dipole

Antennas are new to me so the questions I ask may not be entirely
sensible. But here goes anyway:

My understanding is that a dipole antenna is in the shape of a T where
the length of each horizontal branch is equal to a quarter wavelength.
So, for 2.4 GHz, each horizontal branch would be about 1.2 inches
long. But I've seen articles and spec sheets that call the stock
"rubber duck" antennas that come with many routers and client radios
dipoles. They don't seem to be in the shape of a T at all. Rather they
just seem to be a length of wire. What gives?

Would it be possible to build my own T-shaped dipole by soldering a
couple of 1.2 inch pieces of wire on to the end of a coax cable? Well,
it would be possible, but would it work ok and why or why not? :)
Would I get more signal if I made the dipole a full wavelength long or
longer? How about many wavelengths like a couple of wires running all
the way across a room?

I know this borders on "turn your electrical wiring into a giant
antenna" but, as a novice, I can't help thinking that I can "grab" a
larger signal by putting up a bigger antenna.

Thanks,
Bruce

> My understanding is that a dipole antenna is in the shape of a T where
> the length of each horizontal branch is equal to a quarter wavelength.
> So, for 2.4 GHz, each horizontal branch would be about 1.2 inches
> long. But I've seen articles and spec sheets that call the stock
> "rubber duck" antennas that come with many routers and client radios
> dipoles. They don't seem to be in the shape of a T at all. Rather they
> just seem to be a length of wire. What gives?

You're not wrong. The di-pole is both the centre conductor and the
outer conductor, one going one way, the other going the otherway.

With the rubber duck antenna's what they've done is essentially wrap the
coax back on itself by using a brass sleeve that fits over the outer
insulation. In effect the T is still there but the tail of the T fits
inside the brass tube.

Here are a couple of examples. Although I don't explicitly show the
dipole but you can see where it enters, a parabolic:-

http://www.nodomainname.co.uk/parabolic/parabolic.htm

and the guts of a PCI card being modified to stick in a laptop.

Underneath the heatshrink sleeving is a piece of brass tube that is
soldered inside to the coax braid, the centre conductor is then allowed
to protude the 1/4 wavelength amount, 31mm.

http://www.nodomainname.co.uk/dwl650g

Sorry, never got round to doing the page for the second one, just raw
pics.

David.

On Mon, 10 Oct 2005 12:56:48 GMT, David Taylor <djtaylor@bigfoot.com>
wrote:

>> My understanding is that a dipole antenna is in the shape of a T where
>> the length of each horizontal branch is equal to a quarter wavelength.
>> So, for 2.4 GHz, each horizontal branch would be about 1.2 inches
>> long. But I've seen articles and spec sheets that call the stock
>> "rubber duck" antennas that come with many routers and client radios
>> dipoles. They don't seem to be in the shape of a T at all. Rather they
>> just seem to be a length of wire. What gives?
>
>You're not wrong. The di-pole is both the centre conductor and the
>outer conductor, one going one way, the other going the otherway.
>
>With the rubber duck antenna's what they've done is essentially wrap the
>coax back on itself by using a brass sleeve that fits over the outer
>insulation. In effect the T is still there but the tail of the T fits
>inside the brass tube.
>
>Here are a couple of examples. Although I don't explicitly show the
>dipole but you can see where it enters, a parabolic:-
>
>http://www.nodomainname.co.uk/parabolic/parabolic.htm
>
>and the guts of a PCI card being modified to stick in a laptop.
>
>Underneath the heatshrink sleeving is a piece of brass tube that is
>soldered inside to the coax braid, the centre conductor is then allowed
>to protude the 1/4 wavelength amount, 31mm.
>
>http://www.nodomainname.co.uk/dwl650g
>
>Sorry, never got round to doing the page for the second one, just raw
>pics.
>
>David.

Oh, I see. Like a T with the top turned 90 degrees. Thanks David.
Still wondering if making the overall dipole length equal to a large
numbe of half wavelengths will help.

Bruce

> Oh, I see. Like a T with the top turned 90 degrees. Thanks David.

Yes, or if you make one half of the horizontal member of the T hollow,
tuck the tail down inside it.

> Still wondering if making the overall dipole length equal to a large
> numbe of half wavelengths will help.

Jeff is much better on antennas, Jeff...

David.

David Taylor <djtaylor@bigfoot.com> wrote:
> Underneath the heatshrink sleeving is a piece of brass tube that is
> soldered inside to the coax braid, the centre conductor is then allowed
> to protude the 1/4 wavelength amount, 31mm.

A couple of years ago, Don Widder proposed the "simplest" antenna, but he
specified 32mm for the sleeve and the exposed conductor.
I haven't figured out how far off that is. I wonder if he's at the edge of
the channel range, or maybe free space instead of coax, by mistake.

Once upon a time From Don Widders:

You can just make an antenna out of the end of the LMR195. Remove about 3
inches of the plastic outer 'jacket' of the coax. Then pull the copper
braid back over the remaining jacket. Get a piece of brass tubing at a
hobby shop that will just slide over the braid and cut a piece of the
tubing to 32 mm. Trim away any excess braid. Cut the center conductor so
that it extends exactly 32 mm from where it exits from the braid.

--
---
Clarence A Dold - Hidden Valley (Lake County) CA USA 38.8,-122.5

> You can just make an antenna out of the end of the LMR195. Remove about 3
> inches of the plastic outer 'jacket' of the coax. Then pull the copper
> braid back over the remaining jacket. Get a piece of brass tubing at a

Yes, that's what I did with the miniPCI card conversion here:-

http://www.nodomainname.co.uk/dwl650g/dwl650g-3.jpg

On Mon, 10 Oct 2005 08:04:47 -0400, bjs555 <aaa@bbb.com> wrote:

>Antennas are new to me so the questions I ask may not be entirely
>sensible. But here goes anyway:
>
>My understanding is that a dipole antenna is in the shape of a T where
>the length of each horizontal branch is equal to a quarter wavelength.

Correct. Actually, it's more like 0.95 * 1/2 wavelength because of
"end dispersion" effects.

>So, for 2.4 GHz, each horizontal branch would be about 1.2 inches
>long.

Watch your accuracy. At 2400Mhz a wavelength is about 125mm. However,
each MHZ is equal to:
125mm / 2400 = 0.052 mm/MHz
The band is 83.5 MHz wide, so your overall tolerance on cutting the
elements is:
83.5 * 0.052mm/MHz = 4.35 mm.
it doesn't take much cutting error to end up with a non-functional
antenna.

>But I've seen articles and spec sheets that call the stock
>"rubber duck" antennas that come with many routers and client radios
>dipoles. They don't seem to be in the shape of a T at all. Rather they
>just seem to be a length of wire. What gives?

Well, there's many ways to make a dipole. In the case of the rubber
ducky, it's called a coaxial antenna or vertical colinear. The
antenna consists of a 1/4 wave driven element and a 1/4 wave sleeve
fitted over the coax cable. If you built your dipole out of tubing
instead of wire, and shove the coax cable feed down one of the tubes,
connected at the center as usual, you would have a coaxial antenna.
It's very cheap and easy to make out of just coax cable. It yields
about 2.1dBi of gain.

However, if you disembowl some of the rubber ducky antennas, you'll
find several different types and additions. The simple dipole is
found in the antennas that are about 100mm long. The 200mm long
antennas have an extra sleeve soldered to the coax cable braid about
1/4 wavelength below the feed point. This is intended to reduce VSWR
and radiation from the coax cable feed. It also improves the gain
slightly.

>Would it be possible to build my own T-shaped dipole by soldering a
>couple of 1.2 inch pieces of wire on to the end of a coax cable?

Yes. It's done all the time in feeds for dish antennas which are
often simple dipoles. The antenna will be 75 ohms instead of 50 ohms
but the mismatch loss in nominal. A balun (balance to unbalanced)
transformer might be a good improvement.
http://www.poynting.co.za/tech_training/tut_balun.shtml

>Well,
>it would be possible, but would it work ok and why or why not? :)

It would work no better than the rubber ducky antennas. however, it
could be used to position the antenna in a better location. Try to
use fairly low loss coax and not junk.

>Would I get more signal if I made the dipole a full wavelength long or
>longer?

No. Without phasing the additional 1/4 wave sections, you would end
up with a pattern that vaguely resembles a cloverleaf at 1 wavelength
per element.

What you should try is building a vertical colinear as in:

a b b b b a
===== =========== ========< >======== =========== =====
| | | | feed | | | |
a | | a a | | a a | | a a | | a
| | | | | | | |
=== === === ===

The long pieces (b) are 1/2 wavelength long. The stubs (a) are 1/4
wavelength long. The short cross pieces at the 1/4 wavelength stubs
(a) are as short as possible. The end pieces are 1/4 wavelength long.
The antenna can best be made by bending copper wire or brass rod. A
sleeve balun might be used at the coax feed point if you want to
squeeze every bit of gain out of the antenna.

It's not a perfect or great antenna but is very easy to build. You
can expand it forever, but there's a catch. Doubling the size of the
antenna only yields 3dB of gain. Most of the radiation comes from the
two sections near the feed point.

>How about many wavelengths like a couple of wires running all
>the way across a room?

A "long wire" antenna has problems with matching to 50 ohms. It
starts to look more like a big inductor than a proper antenna. Getting
RF to the end points of the antenna is difficult. Don't bother.
Bigger antennas don't necessarily imply better antennas.

>I know this borders on "turn your electrical wiring into a giant
>antenna"

That's how I got my start in electronics in the 1950's. There was
this crook in New York that was selling "Turn your House Wiring Into a
Giant TV Antenna" kits. It had a "capacitator" inside and was rather
dangerous with the AC/DC TV's of the era.

>but, as a novice, I can't help thinking that I can "grab" a
>larger signal by putting up a bigger antenna.

Nope. Think phase cancellation. If all the parts of your bigger
antenna received the signal at exactly the same time, and exactly in
phase, then you would have some gain. (Double the antenna gives 3dB
gain). However, just a random wire antenna doesn't do that. Different
parts get the signal at different times. That causes them to randomly
cancel as well as add. I can model anything reasonable with an
antenna modelling program (4NEC2). Methinks you'll find that long
antennas for 2.4GHz don't work at all and are actually worse than a
1/2 wave dipole or rubber ducky. Think phasing.

If you want to build something, I suggest a Biquad antenna. This is
basically two full wave loop antenna in parallel with a reflector.
http://martybugs.net/wireless/biquad/
http://www.weijand.nl/wifi/
http://homepages.ihug.co.nz/~redwood4/


--
Jeff Liebermann jeffl@comix.santa-cruz.ca.us
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

David Taylor <djtaylor@bigfoot.com> wrote:
>> You can just make an antenna out of the end of the LMR195. Remove about 3
>> inches of the plastic outer 'jacket' of the coax. Then pull the copper
>> braid back over the remaining jacket. Get a piece of brass tubing at a

> Yes, that's what I did with the miniPCI card conversion here:-

> http://www.nodomainname.co.uk/dwl650g/dwl650g-3.jpg

Oh, that's what _you_ did. I thought you had surgically removed the rubber
duckie, exposing the existing internals.

Is Don's 32mm, verses your 31mm in this thread, verses your 30.5mm in the
parabolic dipole, important? Why do yours differ by .5mm, or is that so
miniscule as to be meaningless?

--
---
Clarence A Dold - Hidden Valley (Lake County) CA USA 38.8,-122.5

> Oh, that's what _you_ did. I thought you had surgically removed the rubber
> duckie, exposing the existing internals.

Nope, just unsoldered the wire which went to the SMA on the backplate
and thought simplest thing to do was what I did. :)

> Is Don's 32mm, verses your 31mm in this thread, verses your 30.5mm in the
> parabolic dipole, important? Why do yours differ by .5mm, or is that so
> miniscule as to be meaningless?

Well I was rounding up but while I'm sure it's signicant to someone who
is going to measure it, when in a laptop, you can probably cancel out
the difference by turning the laptop around on the table!

David.

Those are great ideas. Looks like I'll be bending wire all week :)

Thanks,
Bruce


On Mon, 10 Oct 2005 09:22:20 -0700, Jeff Liebermann
<jeffl@comix.santa-cruz.ca.us> wrote:

>On Mon, 10 Oct 2005 08:04:47 -0400, bjs555 <aaa@bbb.com> wrote:
>
>>Antennas are new to me so the questions I ask may not be entirely
>>sensible. But here goes anyway:
>>
>>My understanding is that a dipole antenna is in the shape of a T where
>>the length of each horizontal branch is equal to a quarter wavelength.
>
>Correct. Actually, it's more like 0.95 * 1/2 wavelength because of
>"end dispersion" effects.
>
>>So, for 2.4 GHz, each horizontal branch would be about 1.2 inches
>>long.
>
>Watch your accuracy. At 2400Mhz a wavelength is about 125mm. However,
>each MHZ is equal to:
> 125mm / 2400 = 0.052 mm/MHz
>The band is 83.5 MHz wide, so your overall tolerance on cutting the
>elements is:
> 83.5 * 0.052mm/MHz = 4.35 mm.
>it doesn't take much cutting error to end up with a non-functional
>antenna.
>
>>But I've seen articles and spec sheets that call the stock
>>"rubber duck" antennas that come with many routers and client radios
>>dipoles. They don't seem to be in the shape of a T at all. Rather they
>>just seem to be a length of wire. What gives?
>
>Well, there's many ways to make a dipole. In the case of the rubber
>ducky, it's called a coaxial antenna or vertical colinear. The
>antenna consists of a 1/4 wave driven element and a 1/4 wave sleeve
>fitted over the coax cable. If you built your dipole out of tubing
>instead of wire, and shove the coax cable feed down one of the tubes,
>connected at the center as usual, you would have a coaxial antenna.
>It's very cheap and easy to make out of just coax cable. It yields
>about 2.1dBi of gain.
>
>However, if you disembowl some of the rubber ducky antennas, you'll
>find several different types and additions. The simple dipole is
>found in the antennas that are about 100mm long. The 200mm long
>antennas have an extra sleeve soldered to the coax cable braid about
>1/4 wavelength below the feed point. This is intended to reduce VSWR
>and radiation from the coax cable feed. It also improves the gain
>slightly.
>
>>Would it be possible to build my own T-shaped dipole by soldering a
>>couple of 1.2 inch pieces of wire on to the end of a coax cable?
>
>Yes. It's done all the time in feeds for dish antennas which are
>often simple dipoles. The antenna will be 75 ohms instead of 50 ohms
>but the mismatch loss in nominal. A balun (balance to unbalanced)
>transformer might be a good improvement.
> http://www.poynting.co.za/tech_training/tut_balun.shtml
>
>>Well,
>>it would be possible, but would it work ok and why or why not? :)
>
>It would work no better than the rubber ducky antennas. however, it
>could be used to position the antenna in a better location. Try to
>use fairly low loss coax and not junk.
>
>>Would I get more signal if I made the dipole a full wavelength long or
>>longer?
>
>No. Without phasing the additional 1/4 wave sections, you would end
>up with a pattern that vaguely resembles a cloverleaf at 1 wavelength
>per element.
>
>What you should try is building a vertical colinear as in:
>
> a b b b b a
> ===== =========== ========< >======== =========== =====
> | | | | feed | | | |
> a | | a a | | a a | | a a | | a
> | | | | | | | |
> === === === ===
>
>The long pieces (b) are 1/2 wavelength long. The stubs (a) are 1/4
>wavelength long. The short cross pieces at the 1/4 wavelength stubs
>(a) are as short as possible. The end pieces are 1/4 wavelength long.
>The antenna can best be made by bending copper wire or brass rod. A
>sleeve balun might be used at the coax feed point if you want to
>squeeze every bit of gain out of the antenna.
>
>It's not a perfect or great antenna but is very easy to build. You
>can expand it forever, but there's a catch. Doubling the size of the
>antenna only yields 3dB of gain. Most of the radiation comes from the
>two sections near the feed point.
>
>>How about many wavelengths like a couple of wires running all
>>the way across a room?
>
>A "long wire" antenna has problems with matching to 50 ohms. It
>starts to look more like a big inductor than a proper antenna. Getting
>RF to the end points of the antenna is difficult. Don't bother.
>Bigger antennas don't necessarily imply better antennas.
>
>>I know this borders on "turn your electrical wiring into a giant
>>antenna"
>
>That's how I got my start in electronics in the 1950's. There was
>this crook in New York that was selling "Turn your House Wiring Into a
>Giant TV Antenna" kits. It had a "capacitator" inside and was rather
>dangerous with the AC/DC TV's of the era.
>
>>but, as a novice, I can't help thinking that I can "grab" a
>>larger signal by putting up a bigger antenna.
>
>Nope. Think phase cancellation. If all the parts of your bigger
>antenna received the signal at exactly the same time, and exactly in
>phase, then you would have some gain. (Double the antenna gives 3dB
>gain). However, just a random wire antenna doesn't do that. Different
>parts get the signal at different times. That causes them to randomly
>cancel as well as add. I can model anything reasonable with an
>antenna modelling program (4NEC2). Methinks you'll find that long
>antennas for 2.4GHz don't work at all and are actually worse than a
>1/2 wave dipole or rubber ducky. Think phasing.
>
>If you want to build something, I suggest a Biquad antenna. This is
>basically two full wave loop antenna in parallel with a reflector.
> http://martybugs.net/wireless/biquad/
> http://www.weijand.nl/wifi/
> http://homepages.ihug.co.nz/~redwood4/
>
>

Jeff Liebermann <jeffl@comix.santa-cruz.ca.us> wrote:
> On Mon, 10 Oct 2005 08:04:47 -0400, bjs555 <aaa@bbb.com> wrote:

>>My understanding is that a dipole antenna is in the shape of a T where
>>the length of each horizontal branch is equal to a quarter wavelength.

> Correct. Actually, it's more like 0.95 * 1/2 wavelength because of
> "end dispersion" effects.

Isn't he speaking quarter wave and you half wave?
He said quarter wave, and you counter with .95 * half.


Maybe I'm too fussy, measuring with a micrometer, marking with a piece of
chalk, and cutting by biting the wire with my teeth.

My dilema: I see 30.5, 31, and 32 suggested, all of which seem to long,
given a center frequency of 2437MHz and the .95 factor.

It would seem that the optimum would be 29.2mm for Channel 6.
299792.458 / 2437 / 4 * .95

Why, then, do I see suggestions of 30.5 (which would be 100% of 2400MHz,
outside of the band), or 31, or 32mm? Where is the .95 applied?



>>So, for 2.4 GHz, each horizontal branch would be about 1.2 inches
>>long.

> Watch your accuracy. At 2400Mhz a wavelength is about 125mm. However,
> each MHZ is equal to:
> 125mm / 2400 = 0.052 mm/MHz
> The band is 83.5 MHz wide, so your overall tolerance on cutting the
> elements is:
> 83.5 * 0.052mm/MHz = 4.35 mm.
> it doesn't take much cutting error to end up with a non-functional
> antenna.

A margin of error of 4.35mm on an element that is only 31mm long is
obvious enough that I would hope people can get it right.

But that's a full wave, not the margin on each 1/4 wave element. It also
suggests that such an antenna would be perfect for some frequency with the
range of 14 channels, but doesn't say anything about the tolerance for an
acceptable antenna that will be used specifically on channel 6.

If that were applied to each 1/4 wave element, are you saying that the
margin of error on cutting the element is 1mm?

--
---
Clarence A Dold - Hidden Valley (Lake County) CA USA 38.8,-122.5

>Isn't he speaking quarter wave and you half wave?
>He said quarter wave, and you counter with .95 * half.

As I understand it, neglecting the .95, a dipole with each leg 1/4
wave long (1/2 wave total length) has an input impedance of 50 Ohms. A
dipole with each leg 1/2 wave long (1 wave total length) has an input
impedance of 75 Ohms. I'm using 75 Ohm RG-6 cable to keep costs down,
so I'm probably better off with 1/2 wave legs. At least I'd be half
right that way.

On Mon, 10 Oct 2005 20:39:36 +0000 (UTC),
dold@XReXXdipol.usenet.us.com wrote:

>Jeff Liebermann <jeffl@comix.santa-cruz.ca.us> wrote:
>> On Mon, 10 Oct 2005 08:04:47 -0400, bjs555 <aaa@bbb.com> wrote:
>
>>>My understanding is that a dipole antenna is in the shape of a T where
>>>the length of each horizontal branch is equal to a quarter wavelength.
>
>> Correct. Actually, it's more like 0.95 * 1/2 wavelength because of
>> "end dispersion" effects.
>
>Isn't he speaking quarter wave and you half wave?
>He said quarter wave, and you counter with .95 * half.

Yah, sorta. I tend to refer to a dipole antenna as a "half wave
dipole" because of the overall length. The 0.95 is the calculated
fudge factor based on the total length of the dipole. It is NOT a
fixed number but an estimate. I need to know the wire diameter to
calculate the exact value.

>Maybe I'm too fussy, measuring with a micrometer, marking with a piece of
>chalk, and cutting by biting the wire with my teeth.

Ummm... I've seen worse.

>My dilema: I see 30.5, 31, and 32 suggested, all of which seem to long,
>given a center frequency of 2437MHz and the .95 factor.

It's difficult to define the length because I don't know the aspect
radio. Tell me the wire diameter and I'll give you the free space
numbers to about 4 decimal points. I usually cheat and just pound the
numbers into an NEC2 modelling program such as EZNEC or 4NEC2.

A nifty trick to inscreasing the bandwidth of a driven element is to
use conical shaped radiators or rounded ends on the rod. Most of my
900MHz rod antenna have rounded ends on the elements to improve the
bandwidth. Fat elements also dramatically increase the antenna
bandwidth.

>It would seem that the optimum would be 29.2mm for Channel 6.
>299792.458 / 2437 / 4 * .95
>
>Why, then, do I see suggestions of 30.5 (which would be 100% of 2400MHz,
>outside of the band), or 31, or 32mm? Where is the .95 applied?

The 0.95 is my guess as to the fudge factor for the overall 1/2 wave
dipole. I don't know the magic correct number without knowing the
construction details.

>> Watch your accuracy. At 2400Mhz a wavelength is about 125mm. However,
>> each MHZ is equal to:
>> 125mm / 2400 = 0.052 mm/MHz
>> The band is 83.5 MHz wide, so your overall tolerance on cutting the
>> elements is:
>> 83.5 * 0.052mm/MHz = 4.35 mm.
>> it doesn't take much cutting error to end up with a non-functional
>> antenna.

>A margin of error of 4.35mm on an element that is only 31mm long is
>obvious enough that I would hope people can get it right.

Try again. The above calcs are for a full wavelength. The necessary
accuracy for a 1/4 wavelength would be 1/4th of that or about 1.1mm.
That's 1.1mm difference to mistune the antenna over the entire 83.5MHz
wide band. If you model the antenna and get exact numbers for
building the beast, then to keep the antenna inside the 2400-2483.5MHz
band, you need +/- 0.55mm accuracy on the 1/4 wave elements.

Another way of demonstrating the accuracy is:
1/4 wavelength at 2400.0 MHz is 31.25mm
1/4 wavelength at 2483.5 MHz is 30.20mm
Therefore, the total allowed cut range accuracy is:
(31.25 - 30.25) /2 = +/- 0.5mm
If you want it to operate on a specific channel (1, 6, or 11), then
the accuracy required is one third of the 0.5mm.

Can you say critical?

>But that's a full wave, not the margin on each 1/4 wave element. It also
>suggests that such an antenna would be perfect for some frequency with the
>range of 14 channels, but doesn't say anything about the tolerance for an
>acceptable antenna that will be used specifically on channel 6.

Bandwidth is directly affected by the antenna gain. Very roughly, for
a given antenna type, a 3dB increase in gain (by doubling the size)
will also double the Q (quality factor of the antenna) which
effectively cuts the bandwidth in half. That means that fairly low
gain resonant antennas (<12dBi) will have no problem operating over
all the channels without retuning, while higher gain resonant antennas
may only cover a few channels.

>If that were applied to each 1/4 wave element, are you saying that the
>margin of error on cutting the element is 1mm?

Worse. +/- 0.5mm. See above calcs.

--
Jeff Liebermann jeffl@comix.santa-cruz.ca.us
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558