"Charlie" <ctuckey@gmail.com> hath wroth:

>As posted before, we are trying to set up a test bed with a bunch of
>Linksys WRT54GL routers. I would like to be able to cable the routers
>together so that they are not using OTA to communicate. Ideally, there
>would be an adjustable attenuator in the cable as well so we could
>simulate distance.

Yep. Been there and done that. What you want is a portable RF screen
room or shielded box. Since you're only trying to block a small range
of frequencies, this should be easy. See:
<http://www.smallnetbuilder.com/content/view/26746/100/>
<http://www.azimuthsystems.com/index.asp?p=198>
<http://www.ramseyelectronics.com/te/default.asp?page=rften>
<http://www.saelig.com/miva/merchant.mvc?Screen=PROD&Product_Code=TE001&Catego ry_Code=TE>
for examples. I'll spare you my photograph of the mylar foil and
aluminum duct tape "tent" that I constructed for my first attempt at
shielding. Search Google for "RF test enclosure".

>In a naive attempt at doing this we tried wrapping the routers in
>tinfoil (after removing the antennae) to isolate them. This worked
>pretty well.

Until they overheated. That's why I gave up on foil. Also, there was
considerable leakage from the CAT5 cables on some routers.

>Then we tried wrapping a cardboard tube in tinfoil and
>connecting the routers by placing the tube over the antennae RP-TNC
>connectors on the routers. (the naive part!) This didn't work at all -
>no signals got through.

Chuckle.

>More research on the internet found me at this site: http://
>www.wirelessdesignmag.com/ShowPR.aspx?
>PUBCODE=055&ACCT=0031546&ISSUE=0610&RELTYPE=PR&OR IGRELTYPE=FE&PRODCODE=W0240&PRODLETT=A
>which shows a test setup of wireless nodes connected with 'RF cable'.

In the future, please wrap your URL's with <> so that they don't wrap.
<http://www.wirelessdesignmag.com/ShowPR.aspx?CommonCount=0PUBCODE=055&ACCT=0031546& ISSUE=0610&RELTYPE=PR&ORIGRELTYPE=FE&PRODCODE=W024 0&PRODLETT=A>

>Would this actually work?

Sure. However, there's quite a bit of detail missing. There are no
combiners shown in the picture. You need some what to linearly
combine all the tx and rx signals. A simple resistive combiner will
do, or you can use a power splitter.
<http://www.hyperlinktech.com/web/signal_splitters.php>
Be sure to terminate any unused ports. I use an 8 port device.

Getting attenuators that don't leak and are accurate is more of a
challenge. You'll need to attentuate the maximum tx power down to the
baseline noise level. That's about:
+20dBm - -90dBm = 110dBm
range. That's much more than can be easily found with the common
switchable attenuator that might have perhaps 60dB of adjustable
range. Fortunately, you only need half the 110dBm of each leg of the
power splitter so a mess of common 60dB switchable attenuators will
barely work. However, I suggest also buying a mess of fixed
attenuators to make life easy on the attenuator.

One big problem will be calibration. You're going to have quite a bit
of hardware floating around that may not necessarily be accurate at
2.4GHz. I strongly suggest you spend the time calibrating the
attenuation of the various components at 2.4/5.7Ghz before you
discover that your switchable attenuators were previously used as
someone's dummy load.

>Assuming it does:
>Are attenuators required so that the output from one router doesn't
>damage the other router?

Oh yes. LOTS of attenuators. The typical wi-fi receiver has a very
limited dynamic range. My guess(tm) is that they can't really handle
signals much stronger than perhaps -20dBm before blocking, saturating,
or overloading. You might damage the receiver with anything over
perhaps +30dBm (1 watt), but that's not possible. I've done some
crude damage testing and found that most access points can handle at
least +20dBm before doing any damage. However, that's because the
diversity switch PIN diodes effectively limit the input power. I
don't know what will happen is the PIN switch wasn't there.

>What cable is best to use for this application?

Well, for TNC and SMA connectors, I use RG-400 and LMR-240.
For N connectors, it's all LMR-400 or LMR-400 ultraflex. The stuff is
really stiff. Make sure you use decent crimp connectors.

>Any other advice (note - we don't have any experience in
>this field)?

Yes. Find someone with microwave experience to help. Be prepared to
buy some rather expensive RF test equipment.

--
Jeff Liebermann jeffl@comix.santa-cruz.ca.us
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558