Wireless on the ranch

Here goes: I'm putting together an internet system for our "ranch" down here on the coast in Mexico to share among six neighbors. We are spread among three ridges and two arroyos, but pretty close, all about 75 - 150 meters from the central distribution point.

We got the satellite system installed at a central point and have a Linksys wrt54g hooked up to it. The sat modem and router are in a box on the side of a building, where I hope to be able to mitigate the corrosion somewhat. Three neighbors are able to connect by ethernet, and three will be wireless, including me, at the farthest, about 150 meters line of sight across the arroyo.

Monday we are going to cut some trees, stick the router up on a pole for a test and see how the signal reaches for the various houses. This will help decide which antenna to get.

My plan is to run an omni antenna on a pole at the center to get the

200 degrees coverage I need horizontally. Then use matching repeaters in each of the three houses (with optional ext. antennas) to allow us to use our laptops around each house.

My questions have two parts: ANTENNAS and RADIOS. Mostly, I'm explaining my plan to see where the flaws are, if there are better solutions, if it should work.

RADIOS to start as my neighbor wants to know what to buy as soon as we do the tests.

We already have a linksys WRT54G. They are common and easy to replace when they go out in 6-24 months as everything here does.

It looks like if we go with DD-WRT on these, then I can boost the power on each end a bit (avoiding alligators-right?). If using aftermarket linux firmware, then I understand that I can use a WRT-54G at each house as well, as these are supposed to work as repeaters, clients whatever on DD WRT. I like this idea of having all the same boxes easy to swap out for problem solving. (They will be the WRT54GL actually)

So, does this seem like it will work? I know that many dislike repeaters, I understand that using a single radio as one is pushing it. But we each want the signal in various parts of the house, not just in one corner or cable. And it seems like an elegant solution. Only one laptop will be on each repeater, and the repeaters will be apart from each other, in a star around the central antenna.

After the tests, Monday, I would also like to discuss my antenna choice.

I hope this is clear enough to follow...

Reply to
seaweedsteve
Loading thread data ...

"seaweedsteve" hath wroth:

Try it WITHOUT the repeaters first. Repeaters tend to be compatibility and RF pollution problems. At 75 to 150 meters range, you might be able to communicate inside the houses if they are fairly RF transparent (made of wood). However, if adobe or concrete, you may be stuck with the repeater.

Also, an improved antenna system at the central access point will improve things for everyone. I have some suggestions.

I have had zero hardware failures among about 15 WRT54G/GS boxes that I have sold or maintain. Most are running DD-WRT firmware. We have a local free wireless network also running WRT54G radios, running OpenWRT. No failures except for one dingbat that plugged in an 12v AC adapter and blew the protection diode. I fixed it in a few minutes. However, I have seen blown WRT54G among individuals that have cranked up the xmit power to maximum (250mw). When the weather gets hot, so does the radio, and the xmitter blows up. Lifetime seems to vary but

2-4 weeks is my guess. 100mw seems safe or 150mw if you can keep it cool.

It's not an alligator if the power is roughly equal on both ends of the link. However, my guess is that you don't have any neighbors that might cause or be affected by the increased interference. At worst, you'll just cause your own interference which is not necessarily a bad thing as it sorta reduces the "hidden node" problem where collisions occur because the clients can't hear each other.

Correct. In client mode, they can handle more than one MAC address so you can just plug in more than one computah per client radio. They can also play repeater, but they cannot do repeater and client mode at the same time. However, there's a way around it. Look into the WDS (wireless distribution service) mode. It allows using the radio as an access point for connecting wireless laptops and a wireless bridge for providing the backhaul back to the central router. In short, it's what you're trying to accomplish, so try it.

DD-WRT online emulator (so you can see what's available): |

formatting link

Sounds like your anticipating disaster. Yeah, I guess that's a good idea. Note that the WRT54G is not the only product that supports alternative firmware: |

formatting link
|
formatting link

  1. Client radios, yes but laptop roaming will be an issue.
  2. Repeaters, maybe. I hate repeaters.
  3. WDS. Yep, especially if everything is using the same firmware.

So there's no misunderstanding. I consider repeaters and mesh networks to be an abomination and am considering a jihad to wipe them from the face of the internet. The very mention of repeaters makes my wallet itch from the anticipation of selling the suffering customer a suitable replacement.

Sorta. The problem is that everything is on one RF channel. With a store and forward repeater system, only one xmitter can be on the air at a time. This cuts the maximum speed through the repeater in half. That's not a problem is the shared internet connection is relatively slow. Going from perhaps 25Mbits/sec to 10Mbits/sec through put through a repeater will never be noticed if your satellite link can only do 400Kbit/sec.

Look into WDS. Unfortunately, under the WDS mode lies a store and forward repeater to the wireless laptops (and a client bridge to the directly connected computahs). It's really a repeater in disguise. As long as you stay with one manufacturers interpretation of the WDS mode, it will work. You take your chances if you mix hardware and firmware versions.

I'm glad you mentioned that, because it's possible that this can be done without any repeaters, WDS bridges, or divine intervention. Cranking up the gain of the central antenna might give you enough range and power to cover the inside buidings at 150 meters. It really depends on the building construction and if there's a clear line of sight in between.

What I recommend is a Franklin sector antenna. They're very easy to build. I just built 3 of them and an testing them this weekend. So far, it looks really good. See:

formatting link
formatting link
formatting link
formatting link
'm doing the inverted AMOS antenna style, which is only about a 90 degree sector (instead of 120 degrees), but has a few dB more gain.

Not too bad. You left out the topography and house construction, but otherwise quite clear.

Reply to
Jeff Liebermann

If you're using a pair of routers in the outer buildings I'd consider placing the connecting router in the surrounding buildings at their farthest side. This way any in-building coverage they offer will be farther away from the central point. If you're using external antennas this shouldn't be hard to setup. Since you're in charge of the whole setup you should be able to setup subnets and routes to get everything coordinated effectively. You've got the choice of having everyone on one network or each on their own NAT routed subnet.

The biggest hassle you're going to run into as needs or expectations grow is outside traffic being routed internally. You can't expect to use only one external IP address and get services that expect inbound ports to work. At least not for more than just one computer per service. Just be aware of it. If all everyone's doing is basic web surfing and e-mail you're set. But if they wanted to get into VOIP and the like it starts getting a lot more complicated. Not impossible but not as convenient as them being able to add their own stuff to their own ISP connection.

-Bill Kearney

Reply to
Bill Kearney

OK. Jeff! Thanks for such a rich answer.

I'd better fill in the info gaps:

The significant part of the terrain is two ridges coming to the sea with an arroyo in between. One ridge (where I am) is straight and narrow. The ridge I am looking across at , (where the dish and gear are) is more of an L - it comes down, then follows the coast for a hundred meters or so. There is a house along that L, about 70 meters from the antenna that gets wifi. Behind the "L" is a tiny valley. I may need to hit a house there in the future.

Thus the 200 degrees of horizontal coverage and 10-30 degrees of vertical coverage (below horizontal).

Here's a photo of the view from my house: tagged:

formatting link
formatting link
labels on the photo blurred. On the left below my balcony I indicate where the house with the desktop is. It has LOS.

House construction/position: Yep,several of us will need antennas up high. My house is open above, but adobe below, where I want the signal. Also, below has no LOS. The neighbor on the opposite side of the central antenna from me has a concrete wall between her and the antenna. My neighbor below me has a desktop, so actually, I lied, no repeater there; he could either have his PCI adapter with a cantenna, or else a USB cookware solution. Or maybe the signal will make it into his "house".

The fourth is down around and below the central antenna in my side of the arroyo. He is so low that his LOS is buried in tree trunks on other properties that cannot be cut. I want him to run a cable (about

100 meters) but he does not want the trouble, is hoping I'll find a way to hook him with some super tree-busting antenna.

Good to hear that you have good experience with the Linksys. I only meant to say that everything here fails in a year or two, stainless steel rusts here.

It seems that the best cheapo swiss-army-knife router is a DDWRT router. Linksys or Buffalo, right? I tend towards the Linksys over the Buffalo because 1) we have one already 2)linksys updates thru web interface. TFTP seems more complex and likely to mess up 3)it's easiest to buy in a pinch anywhere 4) most supported online.

But maybe I should reconsider. Will the Buffalo/Linksys work together on ddwrt /WDS as brethren? Or, does the different manufacturer make a difference?... Is the Buffalo any better or just cheaper? And oh, the flashing....What about the flashing?

I understand that by having the same power on matching ends, I will avoid the alligator syndrome. As far the the neighbor with the desktop, his usb or PCI will not likely match the power level, should that be be a problem?

I was figuring to go with 100 mw, maybe 150 as you said. It's good to hear that one could run the linksys up to say, 200mw. I understand that's a bad idea- is it only due to heat? (Enough reason!)? As far as radio noise/interference, we should not affect anybody here on that count.

Our roomy, enclosed wooden gear box is in a breezy shady spot. I am still deciding on whether to fan cool it or leave it sealed so the rust blocker can work best. The box contains a power regulator, the Hughesnet "modem" with it's power brick and the router.

OK. The repeater bugabear: Forgive me if I'm stubborn, but I'm still struggling to understand what part about it is going to be bad for me...apart from your jihad, that is!

1) first of all, proprietary factor. If we stick to Linksys and DD-WRT, there should be no problem, right? And we will do it in WDS

"As long as you stay with one manufacturers interpretation of the WDS mode, it will work. You take your chances if you mix hardware and firmware versions."

I also like the idea that according to reports, with WDS an Apple Airport express could fit in someday as well. (Apple certainly pushes the repeater solution!)

2)Speed:

"The problem is that everything is on one RF channel. With a store and forward repeater system, only one xmitter can be on the air at a time. This cuts the maximum speed through the repeater in half."

Then you go on to say " That's not a problem if the shared internet connection is relatively slow. Going from perhaps 25Mbits/sec to 10Mbits/sec through put through a repeater will never be noticed if your satellite link can only do 400Kbit/sec.

Well, at times our link does 1,500kbps. But that still seems to be under a quarter of the 11 or 12 that I''m hoping for as a connection...

3) Signal confusion: Is there such a thing? As I look at my neighber,the one below the antenna towards the sea, I see that if he has success with a repeater in his house, mine will see the same signal on the same channel from the central antenna and his antenna. Will that be a problem?

Is repeating bad in itself if the compatibility and speed are not problems?

I figure that as a fallback, if there is trouble, I will get a cheap extra WAP and cable it to the DDWRT device running as a client. Then I'll have two radios, one for the long haul and one for the local for $100 total.

ANTENNA: I checked out the sector antennas. They look veeery interesting. I would need two, I suppose. Hmm. There is a diagram at one of the antenna links you gave for a signal splitter. Is that a good workable idea?

Unless the house down below and behind the antenna goes with cable, I need to cover about 200 degrees around with a beam from horizontal down about 25 degrees. What is that- a 45 straight beamwidth or a 22 with a

12 downtilt, or..

In other words, I need a donut. Or a donut cut in half. Or quartered. As in, slice it like a bagel, then cut that in two pieces. I need an antenna with that.

If that house drops out of wifi and goes with cable, then a more reasonable beamwidth, say 15 degrees down from horizontal should work. Here's two contenders:

formatting link
If I can use two sector antennas then two of these would work well too:

formatting link
I wonder if somebody has worked out how to make these?

Steve

Reply to
seaweedsteve

Thanks Bill. If I understand you correctly about antenna placement, you mean, the router acting at the clients house as a repeater should be on the side of the house closer to the central wap antenna?

Um, I understand enough of this to think I want to put it off for now...rain check?

Yep, thanks for the heads up on that. We will want VOIP, though it's not currently working over wireless, just by ethernet. I've used yahoo voice and Gizmo on this connection and both worked out the port stuff on their own....on ethernet only.

Reply to
seaweedsteve

For what it's worth (tough to tell from photos), but when I did the place in idaho, I had the sat (part way up) and the wifi stuff at the top of one of the hills on my property (about 1000ft AGL) that I could see from all over.. Worked great not only with the laptops, but also with my handheld, specially in the spring when the stuff grew....

Reply to
Peter Pan

"seaweedsteve" hath wroth:

That means at least two sector antennas. Typical is 12dBi gain with

120 degrees horizontal beamwidth and about 15 degrees vertical beamwidth. You can get more gain, but the vertical beamwidth will shrink. Tilt the antenna at an angle from vertical?

Los Hamacas, Oaxaca. Nice. Wireless looks easy if you do some deforestation.

Wood roof? You can usually go right through that. Adobe is like a brick wall. Well, it *IS* a brick wall.

Great. Why make it easy? At 75 to 150 meters, you might wanna consider running copper cable or fiber. If anything, it's cheaper. If you observe some precautions (like run everything at 10baseT-HDX), then you can go up to about 300 meters with CAT5, RG-58a/u, or my favorite hack, RG-6/u CATV coax. Details on request.

Does this neighbor have line of sight? You might wanna number the neighbors in your description as I get easily confused.

What happened to the third neighbor? Trying to cover all these people from a single location may not be possible. Is there a nearby hilltop that everyone has line of sight? Put an access point and antenna(s) up there and run cable to your house.

Oh yeah. Corrosion. I used to work for a marine radio manufacturer. Some of the radios that were returned from the field had more corrosion than base metals. We had to spray some stuff with various acrylic, urethane, and wax conformal coatings. Always a mess and it would often ruin connectors, adjustments, plastics, etc. Corrosion inhibitors wouldn't work unless the radio was hermetically sealed. If hermetically sealed, it had to be pressurized or condensation would accumulate inside. Good luck.

I don't have a huge amount of experience with these. The WRT54G/GS boxes I deal with run DD-WRT v23 sp2. The local freenet uses OpenWRT. They work. TFTP is nothing to be afraid of. It works quite well and is necessary if you manage to "brick" the router. There are lots of good Windoze TFTP clients available. Linksys has a really simple Windoze version. | (somewhere in the Linksys ftp mess) Command line TFTP client: |

formatting link
's also a command line TFTP client in Windoze 2000 and XP: |
formatting link
I haven't used much Buffalo hardware (except for their NAS products). John Navas wrote me proclaiming that the Buffalo wireless is superior to Linksys in terms of range and features. So, I guess I'll try one next time I have an excuse.

Probably will work. Same chipset. Same firmware. What can go wrong, go wrong, go wrong, go wrong, etc...

RTFM. See: |

formatting link
simple enough with lots of precautions. Looks like you have to use TFTP to get the DD-WRT image into the router. After that, you can use the web interface. You'll probably get better information from the DD-WRT forums.

It won't be a problem because there's nobody else to interfere with. If your xmit coverage area is substantially wider than your receive range, there's nobody else in the area that will suffer from the interference. In the middle of nowhere, you can get away with quite a bit of power.

I've only done one post mortem and I was unable to repair the WRT54G. I'm sure that the power amplifier chip was dead, but I think the damage also propagated to the Broadcom RF chip. I didn't spend much time on it. Yeah, I think it was totally due to local heating.

Agreed.

Yep. It's a fair assumption that identical hardware and firmware will be compatible. The good news is that only the routers need to be compatible with each other. The client radios can be anything. However, I don't think you'll be thrilled with the results. When you get all the hardware together, try assembling everything and running it all in a single room, where every radio can hear every other radio including the clients. I think you'll have problems with "reliable" thruput and erratic download speeds. I recently did this demonstration for a customer that had bought a mess of repeaters and was wondering why all the guests in his motel were complaining. I'll admit that this is the worst case and that it will get better as the clients and repeaters are separated from each other.

Reminder: In WDS mode, you can't use WPA encryption. Only WEP works.

One of my friends has an Apple Aiport Express running WDS that works with his WRT54G running DD-WRT.

formatting link

I like to set the access points to "802.11g only". This insures that the connection speed never drops below 6Mbits/sec (or about 3Mbits/sec thruput). If you have a strong enough signal, this will keep the radios from going too slow when they hit with some noise or interference (i.e. microwave oven).

No.

No. Everything is orchestrated by the access point in infrastructure mode. The only potential problem is called "hidden node", where the clients can't see each other, and therefore don't know when the other nodes are transmitting. The result is some collisions. The solution is to enable CTS/RTS flow control to allow the central access point to decide when each client can transmit. It constitutes a substantial performance hit. However, you don't have enough clients to justify CTS/RTS flow control. Just let the clients collide occasionally.

Two answers.

  1. As far as your isolated system is concerned, there's nothing wrong with using repeaters.
  2. If you have other 802.11 systems around, then repeaters (and mesh networks) send duplicate packets for each hop. This hogs "air time" and leaves less time for other users of the same channel to communicate.

Incidentally, your assumption that compatibility and speed are the only problems may be wrong. The rule of thumb that repeaters cut the maximum speed in half is under ideal conditions. I've seen thruput as lousy as 10% on a 4 router and 5 client WDS network because of all the mutual interference. Try it all in one room and watch it happen. Remember that only one xmitter can be on the air on a given channel in a give airspace at a time.

That will work. I've recently spent some time tinkering with DD-WRT in the client mode. Works nicely.

Two at least. 200 degrees is impossible.

Yes. That works *IF* each antenna does not appear inside the other antennas pattern. If they see each other, the antenna patterns will "merge" and you will get a strange cloverleaf pattern. Actually, splitters are kinda weird. The reduce the power to each antenna in xmit by half (-3dB) plus the splitter loss (about -0.5dB). However, in receive, there is no power splitting or loss other than that of the splitter itself (-0.5dB). Therefore, there's almost no loss in receive sensitivity. This is one case where doubling the xmit power is justified.

Unless you build your own splitter, it's probably only a bit more expensive to install a separate WRT54G access point for each antenna. See the photos at:

formatting link
3 sector antennas on a pole with downtilt. Each has it's own access point.

Tilt the antennas downward as needed. If you want to cover just one house per antenna, that can probably be done better with a directional (yagi, dish, patch, panel, biquad) antenna.

Nope. That's an omnidirectional antenna with only -5degrees of downtilt. It's also short on gain at only 7dBi. It will probably work, but you can do better.

Worse. Only 6dBi gain with no downtilt. Forget the omnis unless you're desperate or want to mount the antennas upside down.

That will work. 12dBi gain. 65 degrees horizontal. 34 degrees vertical.

I have a guess on what's inside. Photos when I get one.

Reply to
Jeff Liebermann

Thousand thanks again, Jeff. You make me laugh!

Probably it's desperation. I feel further from a decision than ever. That's the nature of the situation, I suppose. There are several workable solutions with different trade-offs. I need to re-read your answers and digest them for a few days, then I will try to finalize my questions.

Tomorrow we are cutting a few trees at the central point and putting a WAP up in the air on a stick to see what a stock antenna and my Belkin laptop card will do. I am online at my house THROUGH the trees on that setup right now. It works about half the time.

Makes me wonder why I need high gain. I keep thinking, "It's not a long reach- keep it simple at the base station....one device (asume it has to be replaced frequently), one antenna that has wide coverage - crank up the xmit power a bit to cover the cable loss and each house is treated as needed.

Just for fun: (if your patience is running out ignore this - I don't want to waste it now!)

What DO you think is in that little boxy Hyperlink antenna? A Bi-quad ? What do you mean about mounting an omni upside down? With a ground plane ?

Reply to
seaweedsteve

"seaweedsteve" hath wroth:

You need gain because the area you're trying to cover (150 meters radius) is much larger than what would normally be achievable with the stock antenna. You could supply the necessary gain at either end of the links, but the gain has to be there to get a reliable connection. You can usually get some manner of connection through the trees. Maintaining it reliably, without dropouts, fades, and data loss is the problem. Try a large file copy from a local ethernet connected server at the access point.

The measure of how much gain is the "fade margin" or "system operating margin". See sample calculation at:

formatting link
that the fade margin can be used to predict the availability , reliability, and annual outage time. Fade Availability Downtime Margin % per year dB 8 90.0 876 hours 18 99.0 88 hours 28 99.9 8.8 hours 38 99.99 53 minutes 48 99.9999 32 seconds 58 99.99999 3.2 seconds

I can't tell what's inside the Hyperlink antenna. It's awfully small for having 12dBi of gain. Buy one, open it up, take some photos, post them.

If you take an end fed collinear antenna, such as is typical with most of the Wi-Fi omnis, and place a ground plane under it, the antenna pattern will raise somewhat above horizontal. It varies with the gain and configuration, but 3-7 degrees is typical. If you're trying to cover something that's below horizontal, then it's often better to mount the antenna upside down. It's really not an issue with omni antennas of about 10dBi or less as their beamwidth is sufficiently wide to deal with the problem. It's when the beamwidth becomes so narrow that uptilt will place the clients out of the pattern is where it's important.

Reply to
Jeff Liebermann

OK. We got the trees trimmed and then yesterday I put the router up on a pole to simulate the mounted antenna position.

Results using the stock Linksys antennas at the stock power level. Measured using Belkin G PCMIA card with utility. (Netstumbler won't show levels)

- #1 My house (across the arroyo - 150 meters to base antenna): Upstairs is 45- 50% signal level on my belkin PCMCIA card. (need

25-30% to function) This is great ! I could live with a 2dbi antenna.. It was connecting at 24 IIRC. Also, the signal reaches inside the lower level of my house (30% level) through the doorway even though blocked by trees a bit there.

- #2 Neighbor just below and beside me (130 meters to base: 50% signal level on Belkin at roof level. 30% inside at desk (blocked by bushes at the house, can be cut if needed)

- #3 House below and around hill from base antenna (about 80 meters) No signal. Due to the many uncuttable trees, he is ready to run a cable pending permission to cross properties.

-#4 House on opposite side of base antenna from me (about 80 meters)

25% signal level at entrance /outside concrete wall. Raising up the antenna did not help at all. Apart from the wall, this house has the same problem as #3: Too many uncuttable trees.

Conclusions:

#1 & #2 The two farthest houses are going to work great with any antenna over 8 dbi and will NOT need repeaters! As Jeff said, I can just blast them. They did ok with 2 db test antennas. Considering a

3 dbi loss cable assembly (25' LMR400+pigtail), a 5 db antenna should give equal results. 8 db will double signal strength? 10-12 will give serious improvement. Should not need any power increase on the radio.

House #3 will do best with a cable and a local WAP. He is now convinced. Plan B: If permission to run the cable is denied, we have good LOS from another position on the hilltop. We can run a second ethernet cabled WAP mounted in a tupperware at that location and hit his house with the stock antenna. At that point he can repeat if need be.

House #4. Really needs a cable and local WAP. There is no hope of clear LOS. Plan B is run a second antenna for #4 on the same mast using a splitter, something very directional else Circ Polarized ? Something that will bust through the trees even when they are green. Best to push hard for plan A.

So, due to some reorientation after checking into this group and also reality checks in the testing, I'm now tending towards a patch antenna for the two farthest houses. The closer two houses will use cables if at all possible or the "Plan B" solutions as noted.

Wireless Routers:

I checked out the Buffalo WHR routers more thoroughly and they really do get good reviews. The plain version is $45 shipped from NewEgg. I will get this one. The WHR-HP (high power) looks super for about $60 shipped through Amazon.

I will have the cabled houses go with the same thing for their local WAPS.

By the way:

|

formatting link
Actually, I had read that link - it's is apparently all there is out there. I was refering to my fears of the somewhat trickier aspect of TFTM (timing, command-line). I know that once one has a bit of experience, this kind of thing is EASY. But the first time around it's a cryptic and mysterious path compared to GUIs.

Anyway, I'm taking the chance and switching over to the Buffalo WHR G54S as our standard here.

As far as antennas, well, those Franklin sector antennas look like a good choice, but again, this hyperlink directional patch looks like a winner:

formatting link
So cheap ($32 shipped Ebay) good specs, mounting hardware, it hardly seems worth it to fabricate.

Thanks again for your help/guidance on this. Any comments?

Steve

Reply to
seaweedsteve

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.