[CUWiN-Dev] New Platform

[Quantum Scientific]

Here's Belkin's cardbus pre-N product:

http://quantum-sci.com/images/preN-cardbus-f.jpg
http://quantum-sci.com/images/preN-cardbus-b.jpg

Again, this is straight from the Airgo reference design.  Believe it or not, 
this is a very simple for a wireless card.  The big chip at top is an ASIC 
which bridges cardbus (PCI) to its internal PCI bus, does the protocol and 
channel arbitration magic, and controls the transceiver chips.  The EEP is 
what it boots to, and the PLD likely is the hardware IPSec and a few other 
functions.  The RF synth makes stable freqs from the crystal, suitable for 
802.11a/b/g.

There are three Tx/Rx channels, separated into RF domains by the can.  First 
time I've seen anyone go to the trouble of separating them.  The transceivers 
handle physical layer error correction, arbitration, etc, and can do a, b, & 
g.  The Tx/Rx switches control their respective antennae, and are in turn 
controlled by the baseband chip.  

Each physical channel has an Rx low-noise-amp transistor (which I didn't 
bother to look up) and the two outside channels have a power amp for Tx.  
Each PA can boost by 33dB, up to the level of +19dBm (80mW) in OFDM 
(802.11g), the most powerful commodity PA I've seen.  The datasheet doesn't 
have figures for 802.11b, which implies Pout is about the same for that.  
Most PA's can do 'b' much better than 'g', but this one is about as good as 
I've seen for 'g'.

This card automatically locates a clear WiFi channel, and uses its three 
physical channels to operate on that, so up to 108Mbps on three channels, 
instead of 9, as with Atheros.  On two physical channels Tx is hot, and on 
one cool.  They all have the same Rx sensitivity.  The meandering line 
antennae (a recent development) are horizontally polarized, and the inductive 
stub vertically.  This allows pickup of signals which have bounced off of 
obstructions and are twisted, a major advancement, I say.

There's room for three more channels, intended to go to SMA connectors, but 
these auxiliary channels' components are unpopulated.  However, replacing the 
factory calibration jacks with U.FL jacks would allow external antennae for 
the existing physical channels.  For this system to be effective, you must 
have three antennae... what a mess of wires for a notebook.  It is pretty 
clear that the antennae are not a directed beam system.  

I do not know what modes this card can be put into.  HostAP (Master) may not 
be available.  This may depend mostly on the driver.  I only know od winduhs 
drivers for this card.  This can be used in Linux with ndiswrapper, but I 
don't know about NetBSD.

Conclusion:  This may possibly not give as good a distance as a 200mW 'b' 
card, but it beats the hell out of it for distance in 'g', and in speed and 
distance in a multipath environment.  I didn't unsolder the can for the 
router, which I shared here recently, but there's good reason to believe it 
is designed the same as this.

Best,

Carl Cook