On Tue, 03 Aug 2004 09:59:08 GMT, John S <(E-Mail Removed)> wrote:
>Not generally a fact but you are not the first to make this common
>mistake!
I never make mistakes (that I can't rationalize or excuse).
>Most cheapie client radios (PCMCIA & USB) have much poorer receiver
>sensitivity (require higher input power for the same BER) the better AP
>receivers. Thus, higher AP transmit power is needed to provide the same
>range in the AP--->Client direction compared to range in the other
>direction.
Really? I've made quite a few field strength measurements of various
wireless devices. Since many devices do not have external antenna
connectors, measurements are a bit tricky. They're also difficult to
duplicate. For example, my "reference" radio will yield about a
+/-2dB variation in field strength. The variations from unit to unit
of the same manufactory and model is about the same variation. I'll
admit that my BER tester is home made and subject to some interpreted
results and that my indoor test range is far from idea. They're more
consistant when I plug directly into an onboard connector, but not by
much.
>For example, peruse receiver sensitivities listed here:
>http://freenetworks.org/moin/index.c...iveSensitivity
>Compare sensitivities of the 200mW Demarc or Senoa/EnGenius cards to the
>run of the mill card like the *VERY* popular 32mW
>Lucent/Agere/Proxim/DellTrueMobile/Diamond(YDI Karlnet) Orinoco Classic
>Gold/Silver Card.
Umm... You left out Avaya and Wavelan. What I've seen with one
Enginius/Seneo card is something like what this user found:
http://lists.nocat.net/pipermail/noc...ry/001475.html
However, others have found the opposite:
http://lists.nocat.net/pipermail/noc...ry/001479.html
On the average, my Orinoco perhaps 1-2dB less sensitive than the Seneo
based cards, often are equal, and occassionally better. The bell
curves are quite wide with lots of overlap.
Incidentally, I don't trust some of those FreeNetworks numbers. I get
suspicious when all the Prism 2 and 2.5 based radios yield exactly the
same numbers without any variations. I've never seen that kind of
consistancy. I suspect some "fudging".
How about these numbers?
http://www.demarctech.com/techsuppor...mw-rx-test.htm
http://www.demarctech.com/techsuppor...eiver-test.pdf
Comparing:
Orinoco My test Orinoco Seneo DemarcTech
Prism 2 Orinoco FreeNetworks FreeNetworks
data sheet silver
11 -87 -85 -82 -89 -91
5.5 -90 -88 -87 -91 -93
2 -93 -90 -91 -93 -95
1 -95 -94 -94 -95 -96
So much for consistancy.
What's really fun is to plot the input signal level vs BER. One would
think it would be a straight line (on a semi-log graph) but it's not.
It has hickups, lumps, glitches, and general wierdness due to the
effects of digital noise and threshold effects on various circuits
(especially the data slicer). A dash of positive feedback and
assorted leakages provide other lumps. I'll see if I can find some
old data and post it. It's ugly. One number just doesn't do justice
to what's happening in the receiver department.
>At 11 Mbps the sensitivities of the cheapies are
>about 8dB worse.
Well, there's a reason why some cards are crap at 11Mbits/sec. The
CCK modulation scheme makes the signal very sensitive to digital noise
across the board. Ditigal crosstalk, noise, crap, and pickup are not
much of a problem 1 and 2Mbits/sec, but really become an issue at 5.5
and 11 due to the amplitude modulation content. Even cleaning the
boards make a big difference. I had one Orinoco Gold card that would
vary radically in rx sensitivity over several days operation. I
suspected condensation and humidity. So, I disassembled the case,
washed the board with an alcohol and trichloretheylene mix, baked it
dry, and re-ran the test. Perfection. I coated it with an acrylic
Humiseal concoction and it's been the more stable and reliable than
any of the others. Moral: Board leakage is a problem.
>Thus, 8dB higher AP transmit power is required for the same BER (range)
>in both directions if we use an AP with Demarc or Senoa/EnGenius quality
>receivers.
>Note that 200mW vs 32mW gets that 8dB!
That assumes the the Orinoco is -8dB worse sensitivity. That hasn't
been my experience.
>The Senoa uses the Prism 2.5 chipset which will work (unofficially) in
>most brands of APs, Bridges, and Repeaters that use a Prism chipset
>card. I and many others have used the Senoa and reported that range
>improvement in several systems (ZyWall 10W and Linksys WET11 for
>example). The advantage is that you only pay once (maybe $35
>difference) for the power improvement at the AP end vs paying for
>improvements at all the client radios.
Good point. However, that also applies to paying once for a better
antenna at the access point. I've watched the S/N ratio on both
access points and client radios to see what craps out first as they
get farther apart. This is mostly in outdoor WISP service. I find
that the access point receiver is the major culprit and bottleneck.
My guess(tm) is that this is mostly because the access point is more
susceptible to interference from other systems and secondarily to
generally crappy receives.
>> 2. 7 times the power is 8.5dB gain. You can get the same improvement
>> with a 1 lb coffee can antenna or decent omni with the benifit of
>> improving the range in both directions. Opening the 2wire box is
>> tricky, but not impossible.
>
>Not entirely correct.
>If we need an omni pattern for the AP, 8.5dB above a stock 3dBi is not
>easy or inexpensive to get and the pattern could be too flat.
Methinks the typical rubber ducky coaxial antenna found on many access
points is at best +2dBi. The PIFA antenna found in the 2Wire box is
no more than 0dBi gain. I said 8dBi can antenna, but if you wanna use
an 8.5dBi omni, the -3db vertical radiation pattern is about +/-20
degrees which is adequate for most applications. There are also
specialized omni's with patterns designed to deal with overhead
antennas:
http://www.maxrad.com/cgi/maxrad_pro...&catalog=10001
http://www.maxrad.com/cgi/images/pro...ss_antenna.pdf
I personally don't like omnis for WISP outdoor use and much prefer
sectored antennas, which solve the downtilt problem.
>But, I
>agree that approx 3-4 dB improvement is easy.
>But if we put antennas at all the clients the cost is much higher.
I'm suggesting a better antenna as an alternative to a higher power
transmitter and of course not touching the client radios.
>So the bottom line is that for the very common situation I described
>above, an 8dB AP power increase is going to get the corresponding two+
>times extra range (in free space).
There's no such thing as free space or a free lunch. All other things
remaining constant, you might get double the range, but only if your
Seneo receiver is as sensitive as you claim. That's hasn't exactly
been my experience.
>Of course this "burn through" may interfere and PO your neighbor but an
>omni at the AP would do the *same* thing.
Again, I didn't suggest an omni. Few outdoor installations require or
desire an omnidirectional pattern. Directionality is a good thing for
reducing interference both generated and received. A high gain omni
with a good view of the world is going to pickup as much interference
as it generates. I have lots of problems with access points in tall
glass buildings hearing literally 100 other access points.
>And, I agree that the ideal (desired) situation would also include an
>antenna improvement for added range in *both* directions like you
>stated. And antennas with high directivity are desired if the situation
>allows that.
Ah, we do agree on something.
Actually, I'm doing more hair splitting than contradicting your
statements. In general, they're correct. It's the details that I
disagree with.
--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831.336.2558 voice
http://www.LearnByDestroying.com
#
(E-Mail Removed)
# 831.421.6491 digital_pager
(E-Mail Removed) AE6KS
Similar Threads
Linear Mode
