomainname.co.uk/Omnico...4collinear.htm
>was the design I had in my head but couldn't find. However, I still
>think it needs to be examined if he is "sitting" under the donut. I
>looked at the amount of work that went into this design and figured it
>would be better to buy a commercial antenna.

I previously itemized why such antennas are not particularly
wonderful. However, I have built them in the distant past and added a
few mistakes of my own. I built a 444Mhz version that was almost 20ft
long. Suspended from a tree, it seemed to work well enough. VSWR was
low and it did have some gain (which I couldn't measure at the time).
Satisfied, I shoved it into a 1" dia PVC pipe and filled it with
urethane fence post foam compound. Suddenly, the antenna didn't work
as well. The PVC and foam combination have what I guess to be a
velocity factor of about 0.95. That moved the resonant frequency
down, raised the VSWR, and ruined the antenna. Instead of a 444MHz
antenna, I had a 430MHz antenna. I used a Dremel tool to slice the
pipe open, extracted the antenna, and used a fiberglass tube instead.
That worked.

>Pretty much the biquad is
>all I would suggest for homebrew after looking at most of the antennas
>on the net.

Agreed. I like biquads. They're easy to build, fairly non-critical,
and offer reasonable 9-10dB gain. My favorite biquad:
http://www.vallstedt-networks.de/?Fotogalerien/quad2
Some analysis:
http://802.11junk.com/jeffl/antennas/Biquad/

>I've built plenty of VHF and UHF super J-poles and wondered two things.
>One, why can't the super J-pole be adapted to wifi. Is it the lack of
>small diameter copper pipe?

A j-pole antenna (inverted Zepp) is a fairly lousy antenna. It's
major advantage is that the entire antenna is at DC ground thus
avoiding messy insulators (and nasty sparks for Zeppelin gas bags).
It's also easy to build out of pipe. On the negative side, it has a
fairly narrow bandwidth and a high angle of radiation if mounted
upwards instead of hanging downwards as in the Zeppelin.

A J-pole would probably scale nicely to 2.4Ghz. Fat copper pipe is
not needed. The reason one uses fat pipe for 50MHz and below is to
increase the bandwidth of the antenna. The larger the OD, the wider
the bandwidth. At 2.4GHz, the equivalent diameter would probably be a
#10 awg solid wire. The feed point gamma match could be made from the
traditional wire wrap, but methinks a very small trimmer or fabricated
tubular capacitor would be better.

>Two, the coaxial collinear, being something
>like a half wave, still would benefit from some sort of ground plane,
>yet this is never mentioned in the designs.

Benefit how? The purpose of the ground radials is only to stabilize
the impedance of the antenna. Without the ground radials, the antenna
will be seriously influenced by the mounting structure. With the
ground plane, the impedance of the antenna is stabilized by the
"artificial" ground.

In the case of 2.4Ghz antennas, one only needs to be a few wavelengths
above the mounting structure to be considered "far" away. Mounting
will have an influence, but not as severely as at lower frequencies,
where the mounting structure becomes part of the antenna. The dual
sleeve decoupler at the base provides the necessary isolation and
ground plane.

See:
http://802.11junk.com/jeffl/antennas...s/tecom02.html
http://802.11junk.com/jeffl/antennas...s/tecom01.html
Note the ground "radials" near the connector end. They don't need to
be sticking out horizontally and can be folded back as in a coaxial
antenna.

--
Jeff Liebermann (E-Mail Removed)
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

Jeff Liebermann <(E-Mail Removed)> hath wroth:

>I wish my measurements were accurated to so many signifigant figures.

Speaking of measurments, ever wonder how accurately one has to cut
such an antenna? Grinding the numbers...

At the low end of the 2.4Ghz band, one wavelength is:
3*10^10 cm/sec / 2400*10^6 cycles/sec = 125 mm
At the high end, it's:
3*10^10 cm/sec / 2483.5*10^6 cycles/sec = 120.8 mm

The tubes are a half wavelength long, times the 0.66 velocity factor
resulting in 41.6 mm and 40.2 mm for the band edges.

If you aim for the middle of the band, with a nominal cut length of
40.9 mm, and you're off plus or minus 0.7mm in cut length, you're
outside the 2.4GHz band. The antenna will still function, but not
very well.

That level of accuracy is impossible to achieve with coaxial braid,
which is why the article at:
http://www.nodomainname.co.uk/Omnico...4collinear.htm
uses brass tubing.

--
Jeff Liebermann (E-Mail Removed)
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

Jeff Liebermann wrote:
> (E-Mail Removed) hath wroth:
>
> >http://www.nodomainname.co.uk/Omnico...4collinear.htm
> >was the design I had in my head but couldn't find. However, I still
> >think it needs to be examined if he is "sitting" under the donut. I
> >looked at the amount of work that went into this design and figured it
> >would be better to buy a commercial antenna.
>
> I previously itemized why such antennas are not particularly
> wonderful. However, I have built them in the distant past and added a
> few mistakes of my own. I built a 444Mhz version that was almost 20ft
> long. Suspended from a tree, it seemed to work well enough. VSWR was
> low and it did have some gain (which I couldn't measure at the time).
> Satisfied, I shoved it into a 1" dia PVC pipe and filled it with
> urethane fence post foam compound. Suddenly, the antenna didn't work
> as well. The PVC and foam combination have what I guess to be a
> velocity factor of about 0.95. That moved the resonant frequency
> down, raised the VSWR, and ruined the antenna. Instead of a 444MHz
> antenna, I had a 430MHz antenna. I used a Dremel tool to slice the
> pipe open, extracted the antenna, and used a fiberglass tube instead.
> That worked.
>
> >Pretty much the biquad is
> >all I would suggest for homebrew after looking at most of the antennas
> >on the net.
>
> Agreed. I like biquads. They're easy to build, fairly non-critical,
> and offer reasonable 9-10dB gain. My favorite biquad:
> http://www.vallstedt-networks.de/?Fotogalerien/quad2
> Some analysis:
> http://802.11junk.com/jeffl/antennas/Biquad/
>
> >I've built plenty of VHF and UHF super J-poles and wondered two things.
> >One, why can't the super J-pole be adapted to wifi. Is it the lack of
> >small diameter copper pipe?
>
> A j-pole antenna (inverted Zepp) is a fairly lousy antenna. It's
> major advantage is that the entire antenna is at DC ground thus
> avoiding messy insulators (and nasty sparks for Zeppelin gas bags).
> It's also easy to build out of pipe. On the negative side, it has a
> fairly narrow bandwidth and a high angle of radiation if mounted
> upwards instead of hanging downwards as in the Zeppelin.

Ah, but the super j-pole is another story. The collinear again tends to
squash the donut. People love or hate j-poles.

>
> A J-pole would probably scale nicely to 2.4Ghz. Fat copper pipe is
> not needed. The reason one uses fat pipe for 50MHz and below is to
> increase the bandwidth of the antenna. The larger the OD, the wider
> the bandwidth. At 2.4GHz, the equivalent diameter would probably be a
> #10 awg solid wire. The feed point gamma match could be made from the
> traditional wire wrap, but methinks a very small trimmer or fabricated
> tubular capacitor would be better.

My point is the construction at 2.4Ghz is actually hampered by the
small dimensions. Half inch copper pipe would make the phasing element
in a super J-pole somewhat nebulous.
>
> >Two, the coaxial collinear, being something
> >like a half wave, still would benefit from some sort of ground plane,
> >yet this is never mentioned in the designs.
>
> Benefit how? The purpose of the ground radials is only to stabilize
> the impedance of the antenna. Without the ground radials, the antenna
> will be seriously influenced by the mounting structure. With the
> ground plane, the impedance of the antenna is stabilized by the
> "artificial" ground.
>
> In the case of 2.4Ghz antennas, one only needs to be a few wavelengths
> above the mounting structure to be considered "far" away. Mounting
> will have an influence, but not as severely as at lower frequencies,
> where the mounting structure becomes part of the antenna. The dual
> sleeve decoupler at the base provides the necessary isolation and
> ground plane.
>
> See:
> http://802.11junk.com/jeffl/antennas...s/tecom02.html
> http://802.11junk.com/jeffl/antennas...s/tecom01.html
> Note the ground "radials" near the connector end. They don't need to
> be sticking out horizontally and can be folded back as in a coaxial
> antenna.
>
> --
> Jeff Liebermann (E-Mail Removed)
> 150 Felker St #D http://www.LearnByDestroying.com
> Santa Cruz CA 95060 http://802.11junk.com
> Skype: JeffLiebermann AE6KS 831-336-2558

Jeff, you are so mistaken. When antennas are phased, there needs to be a stacking distance between them. Trying to get them closer to reduce the distance is not sensible. As such, the coaxial colinear is not twice as long as it needs to be.

Again, you are so very wrong. Where do you get this factor of a half (0.5) from? Was it found via computer simulations, a peer reviewed journal, some dubious web site. Did you invent it yourself?

If it was true (which it is not), there would be no practical point in having more than 3 sections, as 7/8 of the power would have been radiated after 3 sections, leaving it possible to only gain 1/8th (0.6 dB) more if you radiated the other 1/8th.

I admit you said *ROUGHLY*, but you really don't understand the antenna.

I don't clam to understand this antenna fully, and I've spent a lot of time working on it.

* Consulting the original paper by Wheeler
* Looking at the peer reviewed journals - mainly in the publications from the IEEE Anteanns and Propagation Society.
* Computer simulations with HFSS - software costing well over $100000. (HFSS uses the finite element method and the differential form of Maxwell's equations).
* Comparison with others who have used expensive commerical software based on the finite difference time-domain (FDTD) approach.

There are plenty of idiotic papers on the web about the coaxial colinear, like "Build A 9 dB, 70cm, Collinear Antenna From Coax" by N1HFX.

The author claims 9 dB gain at 70 cm from 8 sections. N1HFX does not state if his 9 dBbi or 9 dBd, but my belief is he probably thinks it is 9 dBd. I suspect his logic is this

* A half wave has a gain of 0 dBd
* Phasing two of them gives a 3 dB increase in gain (so 3 dB)
* Doubling the elements to 4 gives another 3dB increase in gain (now 6 dB)
* Doubling the elements to 8 gives another 3 dB in gain (now 9 dB).

(I have attempted to contact N1HFX, but he has failed to reply to my emails)

Here is a litle "thought experiment" for you. Let's assume you wanted to make an 8 section coaxial colinear for top band (160 m). Each half-wave section, if made of air spaced line would be 80 m long, so 640 m long in total. A bit too long really. Now dielectrics with permittivities of over 1000000 do exist, but unfortunately they are not very good for RF. But let's assume one could get a perfectly lossless dielectric wtih a permittivity of 1000000 and made the coax from that. The velocity factor of the coax would be 1/sqrt(1000000)=0.001. So we could shrink the 8 sections of the top-band antenna to 640 mm long. It should not be hard to see there is a relationship between gain and permittivity of the material used in the cable.

What I do know about this antenna I am not at liberty to share wtih you, as the information is confidential, but I felt I had to correct you. If you do want to learn about it, I suggest you forget all the amateur radio publications, forget all the web pages from hams telling you how to make them, and concentrate on looking at the peer reviewed scientific journals.

Dr. David Kirkby BSc MSc PhD CEng MIET
(G8WRB)