Will I extend range on my wireless home/office project

Sounds like fun.

External antennas will be necessary on both ends.

The WRT54G is a good choice.

Neither of those are useful, due to the lack of an external antenna connection, in the scenario you are looking at. (Since you have it, at the end I'll suggest a way to use the WRT54GX.)

I'm not familiar with a WR3000. I would *not* rely on any "guaranteed 1 mile range", but would want to know what the power output is and what the gain of the antenna is.

I would get another WRT54G... and one major reason for that is because you are talking about using the WRT54G as a WDS repeater, and it is very difficult to predict how well it will work with another brand of equipment.

That will probably work. It requires the trailer WRT54G be set up as a WDS repeater. It also means, with the two antennas both actually being used, that the unit will be switching back and forth between them. I'm not sure just what the total effect of that will be, but given that your Internet bandwidth is significantly less than the potential bandwidth of the wifi link, it probably does not matter at all.

The WRT54G, out of the box, has a Linux kernel. I assume you mean you are thinking of installing third party firmware. That isn't necessary to make this all work, but is certainly nicer and more fun to do with the third party firmware than with the default firmware.

That should provide some nice versatility.

One question I would ask is how much hardware setup time is acceptable for each time you put this into operation? If you want the antennas permanently mounted on either vehicle, that is very different than if you can put them up and take them down each time. A directional antenna requires at least some adjustment each time, so it could not be just mounted and never touched again. And many high gain antennas simply wouldn't be suitable for leaving up while traveling. But of course higher gain antennas will extend the range, and provide some isolation from interference, so that might be useful on one or both ends of the link.

Non permanent?

....

Not much gain. The antennas that come with a WRT54G, and virtually all that are similar to those, are perhaps 2-3 dBi. A

7 dBi antenna is 4-5 dB more, which will add some distance, but will probably less than double the range. Figure on 6 dB being enough to double the distance over which the link will function at any given data rate. (However, with no obstructions, those 2-3 dBi antennas should be good for at least 3-400 yards.)

Jeff Liebermann will be able to cite a few specific antenna designs off the top of his head, so I won't because I'd have to do research and actually learn something...

If you don't mind setting up the external antennas each time you do this (and I assume the Internet access takes significant setup effort, so this would add only a small effort relatively), you can use just about any directional antenna design, and could easily choose something with significantly more gain than a 7 dBi antenna.

Another possible way to go is get another WRT54G and some low loss coax and mounting hardware to locate its antenna permanently on the top of the trailer. Then use a high gain antenna on your existing WRT54G in the car. The only per use setup required would be at the car.

Inside the trailer, install *both* the new WRT54G (which will be configured as a client, connecting to the WRT54G configured as an AP at the car) and your existing WRT54GX. Connect those two with ethernet between the LAN ports. Your in trailer wifi clients will connect to the WRT54GX, and use the WRT54G units as a point to point link to the satellite system.

Send the WRE54G to me.

If you include the connector and tiny cable loss, the antenna gain is more like 1.3 to 2.0dB gain depending on sloppiness of contruction.

Agreed. The original requirement was for 30ft. I'm wondering why he couldn't just grab a roll of stranded CAT5 and roll it out between the car and the trailer. Methinks 30ft wireless links are a rather wasted effort.

Gee thanks. Methinks an omni makes sense for this application. There's no way to predict exactly in what direction the car and trailer will be oriented. It's one less thing to adjust. The range is short so gain isn't a big issue. Bandwidth is limited by the satellite system, so there's no need for super S/N ratio. Yeah, an omni will work. If you want to spend money, just about any decent omni will work:

(Antennas-Omni) I don't have any idea about the mounting and cable restrictions, so I can't suggest a specific model. If a mag mount seems ideal, then one of the 5.5dBi magentic mount antennas seem tolerable (and cheap).

It is a fun project.

Floyd and Jeff, first thank you very much taking your time and giving so much thought and feedback to my project.

To answer some of your questions: The internet hardware setup very minimal (Model: DataStorm D2), power the system, wait for GPS signal and push the search button on the satellite controller. It will align to satellite and connect to internet automatically in 5-15 minutes.

The satellite connection is pretty slow compare to WIFI. Currently the upload speed: 15-70kbs, Download speed:200-700kbs. I'm getting a different satellite system later (INETVUE .98 meter) which will increase the upload up to 256kbs and download up to 1.5mbs.

I use Cat5 cable now between the car and trailer and I want to keep that option as well. However I had scenario where my car had to be parked the other side of roadway where I could not use cable connection between trailer and car.

I'm looking for permanent installed external antennas and rather not adjust those if all possible. I thought of magnetic because of easy installation.

I'm only looking to maximize my internet speeds through the WIFI. I won't need 1 mile range either but few hundred yards would be nice. There will be no cable connection between the car and the trailer unless I stay longer time in a place. The car will be totally self powered has solar and plenty of batteries to run the system without touching for days (depending how much Sun it gets).

I will have to give another thought of the WR3000, it make sense how different brand would be difficult or impossible to synchronize. I was just not impressed the WRT54G transmission power. It seems to me drops signal much sooner then an earlier B model I had years ago. I could be way wrong since I didn't do any measurements.

Floyed: I thought I have to install a 3rd party firmware to use the WRT54G as a WDS repeater. Under what menu I set it up as WDS on the Linksys OS?

I've been living on the trailer for 5 months, what I learned that I travel and move a lot so I'm looking minimal setup. Stop, push a button, brew a cup of coffee and start working over the internet :)

Thank you Endre

I think you mean INetVu:

you sure StarBand, DirectWay, and others offer 1500/256 service? I don't think so (but might be wrong).

Well, running CAT5 across the road isn't a great idea.

The problem with the maggot mount antennas is not the antenna or the magnet. It's the tiny coax cable that goes between the antenna and the radio. It's invariably too long and VERY lossy. You may have

5.5dBi of gain in the antenna, but if the coax has a loss of 5.5dB, then you have a net gain of zilch. Might as well put the access point in a window and use the stock antennas (assuming you have line of sight). Looking at the maggot mount omnis on: find that they come with 5ft of RG-174 coax for a loss of about 3dB. That's half your tx power and half your receiver sensitivity gone up in smog. Add the loss of the pigtail (if any) and coax connectors (1dB per pair), and the maggot mount antenna offers only a slight net gain.

Nail down the numbers please. At short range, almost any antenna will work. At about 300ft, you'll be into a directional antenna at one end. At "a few hundred yards", you'll certainly be into direction antennas at both ends unless you want to tolerate a VERY narrow vertical beamwidth on a high gain omni antenna. That will only work if the camper and vehicle are both level and at the same elevation.

I just measured my WRT54G. About +13dBm in 802.11g. +15dBm in

802.11b. However, my accuracy sucks because my test equipment is either ancient or home made. Increasing the TX power to improve the range must be done at both ends. If you want high power (+20dBm and +23dBm) and a rather good quality receiver, look into one of the Senao based wireless bridges: that these are 802.11b, not 802.11g. That will limit you to about 3Mbits/sec thruput. However, that should be sufficient for your satellite extension. I would still go with the WRT54G.

It's not in the stock Linksys firmware. However Sveasoft Satori has it as well as other firmware mutations. The availability of such features is one one reason why I like the WRT54G.

OK, that means omnidirectional antennas on both ends. Let's run the numbers and see how far you can go using maggot mount antennas and a pair of WRT54G radios: Distance = Unknown TX power = +15dBm TX coax loss = 4dB TX ant gain = 5.5dbi Fade Margin = 20dB RX ant gain = 5.5dBi RX coax loss = 4dB RX sens = -88dBm (at 6Mbits/sec OFDM)

Plugging into:

trying various distance values until we get 20dB fade margin, I find that this arrangement will work to about 0.12 miles or 753 ft. Good enough methinks. I think the 4dB coax losses are a bit optimistic, but good enough for now.

Do those things actually hold on that well? I've never tried one, but I just can't bring myself to trust something like that. I'd want it *bolted* down.

For Endre's benefit, let me say that it *absolutely* the case.

The fewer connectors and the fewer/shorter the lengths of lossy coax, the better. Spending good money on high gain antennas and saving money on cheap feedline is counter productive. The ideal physical design is a relatively short run of low loss coax and does not require a pigtail (a short length of more flexible, but high loss, coax with the right connections on each end).

Antennas with 'N' connectors are the right ones. The ideal run would use less that 4 feet of 1/2" semi-rigid low loss coax with an 'N' to whatever converter at the end. Rather than a short bit of flexible coax, a non-rigid mount for the radio is good too.

Even a very plain 3 dBi antenna with that setup will out perform a much more expensive arrangement using coax and various connectors chosen to make the installation simple and/or easy to look at.

The stock antenna on a unit mounted in a window has just about

*ZERO* loss from connectors and cables!

I just *like* the added flex available with at least one high gain antenna. It's difficult to know what sort of requirements will come up with a mobile unit, so I'd go with at least a high gain antenna for the car, and would probably settle for an omni permanently mounted on the trailer.

Is that with it set for the default 28 milliwatts output? Doesn't look too bad. 28 mw would be 14.47 dBm, so it certainly tracks.

Of course the WRT54G(S) can be cranked up at least 4-6 dB, so it is then right at that "high power" range too. They can do 254 mw, which would be +24 dBm, but I'm not sure how long it would last at that power. I know that short term they do in fact work at that output, and at least don't burn out in one day.

Everything I seen in writing suggests the Linksys has pretty good receivers, though I'm not able to verify that.

These things are just too much fun!

Sveasoft just released the free 1.0 version of Alchemy, the successor to their Satori firmware. It very definitely improves on the Web interface, though there doesn't seem to be any significance for those who is using a command line interface.

Specifically they now have a web interface to manipulate the vlan configuration for the ethernet bridge. So what before was just "AP Isolation" is now fairly flexible. They also have a better status display, which will show the received signal strength for the connected clients. It also does a "site survey" which displays information about all received signals.

All of that was of course available via the command line with the Sartori firmware, but those who only use the WEB interface will find the improved functionality very nice.

I sorta went over the edge. I downloaded the Linksys source code distribution as well as the Sveasoft source code. That is right at 250 Mb over a dialup modem... it literally took all day long.

But, I now have a cross compiler set up and can easily generate a binary for any program that I want to run on the WRT54G. (I dislike vi, so the first real program is a stripped down version of MicroEmacs!)

But I also added the ability to nfs mount filesystems. Now, rather than being limited by the 16Mb total RAM split between the filesystem and the cpu, and rather than needing to recompile and reload the entire firmware to get access to new programs... I just mount a few gigs of disk from the box with the cross compiler.

I haven't done anything particularly useful with it, but it certainly has great entertainment value.

That looks accurate to me. I'd note that if enough thought is given to engineering the antenna systems, probably at least 8 dB of extra gain can be picked up. And at least 4-5 dB more power can be used too, with the WRT54G units. Instead of 250 yards, maybe 1000 yards is possible with reliable solid connections.

(Of course no matter how well this all works, the instant an 18 wheeler pulls up and parks right between the car and the trailer, it all goes to Hell in a hand basket anyway.)

They vary substantially. I have about a dozen assorted magnet mounts for my assortment of ham and commerical antennas. (Real hams drill the holes). The mag mounts will not stick to a curved surface making positioning important. Some have such a grip that they are impossible to release without gouging the paint job. Others are so weak that the wind will blow them over. In general, the bigger the magnet base, the better it holds.

Drivel: One of the demos I give erratically for the ham clubs and commerical radio cronies is to take every coax cable adapter in my collection, string them in series, and measure the loss. I usually do th

Fine. Then find me a spring mounted 2.4GHz omni. Something with a

3/8"-24 thread. I could probably build one with a little bracketry, but I can't buy one. Maybe Ed can add it to his product line. The problem is that all of the high gain omni antennas are long and designed for mast, pipe, or tower mounting. However, campers and vehicles need to deal with tree branches which means flexible or spring mounted antennas, not rigid fiberglass tubing. Maybe a tilt over marine antenna mount.

Yes. It's my WRT54Gv1.1 running Sveasoft Satori, with everything on the Wireless -> Advanced page set to defaults. Power level at 28. However, the accuracy of my test equipment and goofy calibration methods are somewhat lacking. I'm guessing I'm withing +/- 2dBm (approx +/- 30%) which isn't great. Measuring the tx power also involves a bit of number juggling. I measure the top of the spread spectra on the spectrum analyzer and add 10dB (bandwidth / IF BW) to get peak power.

I'm guessing(tm) that 250mw output is a bit much for the Broadcom chip. It proabably won't cook it at room temperature, but might do so if mounted outdoors in the sun. I've been tempted to see what happens on the spectrum analyzer when I crank up the power level. Probably distorts and splatters badly.

Yep. However, the main advantage of the WRT54G Broadcom chips is that the digital noise level on the board is minimal and doesn't screw up reception by adding a mess of noise.

Thanks for ruining my weekend. I gotta try it.

Yech. I never could figure out emacs. Vi (or vim) rules.

Never mind the moving obstructions. Most trailer parks and campgrounds I've been in are full of trees. Usually, they're located around the perifery and between parking spaces for privacy. Finding line of sight might be a challenge.

I'm not sure what you were going to say there... But my guess is that the loss from all of those nice connectors strung together isn't that bad. It probably isn't.

But take the same bunch of connectors and put them like that and hang the whole thing out the 2nd floor window, and leave it there for a year or two. Then measure it. It may not be so good by then...

Excess cable is a guaranteed loser. Excess connectors are just potential losers. If they are properly installed and sealed up agains moisture, and then never disturbed, they're fine.

That's the reason for an omni on the trailer (a short whip that can flex will do it), and using the high gain at the car where it all has to be setup up each time.

Pick up an older HP power meter. You *can't* measure average power with a spectrum analyzer.

Everyone has strange quirks. Guess you are entitled to one or two... but along with the three or four we tripped on last week and the month before, frankly I'm beginning to wonder about you.

I couldn't resist. Loaded, reset to defaults, setup, and it works. Alchemy-V1.0 v3.37.6.8sv

SNMP is nice but returns some really dumb comments. Time to find a MIB file. Well, maybe not. It's 1AM and I'm ready to drop.

Thanks much.

I predict that you will.

Barry

I usually string about 6ft of adapters together. That's about 100 adapters. HT, two wattmeters and a dummy load. However, I cheat a bit. I don't use too many PL-259/SO-239 pairs. I also have 6ft length of RG-58/u and LMR-400 with N connectors handy for comparison. At 440MHz, the VSWR is usually perfect and the loss is typically about the same as the RG-58/u at about 0.6dB. Working the numbers, that's about 0.0006dB per connector.

Even more entertaining is stringing all the BNC "T" connectors together and running the same test. By all reason and logic, the large number of approximately 0.5" "stubs" should result in miserable VSWR and high losses. I only have about 50 of those and get about 1dB of loss. VSWR is something like 1.15:1. Not great, but useable.

Extrapolating these results from 440MHz to 2.4Ghz is not trivial. Were I to do the same test at 2.4GHz, the results would only be about

3-4 times worse or perhaps 2-3dB for the 6ft of adapters. That's still only 0.03dB per adapter. So, why does everyone seem to use 0.5 to 1.0dB per connector pair? Because connectors deteriorate. These are good, clean, dry, and new adapters. Add a bit of corrosion and the loss per adapter increases drastically.

Incidentally, another fun test and demo is filling coax cable with water. I start with a 3ft piece of LMR-400 or 9913 with a "slip on" uncrimped connector at one end. At 440MHz, the measured loss is usually unmeasureable. With a glass syringe, I dump in about 50ml of water. VSWR changes a bit, but the loss is usually about the same. Adding more water tends to increase the VSWR slightly, but no change in the loss.

The clue should be why I used a glass syringe instead of plastic water bottle. I cheat and use de-ionized water. I was also very careful to rinse and dry the inside of the coax from any contaminants. Pure water is an insulator. The increase in VSWR is from the change in dielectric constant causing a change in impedance. However, the pure water contributes no loss.

Now, I add a few grains of table salt to the coax. It doesn't take much. VSWR goes up some more, but the loss becomes huge. -3dB or more for a 3ft piece is about typical.

Moral: It's not the water. It's the contaminants that screw up cable loss.

Right, the voice of experience. I have a slightly different way of waterproofing outdoor connectors. The official method is the black rubber goo that never comes off the connectors, makes a huge mess, and cracks if exposed to UV for a few years. I've always throught it was designed solely to sell replacement connectors.

Instead, I wrap the connectors with 1" side teflon pipe tape. The TFE cold flows to make a seal. It also binds tight enough to prevent capillary action, the cause of water incursion. Over the TFE, I wrap a layer or two of whatever electrical tape I can find. It's only purpose is to hold the TFE tape in place, so selection isn't too critical. I've removed connectors that were installed on towers 10 years ago (when I was still climbing towers) and they're as clean and shiny as the day they were installed.

Methinks you mean peak power. Peak power can't be measured directly, but it can be calculated from the spectra. The problem is that the spectra varies sufficiently to make the accuracy of the calculation a problem. Worse, the pulsed nature of the spread spectra means the power meter will need to be measuring peak power. The average power is a function of duty cycle and transmit time, which vary radically. I have an ancient HP432A power meter but am missing the sensor. Even so, the HP47A bolometer/thermocouple sensor reads average power and is therefore useless. I could probably build a diode peak detector type of watts guesser for 2.4Ghz and spend some time getting it calibrated. (Added to project list as item #240)

People tend to prefer the style of editor or word processor they learned first. I learned vi first. Therefore, I am a vi bigot and refuse to even acknowledge that there are any other editors around. Incidentally, when I first learned to program, I would print out my punched card deck into teletype paper, roll it down the hallway, and mark changes with a red felt tip pen. 35 years later, I print out the program on my laser printer, and mark changes with a red felt tip pen. This is the extent of my progress.

As for my few errors, they REALLY bug me. Every time that happens, I get seriously depressed and have visions of premature senility, Alzheimer, brain rot, and such. Making exactly the same mistake twice was really a shocker. No excuse, and I guess my illusion of perfection is now shattered.

Incidentally, Alchemy fixed a nasty error in Satori syslogd, which would send some duplicate messages and fill my log files with useless warnings. No more filters. Nice.

Pure water is an insulator for DC. However, at 2.4GHz it isn't even close. That's why a microwave oven works... it conducts, and resonates to boot.

Contaminants are just even worse... :-)

We use(d) strips of moldable sealant, and then do the same that you are, tape it.

There are many dozens of antenna mounted LNA's in Alaska that need to be adjusted for better crosspol, and can't be. The LNA's can be removed if needed, because the connectors were treated as above. But the positioning adjustments haven't been moved in about 15 years, and short of drilling them out nobody has found a method of breaking the set screws loose.

When everything was SCPC that always had a gateway station with a large antenna at one end of the link it was fine, but with DAMA (Demand Assigned Multiple Access) equipment that can connect between two of the earth stations with small antennas it's necessary to boost the power to make up the difference in antenna gain, and now the crosspol is significant.

The analyzer displays the average power over the bandwidth of the sample, but not over the range of displayed bandwidth. Hence if you set the (sampled) bandwidth equal to the displayed spectrum, yes you get some indication of the average power in that amount of spectrum.

But that's not the way most people use a spectrum analyzer, and many of the lesser one are flexible enough to really do it. If you can distinguish spectral components of the signal, which is what a spectrum analyzer is typically used for, then you are not seeing an indication of average power.

Of course, you are probably well aware of all that... :-)

Because that isn't what many spectrum analyzers are intended to do that.

That is a real problem. Makes it useless... oh well...

Back to the spectrum analyzer...

I don't know what the inexpensive ones can do though, as most of my experience has always been with whatever the latest HP product was. What can be done with a $60,000 unit is fabulous even compared to what we used to do with those cheapie $25,000 units... ;-)

That is relatively true. Moreover, vi *is* a fully functional editor. (And there actually are only two, vi and emacs, though there are several variations of each.) Once a person has invested significant time with either one of them, there is simply no point in messing with anything else. Emacs was not my first editor, but it was the first one that I *liked* and stuck with long enough to *really* learn. That was more than 20 years ago, and after that much practice most of the "knowledge" is in my fingers. They do things and no sense of what is happening ever gets even as far as my elbow, never mind my brain.

It happens that I don't like "mode" style editors, such as vi. I like "modeless" editors like emacs. That's why I do emacs and not vi. But learning either one of them is essential, as is being able to at least use the other, and everything else is a waste of time.

You sir, are a hopeless *Ludite*!

I don't print anything out. My first exposure to programming was in 1964, when we'd punch out a deck of cards and put them in a bin for the high priests if the IBM to run that night. The next day we looked at the printout, which said "Line 3: Syntax error". grrrrr. Fix that and do it again. The next day's print out of course says "Line 7: Syntax error".

I decided that computers were for people with too much time on their hands, and spent the next 15 years twisting knobs and ignoring bits and bytes.

Then one day they installed a DMS-100 in the same room I worked in. We'd heard rumors that these new digital switches had a computer down there in the bottom of a rack in the front row to control them. Everyone else looked at this thing and saw something like that. Not me! I saw hand writing on the wall, painted with a *huge* brush.

That digital "switch" was *NOT* a switch at all. It had no little computer in the corner either! It was a massively distributed computer system, which just happened to be configured to emulate a telephone switch. It could also control every nuclear power plant in the country, play aliens, and balance the bank's records, as well as provide PBX services to the local staff.

It was time to go back to playing with computers... and I haven't really done much of anything else since then.

Poor. They say the 50's are the "infant" stage of old age. A time when we get a small dose of what it's all about for the rest of your life. I've got 6 months left to enjoy infancy.

I'm getting used to the idea of progressing stupidity too... Things like being given an increased dosage of Atenolol (a beta-adrenaceptor antagonist, aka the infamous "beta blocker") and enjoying having a very low pulse rate for 3-4 months... when it became apparent that it was also knocking about 20 points off my IQ! I told the doc we had to cut it back because it was making me stooopid... she casually says, "sometimes it does that".

A good one in comic page the other day (I don't know the name of that strip). Old guy tells his buddy "I feel good, I set a new record today!" He asks what record. The guys says,

"Yeah I set a new record every day! For the longest number of consecutive days that I've stayed alive."

My mother in law passed away recently. We think she was 103.

If we weren't such cheap SOB's we'd have sprung for the $20 version long ago, eh?

Whatever, I'm having all kinds of fun learning how to do systems admin on embedded Linux.

I beg to differ slightly. There's nothing unique about 2.4GHz and water. See:

proclaims in the "Nov 2, 1999" section:

===== It's a common misconception that the microwaves in a microwave oven excite a natural resonance in water. The frequency of a microwave oven is well below any natural resonance in an isolated water molecule, and in liquid water those resonances are so smeared out that they're barely noticeable anyway. It's kind of like playing a violin under water--the strings won't emit well-defined tones in water because the water impedes their vibrations. Similarly, water molecules don't emit (or absorb) well-defined tones in liquid water because their clinging neighbors impede their vibrations.

Instead of trying to interact through a natural resonance in water, a microwave oven just exposes the water molecules to the intense electromagnetic fields in strong, non-resonant microwaves. The frequency used in microwave ovens (2,450,000,000 cycles per second or

2.45 GHz) is a sensible but not unique choice. Waves of that frequency penetrate well into foods of reasonable size so that the heating is relatively uniform throughout the foods. Since leakage from these ovens makes the radio spectrum near 2.45 GHz unusable for communications, the frequency was chosen in part because it would not interfere with existing communication systems. =====

However, water does have a drastically different dielectric constant from air: Air: 1.0 Ice: 3.2 Water: 80 I made sure that I didn't dump in enough water to create a substantial impedance discontinuity.

(...) I'll comment on the rest later. I just fell asleep at the keyboard which my traditional clue to give up and get some snooze.

Have you used the rubbery tape used to insulate electrical connections on submersible well pumps? I replaced a 10 year old pump, and the tape came off the connectors cleanly, and was still keeping the heavily mineral laden water away from my 220VAC supply. Two layers of the rubbery stuff, wrapped in opposite directions to avoid wicking, maybe covered with black Manco pipe tape for better weather protection.

EMACS source code was larger than the free space available on a newly installed Unix SVR2 system. Finding a bigger disk and performing the compile left me with a binary that was larger than the minimal installation of Unix. EMACS has always seemed like the prototype for Microsoft bloatware.