On 27 Dec 2008 06:23:27 +0100, bubblebrain <firstname.lastname@example.org>
>Jeff Liebermann <email@example.com> wrote in
>Go back and read the OP, cuz I don't think you did not understood what I
>said, or more likely you began answering before you had read the entire
Both. I read it through twice before responding. I didn't understand
what you were asking. Too many computer, too vague a description of
your hardware, no clue what you were connecting with, few numbers.
Actually, it made more sense reading it starting from the bottom and
working up. However, I'm lazy and decided to reply in the order in
which you presented your questions. If you want to avoid a repeat
performance, try supplying:
1. What problem are you trying to solve?
2. What do you have to work with? (Hardware, software, equipment,
network topology, topography, environmental issues, etc).
3. If troubleshooting, what have you done so far and what happened?
Also, sorry for the delayed reply. Saturday degenerated into a work
day for me. It's difficult to say no to the brats with their new
Christmas toys. Once the toys were working, I was bribed into
rewiring the home office.
>I am getting a 40-66% signal strength using a wubber ducky pole antenna
>with a parabolic reflector behind it. I was asking what construction
>factors, other than the ones I already know about (proper curvature and
>placement of the wubber duckey pickup at the focal point) might maximize
>the reflector efficiency.
Thanks. The basics first. Signal strength is NOT the primary
determination of download speed and connection reliability. SNR
(Signal to noise ratio), which can also be represented by BER (bit
error rate) and PER (packet error rate) is generally more important.
For example, you can have a good solid connection, but if there's any
interference along the path, you're going to see errors. With
interference, such errors may not cause a drop is signal strength. The
problem is that it's impossible to tune or aim the antenna by watching
the SNR indication.
This is where a high gain direction antenna is useful. Besides
increasing the signal strength by providing gain in the desired
direction, it also decreases the gain to the sides and back of the
antenna, thus reducing the interference pickup. It doesn't do
anything for interference at the other end of the link, but you don't
have any control over that anyway.
As for building your own reflector, the major issues (in order of
1. Accurate parabolic curvature. You're trying to maintain at least
+/- 1/10th of a wavelength accuracy. That's about 1.3cm which
should be quite easy to achieve.
2. Feed illumination efficiency. This is the major problem with the
hang-on reflector. Much of the TX RF goes in directions that
don't hit the parabolic reflector. That RF is lost and cannot
be recovered. See:
in the section on feeds, illumination, spillover, etc at:
3. Different TX/RX gains. In transmit, feeding a dish with a
vertical collinear results in considerable illumination loss.
That's not the case in receive, where everything that hits
the dish will hit the collinear feed.
4. VSWR and matching. Placing a reflector near the antenna will
change the characteristic impedance of the antenna. This may
cause some additional TX loss. It won't be much, but if you're
looking for optimum performance, this should be measured and
5. Width of the parabolic reflector. Most likely, your reflector
is curved into a parabola only in one axis. It's flat in the
other axis. The width of the reflector should be at least as
wide as the antenna (ideally 1 wavelength or 12.5cm). Any
wider will not help as the reflected signal does not go towards
the other end of the link, but instead, goes towards up or down.
For a clue, replace the collinear antenna with a light bulb (point
source) and look at the reflected light pattern on the wall. That's
very roughly how your RF is going to be distributed. It won't be a
nice clean spot on the wall, but more like a vertical smear, with lots
of wasted over spray to the sides coming directly from the lamp.
Here's the Coleman lantern version:
>Reflective material used (using alum foil
Anything that reflects RF will work. Aluminum sheet or foil will work
just fine. Most steel's will also work.
Stainless steel salad bowl with USB dongle shoved down the pipe:
Incidentally, when I do the lamp test with the salad bowl, the result
is an annular ring on the wall. It's not a spot, but more like an out
of focus ring. It's far from optimum, but better than the USB dongle
For single axis parabolics, about 1/2 wavelength (6.25cm) is the
minimum height. I prefer 1 wavelength high (12.5cm). You don't want
a deep dish (high f/D ratio) as reflections from opposite sides of the
dish start to cancel each other. See:
>what common mistakes are mostly made in homemade
>construction of these?
None. Here's how it works. As long as the antenna gain is fairly low
(i.e. <10dBi gain), you can be quite sloppy and still end up with a
usable antenna. The basic design of stuffing a reflector behind a
collinear is at best a kludge, but will still work better than the
collinear by itself. The reflector can be flat or parabolic, and
there will be perhaps 1-2dB differences in gain. For example, compare
the dish and the flat plate reflectors based on the Hawkings USB
7.7dBi for the dish.
8.3dBi for the flat plate.
Fairly close, methinks.
When the gains go above about 10dBi, dimensions and construction
techniques become rather critical. Small errors result in substantial
changes in gain and pattern. Small goofs are less tolerated.
What all this means is that you can be rather creative and sloppy with
low gain reflector antennas, and still end up with a usable antenna.
It also means that there's little you can do with the design or
construction to make any dramatic improvements. Sorry.
>It is already about as large as I want it to be
>for indoor use at about 2' by 1' The wubber duckey is about 7 inches
You probably won't believe me, but the gain of the rubber ducky
(collinear) has no effect on the gain of the antenna. In fact, too
much gain causes only a small part of the parabolic reflector to be
illuminated, resulting in less effective aperture, and therefore less
gain. A 7" rubber ducky will have about 2dBi gain resulting in
perhaps a 60 degree beamwidth, which is sufficiently wide to
illuminate almost any conceivable reflector. You're fine.
You can calculate the maximum possible gain of a parabolic dish
I can't calculate your 2ft x 1ft antenna without a clue as to how it's
>Second question is which brand/ model of usb dongle has the best
>transmit receive sensitivity (more important) for use with the same
>reflector and my desktop computer.
Sorry. I can't answer what's best. There are some high power USB
In my never humble opinion, there's no such thing as a good USB
device. That's not the fault of the radio or chipset. It's the fault
of the tiny antenna that's used in the typical USB device. There are
some that have real antennas, or external antenna connectors that
deserve some notice, but the ones with internal antennas are marginal
at best. The generally crappy Windoze device drivers are also of
Note the external antenna connectors (MMCX):
>I cannot obviously use the pcmcia
>card with my desktop since I have to go about 10 feet to it and it is
>designed for laptops.
Actually, you can. Some PCMCIA cards have external antenna
connectors. Instead of a reflector kludge, get a real dish antenna,
with 19 or 24dBi gain, 10ft of LMR-240 or LMR-400, and approximately a
1ft LMR-100 or RG-316 pigtail adapter, and you can use the PCMCIA
card. I've had good luck with Buffalo cards, but they're banned in
the US thanks to patent litigation by CSIRO.
>Now I recall that seattle wireless did some
>testing of these, which I have to revisit, but their tests were not very
>comprehensive as I recall.
Most of the numbers I saw were copied directly from the data sheets.
There was some comparative field testing done, but that was for early
802.11b only cards.
>I have no idea where the APs are, a visual survey of the area saw no
>tell tale antennas, one AP I am guessing is far away on a tower about 50
>feet high, probably about 3 miles away.
You claimed that this was a municipal WLAN yet you can't find the
antennas? Use your dish antenna, wave it around, and see where the
signal is strongest. Do it from a place where you have a clear shot
of the probable locations. There are maps of most municipal networks,
but since you don't want to disclose identifying information, I can't
help you find them. I'm not sure why you need total anonymity for a
public wireless system, but that's your decision.
>No line of sight, so I have no
>idea what is between us. I have at least 2-3 independent signals I can
>connect to, so the relevant question is what can I use for the desktop
>radio and a good usb radio is what comes to mind. I live in an apartment
>complex, so I am stuck with an indoor reflector or antenna.
Line of sight is important. You might be able to get a connection
without line of sight, but you will probably not be able to maintain
the connection. It will fade in and out as obstructions in the line
of sight move reflections around.
>Since I don't have a line of sight, I am guessing that a dish antenna
>will not work, since I have no reference point to point it to, nor do I
>know what lies between me and the AP.
Actually, I think a dish antenna would work best because it directs
the signal where you most want it (and rejects junk coming from the
sides). Still, it's difficult to predict without knowing what's along
>I am not going to buy fancy signal
>detecting equipment for this, it's a shoestring operation.
>The venetian blinds are oriented horizontally, and metal as I said
>before, but their adjustment makes little to no diff in signal strength.
Very strange. It should block a vertically polarized signal quite
well. Sorry, but I don't have a guess as to what's happening.
>I'm not going to tell you my OS, radio make, etc. That is information
>that can conceivable could be used to identify me individually and I
>don't want that, thanks.
You're more likely to be identified by sniffing your traffic and
identifying your logins passwords and surfing habits. I'm always
suspicious of people that want anonymity as I tend to assume they're
trying to perpetrate a crime of some sorts.
>I have doubts that there exists a usb dongle on
>the market that will match the receive sensitivity and transmit power of
>the pcmcia card I am using.
Agreed. Without knowing what you're using, it's difficult to tell.
List of claimed rx sensitivities of various products:
Everyone lies, but that's ok because nobody listens.
>Most users report they are weaker.
It's mostly the differences between a relatively large diversity
PCMCIA antenna system, and a tiny ceramic patch or PIFA antenna. The
PCB antenna in the PCMCIA card might have about 2dBi gain. The tiny
patch or PIFA are typically -2 to -4dBi. (6dB difference is 2 times
>have any concrete suggestions on how to maximize the efficiency of the
>usb dongle, please state them.
Don't use concrete. It will ruin the USB device warranty.
>At this writing I don't know what else I can use with the desktop since
>I have to go about 10 feet to it and cannot install an outdoor antenna
>in these apartments.
Yep. That's because you haven't asked what to use. You've supplied a
list of questions, some of which are relevant to the problem you're
trying to solve, but most of which are just interesting diversions.
Try asking what *TYPE* of antenna to use.
I think a dish in the window is impractical for an apartment. However,
panel antennas offer almost the same gain and are less obnoxious. The
problem is that you don't know where the other end of the link is
located. Therefore, it's possible that the optimum position of a
window mounted panel antenna might be an inconvenient angle. If you
can fabricate some kind of adjustable antenna positioner, it might be
made to work.
Look though the panels (and dish) antennas at:
They also have 10ft coax lengths and pigtails.
You can use USB but I don't like them mostly because the drivers are
marginal. I suggest an external USB or ethernet connected wireless
bridge radio with an external antenna connector. I'm using various
Linksys WRT54G mutations with DD-WRT firmware. It has a client mode
that works well enough. It's not as convenient as a USB device, but
works much better.
There are also devices made for WISP (wireless ISP) service that might
be suitable. See:
under the various "Wireless ISP Solutions".
>Maybe a USB with a dish reflector behind it might
>be worth a try, all homemade. Have seen some cool ones made from an
Naw, my salad bowl is more cool. However, WokFi is the current high
Far too many Wi-Fi antennas:
>You really did not answer my questions, but you did make me think about it
>more completely, so now I know what to research.
Yep. That's the way I work. I explain how things work and where to
look. You get to answer your own questions and solve your own
Jeff Liebermann firstname.lastname@example.org
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558