[lug] cheap 802.11b for linux...

D. Stimits stimits at idcomm.com
Fri Mar 22 14:12:45 MST 2002


"J. Wayde Allen" wrote:
> 
> On Thu, 21 Mar 2002, D. Stimits wrote:
> 
> > It would be interesting if one of the 802.11b's had an ethernet plug on
> > it, and could be used independently of the laptop. Having the
> > transceiver portion mounted directly at the antenna and running ethernet
> > to it could have some advantages (though need for weather-proofing is a
> > problem).
> 
> In the last couple of weeks I've learned that these are called access
> points.

So I see (I've never used wireless computer equipment). I would guess
though that some wireless LAN access points are purely wireless, and do
not offer ethernet taps, and have as their only communications the
wireless signal (don't know, never really shopped for this).

> 
> > I'm curious if there are any SWR meters for low-powered 802.11b lines.
> 
> It is certainly possible to measure SWR at these frequencies, but the
> question is really why you'd want to and whether or not there is much
> meaning in this for 802.11b?  The standing wave ratio or SWR is a
> relatively poor measure of the effect of impedance mismatch (no phase
> information).  A more complete measurement would be the reflection
> coefficient.  The problem is that 802.11b is a spread spectrum signal.
> You'd need a single frequency signal to generate a standing wave on the
> line, so you couldn't just hook a meter in the line between your 802.11b
> transmitter and the antenna.  You'd really need to look at the mismatch
> characteristics of the antenna/transceiver combination across the signal
> bandwidth.  Very doable, but I doubt you are going to find an inexpensive
> SWR meter for the job.  The best tool would be a network analyzer.  The
> last one of these I bought was from Hewlett Packard (now Agilent) and cost
> $100,000.

Right, spread spectrum. One could use a sweep generator from lowest to
highest frequency, and plot that out, or measure power output from a
uniform noise generator, and graph it for some form of measuring actual
throughput (no, not practical).

> 
> > That is what you want from your antenna line.
> 
> No, you don't want this to be resonant.  You want to couple the energy in
> the line to the free space impedance of typically 377 ohms.
> 
> > If you have a meter (SWR meter) that can measure the phase
> > differences, you can use it to tune the antenna and make sure the line
> > going to it does not radiate.
> 
> There is no phase information in an SWR measurement.  It is the ratio
> between two E-field magnitudes.  This destroys the fact that the E-field
> description is a complex number and simultaneously throws away phase
> information.

Right, the inductive and capacitive fields have +/- sqrt(-1) as one side
of a triangle, with the real resistance as the other leg, and the
effective impedance as the hypotenuse. The hypotenuse is just a
convenient representation of impedance, whereas the exact measurement is
really the phase angle and the real resistance leg (capacitive and
inductive legs are in opposite directions and so cancel out).  The thing
with a 1:1 swr is that it only occurs at the point where inductive and
capacitive cancel out, so at this one ratio, you know the phase angle is
zero; at further ratios, you could calculate phase angle if you also
know the pure resistive component. Ideal would be an antenna that is
resonant on a variable basis that tracks with your sending signal.
Lacking that, antennas and all resonant circuits have a certain Q
(quality) component defining how sharply they change when moving away
from the center resonance, and a spread spectrum device would want
something that looks more like a band pass filter than a single resonant
point. The actual antenna though cannot be made like a band pass filter
with purely element geometry designs...so it must have a resonance and a
Q...capacitive and inductive elements separate from the antenna can be
used to alter how it appears to the transmitter or receiver, but they
are not themselves as good of a solution as an antenna that is actually
behaving directly as desired. There is a tendancy to reduce Q by adding
purely resistive elements, and so it behaves right, but it also is lower
performance. I'm suggesting that many of the consumer broadband products
probably do not put as much design effort into properly "tuning"
(bandpass adjusting) the antennas and their lines as they should, so
they are either more lossy than they need to be, or accept more outside
spectrum noise than is required (either way you lose signal-to-noise
ratio). Being able to sweep the spectrum and display the swr would at
least give an idea of whether the antenna and line system is "centered".
I suppose once you add the sweep generator at that frequency, there is
no way it could be inexpensive for home users while still sensitive and
accurate at low power levels.

> 
> > re-radiated and never make its way to the receiver (plus the line would
> > be receiving energy from wavelengths you are not interested in, and
> > basically require the receiver to separate out more noise). So loss
> > versus distance really depends not just on length, but also on how well
> > tuned it is...if it is perfectly tuned, then it is simple to figure out
> > how well it will do based soley on the wire resistance and dielectric
> > efficiency.
> 
> In a spread spectrum system you need the antenna to be broadband, not
> tuned as you imply.

Yes. Mainly though that is because of how impractical it is to tune the
antenna system with the frequency being generated over a wide range...an
actively tuned antenna. There are however multiple antenna systems
whereby a smaller part of the spectrum is assigned to each antenna, and
filters actually funnel power to the most relevant antenna (probably not
an easy thing to do efficiently, and definitely not possible with a
single antenna). There are some antennas that have elements that are
tuned slightly off in one direction or another, and part of the elements
participate only part of the time, but those are in reality just a
single mount point for multiple antennas, and I think those are almost
entirely directional (whereas the desired effect is omni).

> 
> > An antenna line with a pure vacuum instead of a dielectric would
> > in theory not generate any heat at all, but then the diameter ends up
> > larger for the same capacitance.
> 
> Not true.  Your dielectric loss would go away, but you'd still have ohmic
> losses in both your inner and outer conductors.

It only says that the dielectric component would not generate heat. I
did not intend to make it sound like the other parts disappear. But
dielectric becomes more important as frequency goes up, whereas a pure
resistive element is not sensitive to wavelength. In fact, the note up
above about the Q of the system can be restated to say that a certain
amount of resistive element is actually desired in order to keep the
resonance of any part of the system from being too sharp. In any case,
the larger conductors required to have the same capacitive value when a
vacuum is used will be smaller resistance just because of the larger
conducting surface (at least for the outer grounding sheath; if you used
a balanced line conductors wouldn't necessarily change). The skin effect
at higher and higher frequencies should also be aided by this.

> 
> > better dielectric. The tough part is that as you go up in frequency, to
> > shorter wavelengths, smaller and smaller physical distortions of the
> > line begin to cause more profound defects to the desired resonant
> > frequency, so the nick in the copper, or the slight kink from stepping
> > on the cable, will end up hurting more.
> 
> Not resonant frequency, but the impedance of the line.

Most of the statements about a line's impedance or swr assume a uniform
line. In reality, imperfections can alter a local point that is more
lossy than others. If one were to measure the swr of a single frequency
signal at multiple points along a cable, a proper cable would show the
same swr at all points; if the cable is less than perfect, even though
the transmitter connector shows perfection, then swr would change at
various points along the cable. It is possible to fool either the
transmitter or the antenna so they can work at optimum, but still have a
lossy cable (the use of capacitors and inductors to perform tuning is
inferior to cables and antennas that actually do what is requested
without outside tuning sources). But the real point of all that is that
people often assume they can kink the cable, or bend it, or loop it in
coils, and it won't hurt anything; as the frequency goes up, the quality
of the cable makes a much bigger difference, and quality is not just the
physical product, but also how carefully it is placed. The GHz range is
particularly susceptible to problems when it is kinked, coiled, or
stepped on and distorted.

D. Stimits, stimits at idcomm.com

> 
> - Wayde
>   (wallen at lug.boulder.co.us)
> 
> _______________________________________________
> Web Page:  http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug



More information about the LUG mailing list