!! Location: Narre Warren North {http://www.jeack.com.au/~glenns/images/conifer_rotator_s.jpg Rotatable Conifer} I have a pretty good take off to the South towards Narre Warren and West to Endeavour Hills. Using a DWL-650 card connected to a 18dBi ex Galaxy antenna on an antenna rotator. Note that the antenna polarisation is __vertical__. Is there anyone else active in the area? (http://www.jeack.com.au/~glenns/wireless/conifer_rotator.jpg GMI Antenna) !! DWL-650 Card These are my mods to the DWL-650, I used a short length of RG188 coax from card as it is very flexible and small in diameter (also teflon, so wont melt). An SMA connector is used on the tail. These are heaps more rugged than the MMCX connectors. (http://www.jeack.com.au/~glenns/wireless/dlink_mod1.jpg Dlink mods) I was a bit worried about the strain that the coax tail placed on the card, so I made a little strain relief from plastic and bolted the SMA socket on the tail to the plastic. The idea behind using plastic for the SMA socket mount was to protect the card from damage. Should the plastic strain relief become caught on something solid, it will break, rather than break the PCMCIA card protrusion. (http://www.jeack.com.au/~glenns/wireless/dlink_sma2.jpg Card Strain Relief ) There are some details on the web showing how to add a tail to the DWL-650 and also stating that it does not matter which diversity antenna connection is re routed to the tail. My experience shows it does matter. When I re routed the antenna capacitor as shown in the earlier image, I found that although the card received signals well through the tail, I could not see any RF power on a power meter through the tail. When I made an RF sniffer loop and placed it over the internal antenna, I noted RF only from the internal antenna. To generate a constant RF transmission WLAN Expert was used in Antenna Test mode. Of course the RF transmit signal is bursty, but at least it is regular enough to measure. The coax tail was then re connected to the other internal antenna and the previous connection restored to the other internal antenna. This time RF output was measured on the coax tail. The card was also receiving fine through the tail. My conclusion is that although both internal antenna are used for diversity receive, only one antenna does the transmitting. This in my case is the __opposite__ antenna connection as shown in the previous image. It is possible that the card behaves differently in other modes, but I have only detected RF transmission from the one internal antenna so far. !! Biquad Antenna Made a biquad antenna and covered it with radome made from a hobby box. It is based on the design described in http://www.martybugs.net/wireless/biquad/ The antenna matches and performs very well. Here are some photos: (http://www.jeack.com.au/~glenns/wireless/biquad1.jpg Biquad pic 1) (http://www.jeack.com.au/~glenns/wireless/biquad2.jpg Biquad pic 2) (http://www.jeack.com.au/~glenns/wireless/biquad3.jpg Biquad pic 3) !! Helical Antenna Put together a 12 turn helical antenna based on Remco den Besten's design from http://helix.remco.tk/ and did some tests on air. So far it doesn't perform so well taking in to account that its circular polarisation will be down 3dB from a linearly polarised antenna. Will have to check it out on the network analyser. (http://www.jeack.com.au/~glenns/wireless/helix.jpg Helical pic) Redesigned the match on the helical and now the thing seems to work as expected. Not too sure on the basis for the design for the old match. The match I used is based on a transmission line fabricated from a curved strip of copper foil running almost parallel to the reflector. The theory and implementation is described in the ARRL handbook. (http://www.jeack.com.au/~glenns/wireless/new_helix_match.jpg Helical match) !! Aerosol Questions Using Aerosol for sniffing packets but have a few questions on it as I don't have any documentation. Does anyone have experience with this program? The readme says to set the SSID to nothing for it to work properly, but when I do this the PC slows to a crawl and eventually freezes. 20030428: Sorted it out. Turns out that if you set the SSID to blank while in adhoc things go crazy. A blank SSID while in infrastructure mode is OK. Also WLAN Expert utility leaves the driver in a funny state after it exits. Going to the DLink utility and pressing rescan seems to recover it. Also driver DLink V1.32 only works, the latest from web site doesn't (on my Toshiba 410CS anyway). When using V1.32, must modify the netcw10.inf file after driver installation by adding code so that it works with standard PCMCIA slot not Cardbus slot as driver assumes. Ancient Toshiba 410CS does not have Cardbus slot. Is nothing simple? (http://www.stolenshoes.net/sniph/index.html Aerosol ) !! Polarisation Standardisation Is there a standard for polarisation? Have seen some wireless users with Conifer antennas vertically polarised and some horizontally. There is a huge penality for cross polarisation (up to 20dB). The omnis such as vertically mounted waveguides are horizontal polarisation while coaxial co-linear omnis are vertically polarised. !! Nodes in Range Now can connect to (http://melbourne.wireless.org.au/wiki/?NodeGXT NodeGXT ), over a 7.3km path. Connection is only 1-2Mbit/s, but it's there! Some optimisation of the setup at either end should improve the connection. Using 24dBi Galaxy antenna at either end. Have managed to connect to (http://melbourne.wireless.org.au/wiki/?NodeACD NodeACD ) after a polarisation switch to vertical and a temporary relocation of the antenna. The trunk of a big cypress tree was in the way. Should be enough signal to use after some optimisation of feeder and antenna. [NodeGSJ ] may be LOS. Here is a photo in the direction of GSJ with 30x optical zoom at a distance of 5km. (http://www.jeack.com.au/~glenns/wireless/NodeGSJ_from_GMI.jpg View to GSJ) Here are some path profiles done with Pathloss for the two nodes. Note that the terrain data used by the package has a resolution of about 200 metres so it's probably not that accurate for the 1km path. (http://www.jeack.com.au/~glenns/wireless/path_acd_gmi.jpg Path to ACD) (http://www.jeack.com.au/~glenns/wireless/path_gsj_gmi.jpg Path to GSJ) 2003-07-13: Link to (http://melbourne.wireless.org.au/wiki/?NodeACD NodeACD )has now been stable for 24hrs with -78dBm signals at NodeGMI. Reported link speed of between 5.5 and 11Mbit/sec. 2003-11-29: !! SVEC AP mounted on mast To reduce coax losses the SVEC AP is now mounted on the mast behind the antenna. This results in much more signal on the link between GMI and GXT. The box is a Jaycar weatherproof enclosure with dimensions about 165x220x55mm. Power over Ethernet supplies 12VDC to the enclosure where it is converted to 5VDC by a switchmode DC converter. A switching convertor was chosen so as to minimise heat build up within the enclosure. Has been running for a couple of days with only slight temp increase inside although it will need some sort of sun screen to shade direct sunlight from the enclosure in the warmer months. See picture: (http://www.jeack.com.au/~glenns/wireless/svecAP_in_a_box.jpg SVEC in box) Well it was 33 degrees today and the thermometer probe inside the box only reported a max of 40 degrees. Guess it will survive the summer then. (http://www.jeack.com.au/~glenns/images/ap2.jpg SVEC Temperature) 2003-12-12: !! New dish feed for parabolic This is an experimental feed made from: - half inch copper plumbing pipe - brass tube from hobby shop - RG214 coax - disk of brass sheet for reflector Seems to work OK so far, will compare it with Hills dipole feed. (http://jeack.com.au/~glenns/images/feed_1.jpg Dipole dish feed 1) (http://jeack.com.au/~glenns/images/feed_2.jpg Dipole dish feed 2) (http://jeack.com.au/~glenns/images/feed_3.jpg Dipole dish feed 2) =========================================================================== !! Comments: