Trying to use what I had at hand, I put together a quick cantenna and
ran some tests. I'm describing the results here hoping that I can get
ideas for improvement. I built the cantenna following the instructions
at: http://www.turnpoint.net/wireless/cantennahowto.html
My client card is a Linksys WMP11v4. My router is a Netgear MR814v2. I
built a cantenna as follows:
diameter = 3.25 inches
length = 6 inches (one can) or 12 inches (two cans)
probe 2.5 inches from end of can
probe length = 1.2 inches (seemed short compared to the pictures but
that's what the instructions gave)
coax cable = 1 foot of RG-6 quad shield or 5 feet of same (bad, I
know, but that's what I had)
connectors = one or three F type connectors (again, what I had)
I attached the cantenna to the WMP11 and used Netstumbler to make
some measurements. The overall results were disappointing. The
cantenna didn't give me any improvement over the original rubber ducky
antenna that came with the card. I was able to see drops in the signal
level of up to about 20 dB depending on where I pointed the cantenna,
but even when pointed in the optimum direction the level was about
equal to level with the card's original omni antenna.
Thinking that I might be losing a lot of signal in the coax and
connectors, I shortened the cable from 5 feet to 1 foot and eliminated
2 of the connectors. But that made no difference. That surprised me.
Doesn't that imply that the problem is elsewhere? I know I should get
some LMR cable and the right connectors, but I wonder if it will make
a difference since shortening the lossy cable and removing connectors
didn't help.
I also tried changing the length of the cantenna from 6 inches to 12
inches by adding a second can but that made very little difference
(about 1 or 2 dB).
Another thing I noticed is that, without any modifications, a usb
dongle I had on hand (Zonet ZEW2501) worked much better than the
Linksys WMP11 card. The maximum signal level was about 12 dB higher
for the ZEW2501 than the WMP11. The ZEW2501 is tiny. The WMP11 is a
more serious looking piece of hardware, but it performs much worse
than the tiny dongle. Is there a lot of difference in the quality of
different manufacturer's radios?
> I attached the cantenna to the WMP11 and used Netstumbler to make
> some measurements. The overall results were disappointing. The
> cantenna didn't give me any improvement over the original rubber ducky
What you haven't said is how far away from the AP you were when testing.
If you're in the same room and already have a good signal, the result
will be masked by reflections and the generally good signal that you
already have.
Try it once you've gone somewhere where the signal is really weak and
then see what difference you have. Or find a neighbours signal with the
cantenna and then take it off and see if it still works.
> than the tiny dongle. Is there a lot of difference in the quality of
> different manufacturer's radios?
David Taylor <djtaylor@bigfoot.com> wrote:
>> I attached the cantenna to the WMP11 and used Netstumbler to make
>> some measurements. The overall results were disappointing. The
>> cantenna didn't give me any improvement over the original rubber ducky
....
> Try it once you've gone somewhere where the signal is really weak and
> then see what difference you have. Or find a neighbours signal with the
> cantenna and then take it off and see if it still works.
I agree with David. When I first set up for testing, I was too close, and
I saw very little difference in the NetStumbler signal. It's almost as if
the device doesn't report signal above a certain strength. If you move to
where the signal is in the 70-80 range in NetStumbler, you can see some
antenna improvements. If it's in the 40-50 range, you aren't going to see
much change at all.
It's also helpful to be outside, at least a few feet off the ground, and
have a clean line of sight to the AP.
A single 3.25" can should give you 8-10 dB improvement in NetStumbler.
The 20dB drop that you see by pointing it the wrong way is impressive.
Or maybe your cable shorted out when you twisted it that way ;-)
> probe length = 1.2 inches (seemed short compared to the pictures but
That is 1.21" of _exposed_ probe, including the exposed center conductor of
the connector. The wire also goes down into the connector a bit.
---
Clarence A Dold - Hidden Valley (Lake County) CA USA 38.8,-122.5
Wow, was I wrong! Now I see a 30 dB difference in Netstumbler when I
aim the cantenna for minimum and maximum signal. I get about 10 dB
better than the maximum signal with the original rubber duck antenna.
Heres what happened: I decided to put up some pictures to ask if
anyone saw anything I missed. After taking the pictures, it was clear
that that the hot coax lead had lifted off the point where I soldered
it to the client card. It must have broken off due to moving the cable
around after I put the card into the computer since I checked it just
before. Slap me with a fish.
I'll put up the Netstumbler results and some pictures of the cantenna
in a little while. I'll post the web address in this post.
I am still puzzled by some things. I get a very strong signal from my
own two APs (in the same room about 20 feet away) but none from other
nearby APs. There are about three others in houses close to me that I
can pick up easily with my Zonet ZEW2501 usb dongle. But they don't
show up when I use the cantenna even if I point it in all kinds of
different directions and turn off my own routers to prevent
interference. Any ideas on why this is happening?
bjs555 wrote:
> Trying to use what I had at hand, I put together a quick cantenna and
> ran some tests. I'm describing the results here hoping that I can get
> ideas for improvement. I built the cantenna following the instructions
> at: http://www.turnpoint.net/wireless/cantennahowto.html
>
> My client card is a Linksys WMP11v4. My router is a Netgear MR814v2. I
> built a cantenna as follows:
> diameter = 3.25 inches
> length = 6 inches (one can) or 12 inches (two cans)
> probe 2.5 inches from end of can
> probe length = 1.2 inches (seemed short compared to the pictures but
> that's what the instructions gave)
> coax cable = 1 foot of RG-6 quad shield or 5 feet of same (bad, I
> know, but that's what I had)
> connectors = one or three F type connectors (again, what I had)
>
> I attached the cantenna to the WMP11 and used Netstumbler to make
> some measurements. The overall results were disappointing. The
> cantenna didn't give me any improvement over the original rubber ducky
> antenna that came with the card. I was able to see drops in the signal
> level of up to about 20 dB depending on where I pointed the cantenna,
> but even when pointed in the optimum direction the level was about
> equal to level with the card's original omni antenna.
>
> Thinking that I might be losing a lot of signal in the coax and
> connectors, I shortened the cable from 5 feet to 1 foot and eliminated
> 2 of the connectors. But that made no difference. That surprised me.
> Doesn't that imply that the problem is elsewhere? I know I should get
> some LMR cable and the right connectors, but I wonder if it will make
> a difference since shortening the lossy cable and removing connectors
> didn't help.
>
> I also tried changing the length of the cantenna from 6 inches to 12
> inches by adding a second can but that made very little difference
> (about 1 or 2 dB).
>
> Another thing I noticed is that, without any modifications, a usb
> dongle I had on hand (Zonet ZEW2501) worked much better than the
> Linksys WMP11 card. The maximum signal level was about 12 dB higher
> for the ZEW2501 than the WMP11. The ZEW2501 is tiny. The WMP11 is a
> more serious looking piece of hardware, but it performs much worse
> than the tiny dongle. Is there a lot of difference in the quality of
> different manufacturer's radios?
>
> Many thanks for help.
>
> Confused,
> Bruce
Bruce, Get a Hawking Tech Omni Directional 6DB Antenna, I have one on
the outside of the house sitting on the window sill of the 2nd floor, it
has about a 26 inch cable feeding a NeatGear WG311 card, it works great
for me, I picked up 9 neighborhood AP't this morning.
I don't know if you live near a Micro Center, that are not too many, but
they carry Hawking Products. The Tech Omni Directional 6Db antenna is
only 5 bucks after a rebate, Hawking Tech Wireless G USB Network Adapter
6DB Antenna Combo 802.11G is the same deal, 5 bucks with a rebate, or
was when I bought mine. I have that adapter, and it works great. Go to http://www.microcenter.com/ and check out these two devices, I didn't
read all the details, since There is a Micro Center store in my area
located in Fairfax Virginia. You might be able to get these product
thru an on-line order, but check on the rebate details for on-line orders.
FM
>Bruce, Get a Hawking Tech Omni Directional 6DB Antenna, I have one on
>the outside of the house sitting on the window sill of the 2nd floor, it
>has about a 26 inch cable feeding a NeatGear WG311 card, it works great
>for me, I picked up 9 neighborhood AP't this morning.
>I don't know if you live near a Micro Center, that are not too many, but
>they carry Hawking Products. The Tech Omni Directional 6Db antenna is
>only 5 bucks after a rebate, Hawking Tech Wireless G USB Network Adapter
>6DB Antenna Combo 802.11G is the same deal, 5 bucks with a rebate, or
>was when I bought mine. I have that adapter, and it works great. Go to
>http://www.microcenter.com/ and check out these two devices, I didn't
>read all the details, since There is a Micro Center store in my area
>located in Fairfax Virginia. You might be able to get these product
>thru an on-line order, but check on the rebate details for on-line orders.
>FM
Thanks FM! I've heard good things about Hawking and that's a great
deal. The cable and plug alone are worth more than that. An omni isn't
the best for long distances, I'm told, but I can't argue with success.
I'm definitely getting one of those for experiments.
>Yeah, it's the wrong colour! :)
>
>Why did you fart around soldering to the back of the board when you
>could have just made up a short pigtail to convert from the SMA to N
>type?
>
>David.
Haha. Didn't have the connectors and none in local shops. Didn't want
to wait for delivery. Also, I was trying to minimize the number of
reflections by using as few connectors as possible. But with using the
wrong impedance cable and all, the connectors might have helped as
much as hurt. Anyway, the board had nice big solder pins and I figured
"What could go wrong?". My bad.
I've done some more testing. I don't see any difference by using one
can instead of two. That is, shortening the length to 6 inches. Easier
to aim. Maybe there are differences on the order of a couple of dB
that I don't see in my ramshackle setup.
Rearrange the F connector in the can so that the shielded portion is barely
exposed, just enough for the attaching nut. The exposed inner conductor of
the probe should be 1.21".
--
---
Clarence A Dold - Hidden Valley (Lake County) CA USA 38.8,-122.5
The exposed center conductor of the coax cable is *FAR* too long.
Doing a coax to board transition is not easy, especially with the fat
coax you're using. Continue the bare copper wire shield wrap right up
to the end of the coax. Trim the center conductor to where it just
barely sticks out beyond the shield. Shave some of the insulation to
get the center conductor closer to the SMA center pin. Solder the
copper wrap to the SMA ground pins.
Otherwise, do it right and get a pigtail or adapter.
The probe length should be 1/4 wave (31mm) from the shield on the F
connector. You might wanna try a conical probe for improved
bandwidth.
bjs555 <aaa@bbb.com> wrote:
> Wow, was I wrong! Now I see a 30 dB difference in Netstumbler when I
> aim the cantenna for minimum and maximum signal. I get about 10 dB
> better than the maximum signal with the original rubber duck antenna.
That sounds about right.
> it to the client card. It must have broken off due to moving the cable
;-)
opps.jpg (I assume "oops") shows a lifted center conductor, but I don't see
the shield attachment point.
It should be coax as much as possible, with the shield and center conductor
separated as little as possible, soldered to appropriate contact points.
--
---
Clarence A Dold - Hidden Valley (Lake County) CA USA 38.8,-122.5
bjs555 <aaa@bbb.com> wrote:
> I've done some more testing. I don't see any difference by using one
> can instead of two. That is, shortening the length to 6 inches. Easier
> to aim. Maybe there are differences on the order of a couple of dB
> that I don't see in my ramshackle setup.
That seems to be true for the USB dongle in a can as well.
Two cans gives marginally better gain, but it is also much more
directional. I thought that was good, but in use, the friend that I gave
it to found it too hard to keep pointed at the WAP (using a photo tripod),
and preferred a single larger diameter can. http://www.rahul.net/dold/clarence/u...42-800x600.jpg
--
---
Clarence A Dold - Hidden Valley (Lake County) CA USA 38.8,-122.5
> Rearrange the F connector in the can so that the shielded portion is barely
> exposed, just enough for the attaching nut. The exposed inner conductor of
> the probe should be 1.21".
It looks like the total lenght of your F-Connector and probe is 1.21
inches, so the exposed conductor is substantailly too short.
--
---
Clarence A Dold - Hidden Valley (Lake County) CA USA 38.8,-122.5
>opps.jpg (I assume "oops") shows a lifted center conductor, but I don't see
>the shield attachment point.
>
>It should be coax as much as possible, with the shield and center conductor
>separated as little as possible, soldered to appropriate contact points.
The wire you see twisted around the shield extends out behind the
cable and is soldered to two ground contacts. The shield must be
aluminum or something - I couldn't solder to it so I twisted some bare
wire around it tightly. That's probably bad for long term use as the
connection could oxidize but ok for a quick test. And thanks to you
and Jeff I now know that I should make the exposed center conductor as
short as possible. I'll try that tomorrow.
>The exposed center conductor of the coax cable is *FAR* too long.
>Doing a coax to board transition is not easy, especially with the fat
>coax you're using. Continue the bare copper wire shield wrap right up
>to the end of the coax. Trim the center conductor to where it just
>barely sticks out beyond the shield. Shave some of the insulation to
>get the center conductor closer to the SMA center pin. Solder the
>copper wrap to the SMA ground pins.
I'll try that.
>Otherwise, do it right and get a pigtail or adapter.
I really should but I can't find them locally so I'll have to order
them over the net.
>The probe length should be 1/4 wave (31mm) from the shield on the F
>connector. You might wanna try a conical probe for improved
>bandwidth.
I wasn't sure whether the length should include the connector shield.
I guessed wrong. Also, I'll minimize the amount of connector extending
into the can as Clarance Dold suggests.
I'm still surprised that I don't pick up signals other than my own
with the cantenna like the usb dongle does. Does the high
directionality have something to do with that? But I tried pointing it
every which way. Maybe things will improve after I make some of the
suggested changes.
On Wed, 05 Oct 2005 20:07:09 -0400, bjs555 <aaa@bbb.com> wrote:
>I'll minimize the amount of connector extending
>into the can as Clarance Dold suggests.
That's not really that critical.
>I'm still surprised that I don't pick up signals other than my own
>with the cantenna like the usb dongle does. Does the high
>directionality have something to do with that? But I tried pointing it
>every which way. Maybe things will improve after I make some of the
>suggested changes.
I lengthened the probe in the can so that it's now 1.21 inches (yeah,
10 mil accuracy :) and reduced the amount of connector shield inside
the can. I didn't see any improvement and possibly even a slightly
lower signal than with the shorter probe. But my test setup leaves a
lot to be desired. I'm sure that there are all kinds of reflections
since I can get wildly different readings depending on where I point
the can. As more experienced people have pointed out, a small room
isn't the best place to run tests.
One puzzling thing for me continues to be the fact that I don't pick
up a lot of access points other than my own with my cantenna even if I
try pointing it in many different directions. It seems to me that the
cantenna works best when it's receiving a signal from another radio
that also has a directional antenna attached. I say that because I see
a lot better signal from the cantenna when I put a windsurfer
reflector (www.freeantennas.com) on my ap. That is, without changing
the distance between the cantenna and ap, and aiming the cantenna for
max signal, I pick up a much stronger signal with the windsurfer
attached. Didn't one guy try putting a funnel-like attachment on the
end of a cantenna? I think I saw a picture like that on someone's web
site. Looked goofy, and I don't see how the geometry really feeds more
signal into the can, but the idea of having a larger signal collection
area rings sorta true (like a 12 inch telescope vs a 6 inch one). So
maybe the reason I don't see a lot of other people's access points
with the cantenna is that they are using omni directional antennas.
Would that explain the lack of other access points showing up in
Netstumbler? I know there are at least a few around that I pick up
with a usb dongle.
I haven't yet shortened the exposed center conductor where I soldered
the RG-6 cable onto the pci client card. I'll have to try that later.
For fun, I tried disconnecting my cantenna entirely and seeing if I
could get a signal with just the RG-6 cable attached. I did pick up
enough signal from my own ap 20 feet away to connect even without any
cantenna. Maybe the exposed center conductor was feeding enough energy
into the radio. I also tried using both the cantenna and the rubber
duck antenna at the same time since I didn't remove the client card
connector. Kind of a stupid thing to do, I know, but surprisingly it
didn't kill the connection and the signal level reported by the
Linksys monitor went up by 10%. Guess you can get away with almost
anything at 20 feet. Along those lines, I found this site that seems
to be a quite different kind of antenna: http://www.andrew.com/products/trans...x/default.aspx
Wish I could be more quantitative but I don't have the equipment or
enough knowledge yet to make careful measurements. I'm having some fun
trying to figure out what's going on and I'm very interested in
hearing what others have done.
> One puzzling thing for me continues to be the fact that I don't pick
> up a lot of access points other than my own with my cantenna even if I
> try pointing it in many different directions. It seems to me that the
bjs555 <aaa@bbb.com> wrote:
> cantenna works best when it's receiving a signal from another radio
> that also has a directional antenna attached. I say that because I see
Polarity comes to mind. The can is three dimensional. You may have been
rotating the aziumth and elevation, but the orientation of the antenna
probe itself is also important. Try rotating the antenna on the long axis,
so that the probe is vertical, horizontal, and other points in between.
Especially if your small room is full of reflections, the optimum angle for
the probe may not be obvious.
> a lot better signal from the cantenna when I put a windsurfer
> reflector (www.freeantennas.com) on my ap. That is, without changing
> cantenna. Maybe the exposed center conductor was feeding enough energy
> into the radio.
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
Hey David, just saw on one of your web pages that you tried a horn on
the end of a cantenna. Sorry I called it goofy but it is funny
looking. But then I read some posts by Jeff saying that a horn makes a
smoother transition from the 50 Ohm radio source impedance to the 377
Ohm impedance of free space. Foot in mouth today. Everybody's a
critic, huh?
>Polarity comes to mind. The can is three dimensional. You may have been
>rotating the aziumth and elevation, but the orientation of the antenna
>probe itself is also important. Try rotating the antenna on the long axis,
>so that the probe is vertical, horizontal, and other points in between.
>Especially if your small room is full of reflections, the optimum angle for
>the probe may not be obvious.
Tried that. Rotated slowly through the long axis 360 degrees. Picked
up one ap but just barely.
> 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.
Very interesting. Thanks.
I'm starting to think that the reason I don't see many aps is that the
receive sensitivity of my Linksys WMP11 card isn't nearly as good as
my usb dongle. It's an older card that I just happened to have lying
around. But today I ordered more junk from Compusa and MicroCenter so
I can LearnByDestroying (tm). Compusa has WUSB11s for $4.99 after
rebate and MicroCenter has a usb dongle for $9.99, an SMC router for
$7.99, and a Hawking antenna for $4.99 all after rebates.
I looked up some specs for my radios. The Linksys WMP11 receive
sensitivity is around -82 dBm and the Zonet ZEW2501 is about -85 dBm.
I take it this means I need 6.31 uW to get a signal into the Linksys
and only 3.16 uW for the Zonet. Does that sound right? I think there's
probably some variation these is in these specs from unit to unit but
the lower Linksys sensitivity could explain what I see.
> Hey David, just saw on one of your web pages that you tried a horn on
> the end of a cantenna. Sorry I called it goofy but it is funny
> looking. But then I read some posts by Jeff saying that a horn makes a
I didn't see your "goofy" comment but the novel part was that the whole
antenna other than the rear and the N type connector and probe was all
cardboard. :)
It also happened to be the most effective antenna of the ones that I
made.
bjs555 <aaa@bbb.com> wrote:
> and a Hawking antenna for $4.99 all after rebates.
On one system, I installed a Hawking antenna on a Netgear WG311, which
has a tiny stock antenna. The "Hawking HAI6SDA Directional 6dBi 2.4GHz
Antenna" http://www.hawkingtech.com/prodSpec.php?ProdID=143 is pointed at a
Linksys BEFW11S4 that has a pair of Windsurfers.
It looks a little more professional than the cantenna. I used to have a
pair of corner reflectors, but I caught grief from the local interior
designer, and Jeff Liebermann, for the wrinkly tin foil. http://www.rahul.net/dold/clarence/S...0-strength.htm
I read some mutterings from Don Widders today (posted in 2003, I wonder
what happened to him), that said that the antenna should be centered in a
corner reflector, because it might have corresponding reflections from
three points, one on each side, and the fold itself. He felt that the
antenna should be off-center a little.
--
---
Clarence A Dold - Hidden Valley (Lake County) CA USA 38.8,-122.5
On Thu, 06 Oct 2005 17:52:29 -0400, bjs555 <aaa@bbb.com> wrote:
>I looked up some specs for my radios. The Linksys WMP11 receive
>sensitivity is around -82 dBm and the Zonet ZEW2501 is about -85 dBm.
At what connection speed and what type of modulation (CCK or OFDM)?
The sensitivity varies with the connection speed and modulation type.
If you're going to compare sensitivities, it has to be at the same
speed and modulation type.
>I take it this means I need 6.31 uW to get a signal into the Linksys
>and only 3.16 uW for the Zonet. Does that sound right? I think there's
>probably some variation these is in these specs from unit to unit but
>the lower Linksys sensitivity could explain what I see.
The variations from just the test equipment and test setup is at least
+/- 3dB. There's also some creativity in whether to use the US 10E5
BER (bit error rate) reference, or the ISO 10E6 BER. Also there are
some that measure BER or PER (packet error rate).
The data sheet numbers for the complete radio is often exactly the
same as that for the chipset data sheet. Well, that's impossible
because of all the circuitry between the antenna connector and the
chip. There's the diversity switch, 1 or 2 connectors, a chunk of
coax, and often a matching network.
cantenna experiments 
Linear Mode
