Connecting over 400m. Bridge or Repeater mode?

You can get easily fooled by reflections. If the antenna has side lobes (as is very common with a yagi), it's even more difficult to play direction finder. What I find works is a bit crude. It's a rotating antenna designed for minimum side lobes. The antenna isn't anything wonderful at approx 6dBi gain and a 40 degree -3dB beamwidth. When spinning around, the detected signal and rotation are synchronized to display on an oscilloscope. You can see the signal direction by a solid peak on the scope. Reflections are very fuzzy and are easily recognized, while the direct signal is generally quite steady.

Before anyone suggest using doppler, please read these articles I scribbled many years ago on the subject:

formatting link
formatting link
that this is 1970's technology.

That works if you use your body to block the signal. Hold the USB radio near your chest and spin around. Look for a null. When you find it, the signal is coming from behind you.

Note the pattern at: |

formatting link
shows about a 4 degree boresight error on a typical coffee can antenna. It's not much, but it's there.

This is a more sane and reasonable horn antenna. 16dBi gain with built in stub tuner (for minimum vswr).

formatting link

Reply to
Jeff Liebermann
Loading thread data ...

Nope. It's a motorized rotating antenna. It's usually mounted on an antenna mast with tripod in the back of my pickup truck, or in various mutations that vary with size of the antenna and frequency. I used it for ham radio transmitter hunts and such, but haven't done much with it for many years. If you have a perfect signal source, no reflections, and no VSWR, rotating the antenna will display the exact antenna pattern on the oscilloscope screen. It's almost like radar. It works best when moving as the real signal shows up as a steady lobe on the pattern, while the reflections come and go at random. I intentionally did not describe my 2.4Ghz version because it's not finished, what I have doesn't work quite right, and I have illusions of selling it as a product. That will bring a new dimension to war driving.

Ballet experience is helpful. Actually, it's a common technique also used in ham radio transmitter hunts. Some of the local experts are amazing. Some wrap their radios in aluminium foil, exposing only a small part of the antenna, to reduce the sensitivity when close. Lots of other tricks. The problem with doing it using an 802.11b/g system is that the signal strength meter is often slothish and insufficiently granular. There is software that will give a much better RSSI reading, with good response time. Netstumbler will also do it if you increase the polling rate. The Lucent client manager also works well.

Well, it should work, but I've never tried it with a USB dongle. In a hallway, you will have many reflections which will probably obscure the direct signal. It works outdoors, but I've never tried it indoors. I don't think it will work.

Yes, in the coffee cup. It's described at the bottom of the page. SMA connector with 4mm of brass rod. A simple 1/4 wave feed. The coax is apparently not attached.

It's apparently not a construction article. I could grind out the numbers for the horn if you'd like. I just wanted to illustrate what a properly designed horn antenna looks like and how big a 16dBi gain horn would really look like. Basically, the horn aperature width controls the gain. The transition angle controls the bandwidth (smoother transitions or tapered horns have a wider bandwidth). The stub tuning screws tune for minimum VSWR. For ham radio satellite work, the bandwidth required is very small. Therefore, a rather abrupt transition angle will work just fine. For 802.11 bandwidths, the horn will be much longer.

Yep. Note that the W1GHZ waveguide feed articles are primarily intended to illuminate a dish antenna.

This can get messy fast. Basically, the gain of both the horn and the coffee can is mostly (not totally) dependent upon the aperature diameter. Equal size "mouths" of these antennas should have roughly equal amounts of gain. However, the horn has a much neater coax transition while the coffee can of equal diameter would spray RF all over the place from wall reflections.

Incidentally, selection of materials is important. The only reason the coffee can works is that the surface is tin coated. Skin effect causes all the RF to stay on the surface of the coffee can. Tin is non-magnetic so it's a tolerable RF conductor. However, a stove pipe is a different story. It's usually painted, not plated. The RF conducting surface is yucky magnetic steel, which is not so good a conductor. I'm amazed that it works at all. One of my friends was making 150Mhz cavity resonators out of stove pipe and found the skin effect losses to be substantial.

Gain is the sum of many factors and is difficult to generalize for a type of antenna. In this case, the coffee can waveguide feeds were designed to illuminate a dish antenna. The gain of the entire system is calculated (err... computed) in detail as a system, not just the feed. The gain provided by the dish part is usually much more than that of the feed horn. Therefore, small errors in the feed can be ignored as the overall gain is set almost entirely by the dish diameter. In other words, the gain of the coffee can is not an important consideration. It has a big effect on overall dish efficiency (overspray, under-illumination, vswr), but that is part of the dish gain, not the feed.

Reply to
Jeff Liebermann

I've heard you say "spin" before. I thought you meant turn, like I do with my camera-tripod mounted coffee can. I lost my oscilloscope somewhere. Last I recall, I lent it to someone to hook up to speakers at a party.

There's that spin word again... presumably at a different speed. Looking for the null was not something that I tried. A USB dongle in my shirt pocket might be effective. I tried walking around with a laptop and a WG511. One open Linksys WAP was available everywhere, with signal provided via bounce, and stronger at almost every hallway intersection. A slow spin looking for high signal wasn't helpful. The Orinoco with a chili can was directional enough, again looking for high signal, not low. I needed a cohort though. He looked where the can was pointing, while I concentrated on the screen.

I don't see the feed, or the spacing for it. I presume it's coming in through the coffee cup. The waveguide dimensions are only 2 dimensional. Where are the measurements for the horn? The dimensions are discovered from a program on a web site with no link. It would have been so easy to put in

formatting link
but that reference material looks more useful than this page.

formatting link
has more information. It also references the the Micro-Aerial, which is a handier size, and used as the feed for the big horn on the page. It even includes the picture that you referenced, and the link to the qsl site.

This qsl.net book offers 31 pages of coffee can thoughts, including adding a collar. In the grand scheme of cobbling, it looks like a four inch coffee can with a wood stove 4"-6" adapter added near the open end. That results in a big improvement in front to back ratio. It is more compact than a horn, and might be sufficient for my current project.

I don't see any "gain" expressed in the qsl charts. On your coffee can chart, it looks like the gain is discovered elsewhere, and just noted on the chart, with the chart graphic being dB down from that known amount.

Reply to
dold

Ignore the noise numbers. Going from -35dBm signal to -50dBm is a

15dB *LOSS*. Your can antenna has a problem.

If you set the bridges to a fixed connection rate of about 6mbits/sec, you should get a TCP thruput of between 2 to 3mbits/sec. With the two radios next to each other, you'll probably get close to 3Mbit/sec. Since you're getting 1.5mbits/sec, my guess(tm) is that you're getting about 30% packet loss or resends due to interference and a lousy path with probable reflections. I still predict that if you can get this link to work, it won't stay up and be reliable.

6Mbits/sec is the slowest OFDM rate. Since I guess(tm) that you have some reflections, going slower will only make it worse because CCK is far worse at dealing with reflections than OFDM. Since I also guess that you currently are running with a 30% packet loss, any increase in connection speed will just make that worse. It makes no sense to deliver corrupted packets at higher speeds.

That will help reduce local interference.

You should use WEP as someone could hijack the connection. The only other security is the pre-set MAC address of each end of the link, which is easily spoofed. Also, not enabling WEP will allow someone to sniff the traffic.

Basic is compatibility with the older 802.11 only (1 and 2 Mbit/sec) radios. It also includes an ultra long preamble and other such airtime burners. Since you're NOT concerned with these or even

802.11b radios, it doesn't matter. Once you set the data rate to a fixed speed, all these setting become meaningless. Set it for 802.11g only so that it doesn't send management junk at 1Mbit/sec (another airtime burner).

Put a repeater on the roof of his house. Line-o-sight in both directions. If you make it solar powered, then you won't need any wires going into the house. I have a few of these in the tree tops.

That's my job. You just present the evidence and numbers.

Yep.

802.11g can go for miles with proper antennas (and line of sight). How far do you have to shoot before you get to a house with faster ADSL?

VNC requires good response time to be used effectively. Check your end to end ping times. Packet loss will cause the ping time to dramatically increase. I use fping, which unlike Windoze ping, shows the sequence numbers so you can see the packet loss.

formatting link

Get a real or better antenna. Netstumbler already proved that the can antenna seems to be giving a -15dB loss. Trust your measurments and verify the with/without cantenna test using a more controlled location or known good antenna. I'm fairly sure you'll need some decent antenna gain at each end to make this work.

Reply to
Jeff Liebermann

Not sure if I now understand netstumbler or if it gives truthfull figures. Siting the laptop about 2m from each of the linksys with both of their rubber duck aerials connected gave me approx Signal -35 Noise -94.

I then connect a linksys to the cantenna, go outside and the reading from about 6m below and 6m back (ie about 45deg and 8.5m away) is Signal -50 Noise - 90. I was expecting an improvement on the Signal.

Nonetheless:

This seems to have worked. When all rigged up in the office and setting the linksys (now bridges) to your recommendation of 6mbps I get a throughput of around 1.5mpbs.

When setup as finally required the throughput is around 1mbps and fairly steady. My firewall displays traffic flow rates and was showing anywhere from 90kbytes/sec to 140kbytes/sec with very few if any dropouts. This is a big result and the first time I have had any consistency of throughput with the cantennas talking to eachother.

I also seem to get a better signal from the D-link to the laptops at home.

Once I'm satisfied that things are settled (and after I've got some work I'm supposed to be doing out of the way) I want to see if there is any more to squeeze out of the system. This includes:

i) upping the speed of the linksys bridges from 6mbps ii) setting the antenna from diversity to the one the cantenna is connected to iii) do I need WEP on the the linksys's or is it ignored anyway? I don't want to be double WEPing (ie D-Link is encrypting then the bridge is encrypting again). iv) The D-link has a rate setting called Basic (set to 2mpbs) and another setting called Tx (set to 11mpbs). Might change the Basic if I can find out what it means.

My friendly farmer has cut his tree back for me (a few days ago so it wasn't the reason for the improvement). He hasn't offered to move his house though :(

Don't know what conclusions can be drawn. My uneducated guess is that the home linksys was struggling to know which signals to adjust to (notebooks in the house on one antenna and the cantenna on the other).

Our location is such that ADSL has only just arrived and will be limited to 512k because of the distance from the exchange so the throughput is not going to be limited by the wireless.

I use VNC to drive one of the work computers from home so there is not much load there and then the occasional file transfer so I'm now reasonably satisfied.

Thanks for your input and any further comments/observations will be read with interest.

Reply to
Tony Lewis

Two possibilities: When checking antennas using NetStumbler, I found I had to get some distance in order to get good numbers. Antennas show more obvious improvement at fringe ranges.

I would go to the farthest range where the naked card can see the WAP, and then connect the cantenna to see if there is an improvement.

Your cantenna is no good. Bad cable, bad solder, bad location of the probe, something.

Reply to
dold

He was going outside, so the loss might be loss. We didn't get to see the outside naked number, but I agree. I think his cantenna isn't as good as a piece of coax, stipped back and folded over.

Reply to
dold

Wish I measured both cantennas before sticking them up high. Can't get any further back from the office before I run into the office behind (with the flat roof over which I have a couple of metres clearance). I can try the home one though.

I'll have to borrow a forklift or cherry picker and go into the field in the middle.

Reply to
Tony Lewis

Now I'm beginning to wonder if there is a problem with one or both of the Linksys.

My setting up of the bridge was all in one room, linksys's in bridge mode next to eachother using all rubber ducks, the d-link about 3 metres away and me and my laptop offset in the middle.

I simply copied a 3Mb file from a Win98 PC on the 10mbit network to my XP laptop taking around 20secs (for reference it takes 5.5 secs when the XP is wired into the network).

The implication is that one(?) of the linksys's at least is reducing the throughput then?

Reply to
Tony Lewis

That should be metres of course. Silly error.

What type of coax? I like numbers, not

Barry

Reply to
Barry OGrady

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.