Unfortunately, I haven't had the opportunity to tear open a senao card yet - 
I am considering repeating my tests on the senao because it looks so promising
on paper.  Spec sheets don't always tell the whole story.  No vendor talks
about the shape of their RF passband, nor which mixing products end up in 
their IF band.  I've yet to see a vendor who honestly pointed out their 
product's shortcommings in print - those are only exposed after purchase.
Fortunately, I have the advantage of several RF benches for proving hardware.

See:
http://www2.uwave.com/labrent/index.html           (lab area)
or
http://www.labrent.com        (general services offered)

About two years ago, I was hired as a consultant to evaluate the RF decks
used in several pcmcia cards (which were the heart of many larger designs
from access points to rackmount bridges that accommodated the pcmcia card).
At first, the test plan was just to look at transmit/receive performance, but it
later grew to include temperature and voltage sensitivity, as well as toleration
of drift in the reference crystal and S-parameter testing of the antenna port
under transmit and receive conditions.  Several antenna systems, coaxial 
cable systems, and amplifiers were also tested separately and ranked.  At
the time many cards did not have external RF ports (like the early cisco 350s)
-- these were opened up with an end mill to expose the built in antenna, which
was replaced with pigtail coax.

The general results were that transmit power fell within -3 to +1 dB of the  
product's specifications, but on testing the receivers against a 
known stable source, the cisco hardware showed a +6 to +7 dB better performance
than its "spec" (perhaps this is just manufacturing margin that was being 
"left on the plate" as it's called) - Most other vendors were quoting specs
from the chipset vendor and not taking into consideration their RF and IF 
stages, which reduced their (chipset theoretical) performance to -3 to -6 dB
below "spec" (which most customers wouldn't notice in a residential or light
commercial space).  Several of these are being coppied in china by shops that 
were making VCR remote controlls the week before.  They didn't even match
the 50 ohm impedance of the antenna system.   The difference between the 
"best" and "worst" tolerance from "spec" was > 12 dB - in additon, a few of 
the cards published a slightly worse "spec" so the absolute difference from 
absolute best to absolute worst was substantial (at the time, most all were 30mW
xmit cards except the cisco, which was 100mW with introduction of the 350 
series.  Cisco also improved their receive "spec" by 1 dB.)

I also ran susceptibility tests to intentional interferance (something that 
isn't in the manufacturer's spec at all) by summing a calibrated signal source 
with an artificial noise floor that could be raised or lowered programatically.
This is where the chipsets either showed their strength or  
This test was developed from a site survey that I did for a customer who 
had a line of site view to over 4000 1W ricocet 2.4 Ghz transmitters. 
While each one was technically within FCC compliance rules for emission, the
"air" summation of all 4000 created a -85 dBm signal on a spectrum analyzer 
using a -5 dBi antenna system at 3000 feet MSL - the peak/valley structure 
showed a clear channelization which we tracked back to ricochet's division 
of the 2400-2483 Mhz band (they assigned a unique number of channels, which
matched the number of peaks we found within 2400-2483.  At high altitudes,
Metricom had single handedly polluted the entire 2.4 Ghz band from edge to edge.
Susceptibility to co-channel interferance was a key factor in picking hardware
to be used in densly populated areas.

With regards to senao, I have 2-3 of them inside my network talking to 
existing cisco hardware where they became direct replacements for the 
previous cisco client card, however on my longest links (over 14 miles LOS)
the cisco was the only product which would register and maintain its link 
without downtime.  Most other manufacturers products placed at both ends 
couldn't even register, much less pass error free data.  I have 2 senao's
that are only 7.x miles from a cisco access point and working fine.

Disclosure - I do not, nor have I ever, worked for an 802.11 card manufacturer.
All of the facts stated are from personal experience and lab tests I have done. 
I have not thoroughly analyzed any senao product as of this time.  I just 
use them and they work for most applications.

> In a recent post Everett ([EMAIL PROTECTED]) makes a very interesting comment:
> 
> "There is a subtle advantage to using cisco's hardware (and I've tested a
> wide range of mfgr's products) - the aironet/350 series is much more 
> sensitive on the rx side and more immune to adjacent channel power 
> problems.  Compared to the inexpensive 802.11 brands, it can have
> as much as a 10 dB advantage on the receive end."
> 
> As a ham who has done a lot of weak-signal work I am always concerned about 
> optimizing receiver performance.  Based on published specs for receiver sensitivity, 
> most of the 802.11b cards seem to be in the same ballpark.  The Senao cards I use 
> (Senao/EnGenius NL2511CD Plus EXT2) have about the same receive power required for a 
> given BER for the various modulations as the cisoc cards.  Are these numbers not 
> particularly valid?  I don't have a bench to really test this stuff so I have been 
> relying on specs.  (I know, that is dangerous.)
> -- 
> 
> Brian Lloyd                              6501 Red Hook Plaza, Suite 201
> [EMAIL PROTECTED]                          St. Thomas, VI 00802
> +1.340.998.9447 - voice                  +1.360.838.9669 - fax
> GMT-4
> 
> --
> general wireless list, a bawug thing <http://www.bawug.org/>
> [un]subscribe: http://lists.bawug.org/mailman/listinfo/wireless
> 

--
general wireless list, a bawug thing <http://www.bawug.org/>
[un]subscribe: http://lists.bawug.org/mailman/listinfo/wireless

Reply via email to