Comberton WiFi Project

Menu


Index

Summary of Linksys WAP11 facts

Glossary

Interesting facts

Musings on Ionica dish

External Links



Back to index

I've decided to put up a page of all the facts I know about the Linksys WAP11.

The aim of this project, started on 29/04/2002, and finished on 16/06/2002, was to build a link between the PCs in my house and in my friend Jon's house. Jon lives about 550m from where I am (as the crow flies), across a large communal sports field, so running some CATV cable wasn't going to work. We settled for an 802.11b bridge link between our houses, using a Linksys WAP11 box and some home made aerials (not the pringles can aerial, but something similar). The link is now up and working, which is pleasing.

The report below is a diary in reverse chronological order covering all the problems we encountered and overcame in the process of getting this to work. There are also some useful hints and tips for 802.11b hobbyists and Linksys WAP11 users.


Sunday, June 16, 2002

Success! After a weekend of work, Jon and I can now ping each other. As a test we copied some files from one computer to another, and the actual data transfer rate is about 600 kbps - perfectly good enough for networked games, or playing streamed mp3s off each other's computers.

If you have any questions or comments, or you are interested in Comberton wireless LAN stuff, please feel free to email me at keir @ ongar.org (remove spaces from preceding email address - they are added to foil spammers who trawl for email addresses on the web).

Below is some advice for others out there trying to get Linksys kit to work like this:

  1. The antenna connectors are not well identified in the manual. We performed some tests and found that Antenna 1 (or Antenna A) is the connector on the left, and Antenna 2 (or Antenna B) is the connector on the right, as you look at the Linksys WAP11 from the front (where all the lights are). See here for a picture
  2. Get the Atmel SNMP management tool from this location and download Atmel_SNMP_manager_v1.743.exe. Note that this is a Win9X/NT program. It has a lot more useful features than the management tool that comes with the Linksys box, including a radio test item that allows you to test channels at various strengths to see if packets are getting through.
  3. Make sure your tin can aerials are pointing directly at each other. If they are more than 5 degrees out, the signal received drops off sharply.
  4. We think there is an error in the auto fallback datatransfer rate. Start by using the Atmel configuration tool to set both Linksys boxes to only send/receive at 1 Mbps (and uncheck the auto fallback box). Then check 2 Mbps and try again, until you find the maximum data transfer rate that works for you. (This one held us up for a day, and we were beginning to despair of the whole project working!)
  5. The TCP/IP stack in the boxes is poor - so test by pinging PCs attached to the boxes. We found that if you pinged the Linksys WAP11 with more than 512 bytes at 1 Mbps tranfer rate, that there was a 25% packet loss, but from PC to PC 1 kB packets were fine.

Here is a picture of the tin can aerial on my roof. Jon has his up a tree!

Here is a parts list of some of the more esoteric components

  • Linksys WAP11 Wireless Network Access Point
  • N-Female to RTNC-Female (Plug) Adapter part no. HXNFRTF from www.hyperlinktech.com (about $12 each, but $100 minimum export order, so get someone to buy them in the US and bring them back across the pond for you! These are needed to connect the Linksys aerial connector to N-type connectors for the coaxial cable lead).
  • 10m of URM67 spec coaxial cable (which I think is Thick Ethernet cable) but is pretty low loss 24dB/100m. This connects the Linksys to the tin can aerial.
  • RS stock no. 455-769 N-type silver-plated chassis socket for the tin can aerial
  • RS stock no. 404-941 N-type nickel-plated clamp plug for both ends of URM67 cable.

Incidentally, if you live in Comberton and are interested in joining the project, please contact me!


Friday, June 14, 2002

Started working on the tin can aerial. It's 8.4 cm in diameter, which means the copper wire aerial rod in the can has to be about 6 cm from the back of the can (this is 1/4 of the guide wavelength). And the copper wire aerial rod has to be 3.07 long (due to the wavelength of the signals received).

This can all be calculated from things such as frequency * wavelength = speed of light and standing wave stuff, but I couldn't be bothered, and went to this site, which has a Javascript calculator to do it all for you. You don't need to think these days - someone's already done it for you.

The tin I used contained some Cains Jubilee Ale (5% ABV), with a lovely picture of the Queen on the front and lots of blurb on the back about beer and festivities. You can see the connector mounted on the side.

Here's a picture of the inside of the tin. Soldering the copper rod into the connector was a nightmare with my little soldering iron. My one tip is to ensure you tin the connector and the rod properly before putting the two together. Tinning the copper rod takes especially long.

And here is the final product, sprayed 'underside of jet fighter' grey. I didn't really want a bright orange can stuck to the outside of my house, so this acts as camoflage and I think it makes it look cooler too.


Friday, June 7, 2002

I chiselled out a groove in the hallway wall to run power up to the attic, and ran CATV cable up there two. Now the Linksys box sits up by the gable of the house on a shelf, and I can ping it from the living room. All I need is to make the tin can aerial, mount it on a pole outside and connect it with 50 Ohm connectors and co-axial cable to the Linksys.

Jon and I had to order a special convertor connector from the states - the Linksys has reverse polarity sleeved BNC male connectors on the back, and the female version isn't sold in the UK. So that's one connector needed - a reverse polarity TNC-type female connector to female N-type connector. Also two male N-type clamp plugs for the cable and a flat-top female N-type connector for the tin can aerial.

A note about the fact that all these things are rated 50 Ohms impedance - I think this doesn't have quite the same relevance as resistance (which tells you what the voltage drop over a part of the circuit will be). Instead its about signal reflection - if the parts don't have 50 Ohm inpedence, then echoes of the received signal bounce back from the connectors, screwing up your signal.


Sunday, May 5, 2002

Today I blatted my PC downstairs, wiping Windows NT4.0 and installing Windows 98. Why did I do this? Because I wanted to see Netmeeting running over the wireless connection (I have both linksys devices in my house, and we managed to get ping and file transfer working across them). But the result appears to be a performance drop from 5 Mbps to about 0.005 Mbps. Everything seems to be communicating, only very, very slowly. How annoying!

Update: a couple of reboots and a change from half-duplex to duplex on the LAN card, and the problem appears to be solved. Transfer rates are back to 5-8 Mbps.


Saturday, May 4, 2002

Cabling day. Jon, Jo and I all spent most of a sunny afternoon running cabling down the northern fence of their garden. It is about 60m from their house to the end of the garden by the recreation ground (the diagonally striped area on the map below), and because attenuation is terrible on a coaxial cable from an aerial, we decided to run mains electricity to the garden shed, and an ethernet cable. The idea is that an aerial cable runs from the tin can in the tree to the shed, and the linksys WAP11 is locked up in there. Then the signal is converted to TCP/IP by the linksys, and is piped down the ethernet wire and into the house. The added benefit is that the shed will have power for Jo to use when working down the end of the garden.

We also drove into town to look for aerial brackets and piping for my end of the connection. B&Q has a suitable bracket, and McKays can provide me with an aluminium pipe. We're looking for the aerial portion of the link to be made out of two tin cans, as described here. A hundred years of radio and it looks like we're back to communicating using cans and "string"!


Thursday, May 2, 2002

The Linksys WAP11 devices arrived today. Here's what one looks like from the front:

There are three lights - power on (red), activity over wireless detected (green) and link to ethernet detected (orange).

 

And here's the back:

At the back you can see the aerial connecter (aerial removed - this is Antenna B, or Antenna 2, aerial left on - this is Antenna A, or Antenna 2). The aerial connector on the linksys appears to be a TNC (male) connector). You can also see the power socket, a USB connector and the ethernet socket. It's quite a simple device really - apart from a little IP configuration you just plug it in and it works.


Wednesday, May 1, 2002

Today Jon and I walked out the route that we're hoping to communicate across using Linksys WAP11 web access points. We've checked on the map, and it looks like the distance is about 550m. This should be feasible - the specs for the WAP11 claim up to 500m outdoors, and I assume this is from a PC WLAN card to the WAP11. Furthermore, Jon and I are going to be using directional antennae. Some web reports claim that distances of up to 1000m have been covered.

We appear to be rather lucky - Jon climbed the aerial tree in his new garden, and went up a ladder put against the side of my house, and it looks like we'll have line of sight. This is stroke of luck, as there are some big oak trees, one house and an ash tree that could have been in the way.

We've ordered two Linksys WAP11 web access points, 100m of mains cable, about 20m of 10mm coaxial cable (50 ohms impedence), and 100m of CAT5 cable. It looks like the project is soon to start in earnest.


Tuesday, April 30, 2002

We've both spent a lot of time on the web, reading articles on 802.11b wireless access point devices, and articles like this on how to increase their outdoor range. It looks like 500m or so should be feasible, and that's what we estimate the distance between our houses is.

After looking at a whole host of web access point devices, it looks like the Linksys WAP11 is the device to go for, for the following reasons:

  • It's antennae can be detached, which means we can fit our own directional ones.
  • It can perform bridging, which is a requirement for our machines to be able to talk to one another from wireless access point to wireless access point
  • It comes with monitoring software so we can check its performance
  • It is a popular choice on the web, where there are reports of others succeeding in doing what we hope to do

There were many pitfalls - we were nearly sucked into buying a Linksys BEFW11S4, which appeared to be a WAP11 with extra 4 port switch, DSL/cable modem connector and a built-in firewall, but it turns out that this device can't perform the bridging we want.


Monday, April 29, 2002

I've finished the punt, and there's nothing to do project-wise. So I started talking to Jon, who's moved to Comberton recently. He pointed out that there's a lot of cool wireless LAN stuff being talked about on the web, and noted that our houses aren't too far apart. So perhaps getting a wireless connection between our PCs could be a fun new project.


Mission: a network connection between my PCs and Jon's PCs across the Comberton Recreation Field that is better than a modem connection.