[clue-tech] Practical D.I.Y. WiFi antenna info?

[Nate Duehr]

On 7/14/2010 10:39 AM, Bruce Ediger wrote:
> I need some guidance on the practicalities of WiFi antennas.
>
> Background:
> I bought an IBM a22m Thinkpad at a garage sale for $10.  It was in great
> shape, and I got it to boot Slackware 13.1.  I ended up downgrading to
> Slackware 9.1 for two reasons:
>
> 1. It has an ATI "Rage" card, and which XFree86 supports it beautifully,
>   	you can only use it as a VESA card under Xorg.
>
> 2. I have a PCMCIA WiFi card that I scavenged out of a non-working vintage
>   	WAP-11 access point.  Turns out it's an ACX100 based card, well supported
>   	by a Kernel 2.4 driver, but not so well by 2.6.
>
> I cracked off the two super-nonstanadard SMA (?) connectors on the PCMCIA
> card to get it to fit in the a22m's slot.  That leaves me with 2 empty,
> 5-hole spots for soldering in connectors.
>
> Strangely even without an antenna, I can get it to associate with my DD-WRT
> WiFi router, as long as I'm in the same room with it.
>    

Sure... RF leakage.  You can take the antenna off a hand-held radio and 
talk across the room, but without something that radiates the signal, 
two houses away you won't hear it... put the antenna back on, and 
someone will hear a good 5W transmitter at VHF/UHF about 1/2 mile away...

And this assumes the RF transmitter doesn't fry itself operating into a 
missing/open load.  Most of this super-low-power stuff is designed to 
back off on transmitter power when the antenna is missing.  Most 802.11 
cards have a printed circuit board "antenna" as well as a tap for adding 
an external antenna... so they "work" without having that external 
antenna plugged in.  PCMCIA tells me it's pretty old, and probably 
802.11b and may or may not have been designed to operate without an antenna.


> I tried making a co-linear antenna out of coathanger wire, like this one:
> http://martybugs.net/wireless/collinear.cgi
> but I didn't solder it, I just filed down one end to a taper, and jammed it
> into the PCB that sticks out of the PCMCIA slot.

So the article you reference doesn't appear to be written by someone 
with an RF engineering background, and he doesn't appear to have "swept" 
that antenna he built on a Vector Network Analyzer or similar gear.  He 
did that completely by theory, and didn't factor in the velocity factor 
of any of the conductors he used, etc.  Whether or not his antenna is 
ACTUALLY tuned to the 2.4 GHz band, is very very questionable.  You're 
also talking millimeters of difference between getting that antenna 
"right" and getting it "wrong".  A couple millimeters either way on 
those coils, or the wrong size on them, and who knows if your antenna is 
a proper 50 Ohm load to the transmitter, and/or whether or not it's 
really on-frequency.

So... before I get into the answers to your other questions, be aware 
that you could build TEN of those antennas, and the performance would 
probably vary wildly according to how accurate your build techniques 
are.  In fact, it'd almost be highly recommended, if you don't have test 
gear to sweep the antenna properly at 2.4 GHz. (And I own quite a bit of 
test gear, and none of it at the "hobby" price levels reaches up to 2.4 
GHz.  Microwaves are expensive to do RF engineering design for.)

>    That didn't work so great.
> I need some help with antenna practicalities like this:
>
> 1. Do you have to use hard-to-get LMR-400 coax, or does other coax work too?
>    

The issue at microwave frequencies is loss.  LMR-400 is the BOTTOM of my 
quality level list for 2 GHz microwaves... even then, keeping the run of 
coax as insanely short as possible, and using quality connectors is 
required.


> 2. Do you have to use ridiculously expensive N-type connectors, or can you
> use BNC or other connectors?
>    
BNC are pretty good at 2.4 GHz.  N are expensive, but they tend to be 
very low loss (if you buy from a reputable brand name... there are TONS 
of cheap Chinese knock-offs with cheap nickel plating, etc.)... Amphenol 
is my brand of choice on almost all connectors.  Andrews also.

> 3. When I'm trying things out, do I really need to solder an RF jack on the
> PCMCIA card, use an expensive pigtail, etc, or can I just jerry-rig it with
> alligator clips and stuff?
>    
Alligator clips are way too lossy at 2.4 GHz and introduce Impedance 
"bumps" that will throw off the tuning/happiness of the transmitter.  At 
these power levels, you won't damage anything, but it won't perform 
worth a damn.

> I can solder a little, but I'm an ex-aerospace structural engineer, and a
> software guy, not an electrical engineer.  Where can I go for help on
> these practicalities?  Googling doesn't seem to get me anywhere on this
> occasion, just a lot of feel-good fluff and ads trying to sell me stuff
> that costs more than the a22m laptop did.
>    

The ARRL Antenna Book is a great start... thousands of pages of reading 
on antenna theory, just to get the basics down.  I enjoyed it, but it 
took me months to read the whole thing.  And the book is over $70.  It's 
also quite weak in Microwave theory, since microwaves are considered 
"advanced", even for RF engineers.  Tuning is so critical at these 
frequencies because the physical RF wave itself is only millimeters tall...

A friend who's set numerous world-records for microwave communications 
up into the 100 GHz and beyond range, we all jokingly call "Dr. 
Millimeter-wave".

Now, back to practicality...

What are you trying to accomplish.  Does this laptop/PCMCIA card 
application need to connect to an AP that's 100 feet away, 1000, 10,000?

Give me a hint, and I can tell you what you can probably "get away 
with".  In the great world of microwave antenna experimentation, there's 
really no harm in trying a whole lot of things until something works for 
you... but I'm a little worried that if your budget is tight enough that 
an N-connector is too expensive, ruining a few with the soldering iron 
probably isn't the best way to spend your money...

My "favorite" 802.11 antenna "hack" is using a USB dongle in front of a 
dish... like this...

http://www.usbwifi.orconhosting.net.nz/usbscoop.jpg

No loss in the cable going to the transmitter (USB) and follows more 
sound RF principals than most of the "free WiFi antenna" *junk* 
engineering seen on the Net.  Find the focal point of the parabola, 
stick a USB stick there... done deal.

Life's too short to screw around with microwave antennas.  Unless you 
REALLY like it, or you're REALLY bored. Or unless the point is to 
learn...  ;-)

--
Nate Duehr, nate at natetech.com