"Can" Antennas
Michael Erskine
As I've been rather vocal about these can based antennas. I thought
I'd run some sims in NEC to see what it had to say about the design.
Here is what I have:
1) As in life, so with wave guide based antenna designs. Length
matters.
By this I mean that NEC sims indicate a wave guide based antenna
(cantenna) can be too long. Where I have been saying you should make
these designs 1.25 to 1.75 lambda(g) long... I need to correct
myself. This design when extended to that length exhibits a
bi-directional gain in the horizontal plane. Such a design *MIGHT* be
desirable for a passive repeater but *NOT* for a PtP antenna. NEC
runs show that about 3/4 lambda(g) is a better length. Slightly less
overall gain but a nice broad beam both horizontal and vertical.
2) Effiency matters... If you know what you are doing with the tool.
Things are much more satisfying...
By this I mean that gain figures for these antenna are not what I
thought they were. I thought I was measuring gains on the order of 15
dBi. I wasn't. NEC indicates gains on the order of 6 dBi. Therefore
I had to ask myself why I was seeing such high signal strengths with
these designs. The answer is effiency and not gain. As it happens
these designs are very efficient because their characteristic
impedence is very nearly the same as free space. This means nearly
all the power applied to the antenna is delivered to the ether. This
is not always the case.
3) Placement matters... (SWR)
By this I mean that the coupling between the antenna and the feed line
is critical to delivering power to the antenna so that it can be
radiated. As it happens it is quite easy to couple a 50 ohm line to a
wave guide antenna and achieve a low loss (low SWR) connection. This
means that more power id delivered to the antenna and less is wasted
as heat on the transmission line. As we are working with mw power
levels... SWR REALLY matters.
In general then the reason I was measuring such high RSSI's with
antennas which exhibit about 6 dB of gain turns out to be quite
simple. THEY ARE NEARLY 100 PERCENT EFFICIENT.
As for FB ratios, about 12 dB.
So... Here is a *good* design according to NEC.
1) The antenna is 3/4 lambda(g) long.
2) The antenna is 3.5" in diameter.
3) The antenna is fed by a 1/4 lambda (on the wire) antenna located
1/4 lambda(g) from the closed end of the antenna.
This antenna will exhibit the following characteristics (according to
NEC):
Half power beam width will be 90 degrees horizontal, max 5.72 dBi.
Half power beam width will be 65 degrees verticle, max 5.72 dBi.
In the horizontal plane there will be two side lobes and one backlobe.
The sidelobes are oriented roughly at 90 degrees to the bore sight.
The backlobe is down to -6.53 dB which yeilds a 12.4 dB FB ratio. In
the verticle plane there will be two side lobes and three backlobes.
The front to back ratio will be roughly the same at about 12 dB.
The antenna effiency (mw out / mw in) will be very nearly 1.0. Which
is excetional.
This antenna would make an excellent sector antenna for 90 degree
sectors. My RSSI measurements compare it to both a 16 dBi and an 18
dBi panels. RSSI's place this antenna on a par with those antennas.
FB ratio is no where as good, however and this antenna would not do
well upon a tower with other antennas covering multiple sectors
because of that. It will do well as a PtP antenna over short (.5 mi)
to middle distance (2.5 mi) links. It is trivially easy to construct
one with little or no test equipemnt.
This has taught me that commercial antenna builders really do know
their stuff, but they sometimes overlook effiency figures for gain
figures.
Atleast that is what I *think* I learned.
-m-
Adam Smith
Take a look at the calculations at the bottom of this page
http://www.turnpoint.net/wireless/cantennahowto.html
If you put in 3" for the can diameter, do the lengths
shown make sense to you? They seem a bit long.
Adam
"Michael Erskine" <osi...@deltaville.net> wrote in message news:e59f93b2.0203...@posting.google.com...
Michael Erskine
> Mike,
>
> Take a look at the calculations at the bottom of this page
> http://www.turnpoint.net/wireless/cantennahowto.html
>
> If you put in 3" for the can diameter, do the lengths
> shown make sense to you? They seem a bit long.
They look long to me.
I reduced the equations to a divide operation and I don't have the
formulas I derived the data from. I used the ARRL antenna handbook
and the microwave experimenter's handbook (also ARRL) to derive the
equations here.
fc(TE11) comes out at 2305.8 MHz.
6917.26/3.0 = 2305.8
fc(TM01) comes out at 3011.6 MHz.
9034.85/3.0 = 3011.6
lambds(g) = 11802.85/sqrt(fo^2-fc(TM11)^2) = 12.01"
I used fo=2483.5 MHz. The only reason I computed TM01 was to see how
close it was to fo.
I hope that works.
Michael Erskine
-m-
Michael Erskine
> Ok...
>
One other thing I should mention about the antenna. The antenna feed
point is off center on the can. This causes "skew" in shorter
antennas (length < 1.25 guide wavelength). The skew evidences itself
toward the side of the can away from the feed point and is about five
degrees. Thus if you mount the antenna connector down you would want
to set about five degrees of down tilt on the antenna to get the
optimal pointing along the horizon. If you wanted about five degrees
of down tilt you could just mount the antenna level and put the
connector up...
I did model a half wave wire centered upon the can at various
distances from the bottom. Excitation point on the wire was the
center of the wire. This could be accomplished by making a "T" shaped
driven element and mounting the connector dead center on the bottom of
the can. I did not model the phasing problems which would be
associated with the leg of the "T" but I am sure they would need to be
accounted for. One could feed it with coax but the additional
difficulty of construction seems a waste to me as all it does is
remove the skew and the skew is well within the major lobe of the
antenna. In a practical sense there would likely be no measurable
difference due to the skew if you just ignored it.
Anyway. I still don't really understand why my actual measurements
indicate such high RSSI's... Well later. Not today.
-m-
Adam Smith
This came from http://www.turnpoint.net/wireless/cantennahowto.html
after editing out a lot of text, pictures, etc., getting just to the
relevant javascript.
Maybe you (or someone) can review the calcs and either confirm
they are okay, or make the appropriate changes and repost the
calculate part of the javascript.
---------------------------------------------------------
<html>
<head>
<title>Waveguide Calc</title>
<script language="JavaScript">
<!--
function wgcalc(form)
{
form.fcte11.value=(6917.26/form.diam.value);
form.fctm01.value=(9034.85/form.diam.value);
form.wg.value=(11802.85 / Math.sqrt(5938969 - (form.fcte11.value * form.fcte11.value)));
form.wg4.value=(.25*form.wg.value);
form.wg34.value=(.75*form.wg.value);
}
// --->
</script>
</head>
<body>
<pre>
<form name="wavecalc">
Enter waveguide Can Diameter ........... <input name="diam" value="3.25" size=7> inches
<input type="button" onclick="wgcalc(this.form)" value="Calculate">
Cuttoff Frequency in MHz for TE11 mode.. <input name="fcte11" value="" size= 10> MHz
Cuttoff Frequency in Mhz for TM01 mode.. <input name="fctm01" value="" size= 10> MHz
Guide Wavelength in Inches ............. <input name="wg" value="" size=10> inches
1/4 Guide Wavelength.................... <input name="wg4" value="" size=10> inches
3/4 Guide Wavelength ................... <input name="wg34" value="" size=10> inches
</form>
Solder 1.21" probe to connector and mount at 1/4 Guide wavelength point.
</pre>
</body>
</html>
---------------------------------------------------------
"Michael Erskine" <osi...@deltaville.net> wrote in message news:e59f93b2.02040...@posting.google.com...
Michael Erskine
Those formulas are straight out of the ARRL antenna handbook. I did
not derive them myself.
-m-
Michael Erskine
Change the constant 5938969 to 6167772.25 if you want your numbers to
agree with mine, 'course then we might both be wrong. I'll have a
further look at the handbook when I get home.
-m-
Michael Erskine
The difference in length is due to the fact that the author of the
code chose to use the center frequency of the band as his "operating
frequency" whereas I chose to use the top of the band. In retrospect
I think his choice is likely the best choice and his numbers are "more
correct" than mine.
It may well be that all the computations should be based upon the
center frequency of the band. I'd go with his numbers.
I have reviewed the literature and can see no reason that I chose the
top of the band. I suppose my notes would have reflected the
reasoning behind my choice but they elude me. It may well be that I
just made the wrong choice.
When time permits, I'll exercise the calculator and drop more
information here.
-m-
Adam Smith
Sounds good. The part that has me confused is the reference
to 1/4 and 3/4 wavelengths but the numbers produced look
a lot bigger than 1/4 wave at 2.4 Ghz. Something specific
to waveguides perhaps, or not what I think it means ?
Just cut the HTML code between the lines, put it in the
notepad, save it, and rename the txt file "calc.htm" and open
it with a browser. It works fine.
Adam
"Michael Erskine" <osi...@deltaville.net> wrote in message news:e59f93b2.02040...@posting.google.com...
c....@nospam.blueyonder.co.uk
I also noticed that he used a 3.8" dia stew tin, is that not to large a diameter
can for 2.4GHz ?
Clive
Adam Smith
The larger diameter impacts the bandwidth of the waveguide as well as the
other measurements.
Go to the source of the javascript waveguide-can calculator
http://www.turnpoint.net/wireless/cantennahowto.html
at the bottom of the page you will see some notes that explain
how to use the other calculations to decide if the resulting waveguide
is within certain desired specs.
Adam
<c....@nospam.blueyonder.co.uk> wrote in message news:u6goau00m1kl3cjf7...@4ax.com...
c....@nospam.blueyonder.co.uk
Hi Adam.
>The larger diameter impacts the bandwidth of the waveguide as well as the
>other measurements.
>
>Go to the source of the javascript waveguide-can calculator
>
>http://www.turnpoint.net/wireless/cantennahowto.html
Yes but he used a can of 3.87", the frequency range for that dia works out to be
1.79GHz to 2.33GHz .... too low for a centre frequency of around 2.44GHz
I decided to try things out yesterday, so i made a can antenna using a dried
milk marvel can - 94mm * 157mm (3.7" * 6.18"), it actually worked very well, we
tested a working link at 1.8 miles, the other end used a simple butter fly or
double quad reflector/driven element antenna
<http://www.frars.org.uk/cgi-bin/render.pl?pageid=1064>.
The quad antenna was connected (via a 1 meter length of RG223 50R coax) to a
WAP11 (standard output power level settings - presumably around 40mW) while the
can antenna was connected (again using a 1 meter length of RG223 50R coax) to my
laptop pcmcia compaq WL110 card (a rebadged orinoco gold card), signal strength
reading was approx -85db).
I then replaced the can antenna with another quad antenna - the link failed,
tested showed the can antenna to have an extra 4db gain over the quad, not bad
really, considering the upper TM01 limit is just about 2.44GHz, we used channel
9 for the test, if i'd used channel 1 then things might have been a little
better though i guess.
I made no attempt to fine tune the 1/4wave element in the can for best SWR btw.
All-in-all the can antenna seems to work very well considering it wasn't at
optimum dimensions.
I shall measure it's effective dbi i think soon, just out of interest.
Clive
Adam Smith
Take a look at the "More effective version" design at
http://www.saunalahti.fi/elepal/antenna2.html
that presumably adds another 3db (doubling) of ERP.
It adds a "funnel" to the end of the can.
Do you plan to cut the length down to the 3/4 wavelength
dimension of 5.66 inches? You'd have to in order to
add the funnel.
Your findings are impressive, if you can get that extra
3 dB with the funnel attachment, that would give you
an even better link.
Adam
<c....@nospam.blueyonder.co.uk> wrote in message news:g5qqau0slsm69g6st...@4ax.com...
c....@nospam.blueyonder.co.uk
lo Adam.
>Take a look at the "More effective version" design at
>http://www.saunalahti.fi/elepal/antenna2.html
>that presumably adds another 3db (doubling) of ERP.
>It adds a "funnel" to the end of the can.
>
>Do you plan to cut the length down to the 3/4 wavelength
>dimension of 5.66 inches? You'd have to in order to
>add the funnel.
>
>Your findings are impressive, if you can get that extra
>3 dB with the funnel attachment, that would give you
>an even better link.
Well, if i could attach a funnel to the thing i wud certainly try it, not sure
what i could use for the funnel or how well i could get it to earth itself to
the foil lined card board can though.
I might get some thin copper sheet and make a new one, this time with more
optimal dimensions for 2.44GHz.
One thing i did forget to do was to set the wap11 to use only the one antenna
connected to it, like a fewl i left it set to use both antenna sockets, it still
had the standard wip connected to the other socket, i think if i'd have set it
to just use the proper test aerial then that might have given me upto an extra
3db (maybe just on transmit but still - depends if the wap11 operates it's
diversity system properly) :(
Plus if i use the can antenna at both ends rather than just the one end, that
will give a 4db improvement over the quad aerial too.
So should the length of my current can be 5.6" rather than the current 6.18"
(even without the funnel) ? ... would it make that much difference when not
using the funnel do you think ?
Clive
Adam Smith
I don't know if anyone else but you and I are interested in this project,
we could take it to private e-mail. If anyone else is considering the
Cantenna, please chime in.
I'd go for the 3/4L length given the diagram on the other page (see
below) with the funnel. The attachment of the funnel looks easy
if you're going to use either copper pipe (which is the way I'm
leaning" or galvanized steel ducting, either of which can handle
solder and/or pop rivets. The funnel - all you have to do is make
one end length equal to the circumference of the waveguide,
and extend it about an inch to give it something to solder/rivet.
The Diagram at http://www.saunalahti.fi/elepal/antenna2.html
doesn't explain the reason for the 30 degree angle of the
funnel, but if the ratios are correct, the funnel dimensions can
be easily derived from whatever diameter is used for the original
waveguide since the 30 degree conical section ends up with
a diameter of 1.7x the waveguide diameter (assuming the
figures are correct for the 100 mm dia waveguide in the
drawing).
Am still anxious to hear what empirical gain measurements
you find with a more optimally dimensioned cantenna. I'm
holding off construction until your findings are known, as
it's nice not to reinvent the wheel too many times.
Adam
<c....@nospam.blueyonder.co.uk> wrote in message news:l1orau0c9asl9r2fh...@4ax.com...
Vinny Gullotta
and have been following what you guys are saying. I'm not very good with
the antenna stuff, but I like to tinker :-)
--
Vinny Gullotta
www.wifidirect.com
ORiNOCO Gold PC Cards - $89.95 on sale now!
"Adam Smith" <nom...@stopspam.com> wrote in message
news:9Mnr8.3640$%A.40...@news1.west.cox.net...
bat...@chip.soliddesign.net
> Clive,
>
> I don't know if anyone else but you and I are interested in this project,
> we could take it to private e-mail. If anyone else is considering the
> Cantenna, please chime in.
Please keep the public discussion, so others will benefit from your
work. I'm interested enough to follow the discussion, but not
interested enough to do anything about it.
Joe