Andrew Spurrier and John Samin VK1EME
We want to establish a Wireless LAN link between two houses in Canberra, a distance of around 1.8km. Here is a link to the Conifer Antenna Modification page. We researched the WLAN cards and found that the Cisco Aironet Wireless LAN card is the choice card because of its 100mw output and the best receiver sensitivity. The problem is that it is also the most expensive... We bought the Cisco AIR-PCM352 cards because they are the cheapest of the Cisco cards (at $375) and we thought that they had an external antenna output. But this was not the case. They are a PCMCIA card only, with two internal antennae and diversity reception and transmit. We needed the AIR-LMC352 or AIR-PCI352 but of course these are hundreds of dollars more expensive again...
Our choices now were to return the cards and get a WAVLAN card (only 30mW transmit power), or modify them to accept an external antenna. Well we can't return them because we opened one and voided the return policy. So of course we are going to modify our brand new $375 dollar PCMCIA cards!! And this is how we did it...
Beware! If you decide to follow any of the instructions outlined below, you do so at your own risk! I'm sure the Cisco Kid (and Puncho) won't accept the cards back if you stuff it..... We tested our cards for a couple of days, just to make sure we didn't have a dud.
Step 1.
This was the first card modified so we were a tad carefull... First we removed the cover off the transmit end of the PCMCIA card. It actually came off very easily, using my fingernail, you just need to get something under the cover and prize it off. It his held on with some sticky stuff that seems to stick back on quite well.
From the diagram you can see there are a couple of LEDs an antenna jack of some sort and a hole that shows a barcode from the internal PCB. Don't know what kind of antenna jack it is though it may be OK to use if you can get a connector for it. Most probably it is used for testing the unit. We couldn't get the plug, so surgery IS the only option.
Step 2.
A lot of thought went into opening this baby up, it had to be neat and precise - the PCMCIA card is delicate! In the end we decided to do it with a modified Soldering Iron. Just sharpen the tip with a file, so it is like a small hot knife. Have a look!
This is a 12volt Soldering Iron with removable tip. It was filed down to around 1mm thick. If you file it too thin it will bend. This happened, but the second attempt was better because the tip was wider. We attached the Soldering Iron to an adjustable 12volt power supply. Around 8volts was good, it didn't burn the plastic and it cut really well.
Step 3.
Ok Doctor it's surgery time!! First we opened up a little bit of the case, and took a peek, there were no components (or sardines) around so we opened up the whole thing, cutting just inside the edge so that the cover could be glued back on. The Solding Iron worked very well.
Step 4.
Once we opened up the case it was obvious that this card could be modified to accept an external antenna. In the picture above, just under the "Tracks to be Modified!" text, there are tracks that have been etched to accept another antenna jack. All that has to be done is cut the track to the biscuit antenna, rub off the green PCB coating from the ends of the track and solder on some minature coax. It may be a little hard to see in the picture, and these tracks are small so take great care! We used a hobby knife to cut about 1mm of the track just where it joins the antenna. It's much easier to see in the close up (below).
Step 5.
Solder on the minature coaxial cable. Use a very short length, this stuff is very lossy! Again you will be working with very small tracks, don't use too much heat or you may lift the track and stuff the whole thing...
The PCB track etching is the same on both sides. You can see how the center track on the left side is connected to the antenna. We cut the track (on the right hand side) just where it meets the antenna. We also cut out a channel for the coax to exit the case once the lid was back on.
Step 6.
The coax is a little to high to allow the cover to sit back on properly. We decided to put some heat shrink over the end to hold it on neatly. Before you shrink the plastic tube, wrap all exposed areas of the PCMCIA card in aluminium foil, use around 10 layers of it. The card probably won't want to be cooked in the process. Do the same to the coax pig tail, or the plastic coating will melt. Once the heat shrink was on, use a slightly moist cloth to absorb the heat away from the PCMCIA card. These cost $375 dollars each - I'm entitled to be a little paranoid!
This picture is of the second modified card - BEWDIFULL! And of course the second one was a lot easier... Notice the cable output is offset, we took the cable out from the middle in the first card and this made the soldering more difficult. Also the surgery wasn't so brutal in this one. We measured around the tracks, about 10mm square and cut this section out only. The track etching is symmetrical, so measure around the antenna jack you can see, and apply this to the other side. We made a small channel for the cable to exit though.
Star Log - 13 October 2001:
Of course! Both cards work as before, even without an external antenna attached. There seems to be a bug in the Cisco software. The control GUI lets you select an antenna (left, right or diversity), but this doesn't seem to work, the card status always says "diversity" regardless? Of course our idea was to use the antenna output we modified only (Is it Right or Left?). We connected the output pigtail to our modified Conifer Antenna (see this project) and the diversity seems to choose the loudest signal. We have only tested the link out over a couple of hundred meters so far and 11MBS was no problem.
Star Log - 14 October 2001:
Tested the link over a distance of 1.2km from Andrews house to a my car fitted out with the 12volt powered laptop. Both ends of the link used the modified Conifer Grid Dish Antenna. The nose cone of both Conifer Antennas came loose and we had to support the ends to keep them centered. I'm sure this isn't helping things. It seems the existing glue is breaking down and going mushy, I'm sure some Liquid Nails will solve this minor problem. Signal strength was a bit up and down at around 50%, but we later realised we were not pointing exactly at each other. Link speed was constant 11Mbs. Signal quality was in the 90's. We will try again next week end but these results were promising to start with.
Star Log - 19 October 2001:
Repaired the drooping nose cones on the Conifer Antennas. With a little heat the glue turns to treacle. I cleaned it off with some mineral turps, the applied a liberal amount of Liquid Nails. 24 hours later they are rock solid. Tested the link over 2.5 kms (measured using the street directory), line of sight and established the link no problems. With the repaired antennas the signal strength was a constant 90%, signal quality 97% at 11MBS. Easily did voice over IP (VOIP), ping times <10ms. Looks like these mods have really worked well. Seeing we have no way of measuring the efficiency of the antenna modifications, I'm very happy with the results.
Star Log - 21 October 2001:
Tested the link between our houses. We DON'T have "Line of Sight", so this was going really going to test our comms hopes. I have a 10m antenna mast already installed at my house for my HF antennas. Andrew's house is easily around 20m lower than the hill that separates us. I put one of the modified Conifer Antennas at the top of my mast. There is a 15m heliax run to my laptop. Andrew sat up on his roof with the laptop and the other Conifer Antenna on a camera tripod. We knew we could easily get the 1.8km LOS distance but were amazed to find that we could establish a link without line of sight. The signal strength reported by the cards was 30% (fair) with a link quality of 93% good enough for 11MBS link speed. With a decent mast at Andrew's I'm sure we are going to do even better. We seem to be getting some major reflections off houses. The antenna must be pointed at a corregated iron roof on the top of the hill to get the best signal....
Star Log - January 2002:
Well at last we have our masts up and we have a permanent 11Mbps link. We have installed Mandrake Linux onto laptops, and use these for the Cisco Wavelan cards. The Laptops run the security and act as gateways to the other computers on our network. Also now that we have seen some heavy rain, we have proved that the link is stable in most weather conditions (only fog to go... ). Signal strength varies from 56% to 48% during the day, and I've seen it as low as 26% during a torrential downpour. Link quality ranges from 92% to 98%. Pretty good results considering we don't have line of sight and are transmitting with 100mW of RF!
Questions?
Email: jwsamin@mrx.com.au
13 October 2001
:-p