Ray,
N-type connectors are specified up to 10GHz and should have VERY LOW
LOSS - they are also available specified up to 18GHz but these are
seriously expensive. I use them with my ham set up at up to 10GHz with
no problems (other than the cost). The loss you get when using N-type
connectors is caused by a mismatch when assembling them with incorrect
co-axial cable. The impedance of a co-axial cable and/or connector is
formed by the ratio of outer to inner with a factor for the dielectric
(the stuff in between). Look closely at the N-type connector and
you'll see that it is machined to precisely take the transition from a
PTFE dielectric to air and there is then a carefully machined section
for the co-ax to fit in - and each connector is made for a particular
type (size) of cable as a result. If you don't use the right size
you'll get a discontinuity and a reflection (bad vswr).
The best connector to use for 802.11 purposes imho is the SMA type,
these work well up to 18GHz and, unlike the horrible reverse SMA
connectors used by some AP manufacturers (I gather to satisfy some FCC
requirement that the systems should not be easy to modify) the SMA
connectors are inexpensive. The best ones to use are those that are
designed for copper semi-rigid co-axial cable where the inner of the
cable forms the centre pin of the mail plug. The loss of a couple of
metres of semi-rigid and a couple of SMAs should be less than 1dB at
2.4GHz unless you really mess it up. The SMAs for use with flexible
co-ax are generally of the crimp type and although these work well you
need to know what you are doing to make them up properly and in any
case flexible co-ax is progressively worse at higher frequencies (I
never use it). Then if you need to use an N-Type to form the base of
your feed for a cantenna (basically you are making a co-ax to circular
waveguide transition) you can get an SMA to N-Type adapter that will
have very low loss as the impedance will be properly matched
throughout.
I'd fully recommend soldering the cans together in your cantenna - if
there is a gap in the join and the duct-tape is non-conducting you'll
form an iris in the circular waveguide and could get a surprising
amount of coupling through it (i.e. a sidelobe you didn't want).
Again, sorry that this is a little off-topic but hopefully someone will
find it interesting. I'll shut up now.
Regards
Ian
On 15 May 2004, at 21:12, Ray Dios Haque wrote:
--I had looked at various directional and gain antennas but soon realized
that at 2.4GHz the coax and connector losses are going to eat up most
of the gains unless you keep the coax real short.
Don't waste $50 on something that would cost you less than $3 to build. You
can make a very respectable cantenna with a couple of soup cans and a
connector. Check this page for a nice illustration of one ...
http://www.geocities.com/lincomatic/homebrewant.html
The dimensions are important! It should be 3.25 inches in diameter, and a
little more than 7 inches long. You can get away with connecting two cans
bottom to top to form a nice long barrel (I held mine together with duct
tape).
Just about every one I have seen uses the N type connector (CD/Radio Antenna
style). Those connectors will run you five bucks, and for their size have
lots of loss. Also as Dan mentioned, the longer the cable, the more loss
you will receive. I decided to go with BNC connectors myself, and I ended
up hacking apart a few wireless cards and adding a BNC connector to them.
It also helps to have a crimper that will do BNC connectors. I was able to
make a 3 foot cable, BNC at both ends with solid (not stranded) copper core
inside.
With my cantenna, and a line of sight path (no obstructions) I can sniff
signals from a full mile away (or more). The signal you get with a cantenna
can be deceiving, and it's hard to pinpoint exactly where it's coming from.
Also, you usually cannot send signals as far as you can receive them. In
other words, you may be able to sniff traffic from an AP, yet not join the
network.
-Ray
Tel: +44 1629 733800 Mob: +44 7717 220707 Fax: +44 1629 733807 E-mail: ibeeby@xxxxxxx
