I have an off-grid home. My closest neighbor and I use two-way satellite
(DirecWay). Now we have a third neighbor moving in who was talking about
getting the satellite connection. I said that was nuts :-)
Another neighbor, just about exactly 1km away over open water (foggy at
least 60 days a year, if that matters) has cable access. While she can't
legally share that, I have no problem paying the cable company for a
business connection which we _can_ legally share.
I want to set up an access point where the cable connection is, with a
directional antenna to us. The three houses are within 15 degrees from
that point. Then, at each "client" house, I want to put a directional
antenna pointing back to the base, connected to a repeater, so that the
computers in the client houses have wireless access via the repeater.
The real problem is that I'm confused by much of the terminology used and
I'm not sure what I'm really looking for. Using D-Link as an example, just
because it was easy to find, not because I know enough to want to buy
them :-), it would seem I should be able to use:
- DI-524/624 at the base station, connected to the Cable and external
antenna
- either a DWLG710 range extender or a DWL-G700AP as a repeater. Seems like
they'd both work, but the AP has more capabilities and costs less!
Am I on the right track? And is that fog going to be a problem?
--
derek
Derek,
I don't think you will get 15 degree coverage over 1km, As I learned
the hard way a 15db / 8 degree cantenna goes a lot further than a 20db
/ 30 degree panel antenna. I guess you have to divide the power over
the area covered. I would reccomend several cheap AP's using non
adjacent channels all plugged into a hub or router at the base station
with seperate cantenna's aligned for each client (put it in bridge mode
if client is a network vs an individual machine- you will also need
cantenna on client end to insure hi bandwidth connection. it might be
cheaper to use a pair of dish antenna to bridge the lake and then
redistribute from your side.
>
> Derek,
> I don't think you will get 15 degree coverage over 1km, As I learned
> the hard way a 15db / 8 degree cantenna goes a lot further than a 20db
> / 30 degree panel antenna.
Well, we'll see that when my test equipment arrives :-) I've got a couple
of 14dBi/20degree antennas to test with. My brother-in-law thinks it'll be
no problem, but that's one of the things I want to be sure of.
> it might be
> cheaper to use a pair of dish antenna to bridge the lake and then
> redistribute from your side.
It might even be necessary. My two neighbors are very close and I'm not
certain the one (who _is_ conveniently line of site to all the buildings)
doesn't block the other from the cross-lake line. If a single antenna at
the internet end could serve all three clients, the centremost house would
probably be in the Fresnel zone of the other house.
--
derek
> On Tue, 09 Aug 2005 16:35:13 -0300, Derek Broughton <news@pointerstop.ca>
> wrote:
>
>>I have an off-grid home.
>
> What does that mean?
Thank you for the highly intelligent response to my question. Look it up.
I have no connections to cables of any kind.
>
>>My closest neighbor and I use two-way satellite (DirecWay).
Look into the Linksys WRT54G. It has some big advantages for what
you're trying to do. It can run on anything between about 4VDC and
18VDC (negative ground). Using altenative firmware from Sveasoft, you
can configure it as a router, access point, or client device. It will
do WDS repeater bridging. http://www.sveasoft.com/content/view/3/1/
>My closest neighbor and I use two-way satellite
>(DirecWay). Now we have a third neighbor moving in who was talking about
>getting the satellite connection. I said that was nuts :-)
DirecWay isn't all that fast. There really isn't that much bandwidth
to share. Two of my customers have these with about 3 simultaneous
coputers. It's a battle when everyone is on at once. The headache is
the upstream bandwidth which averages about 40Kbits/sec and is easily
saturated.
>Another neighbor, just about exactly 1km away over open water (foggy at
>least 60 days a year, if that matters) has cable access. While she can't
>legally share that, I have no problem paying the cable company for a
>business connection which we _can_ legally share.
Cable is far better bandwidth suitable for sharing.
>I want to set up an access point where the cable connection is, with a
>directional antenna to us. The three houses are within 15 degrees from
>that point. Then, at each "client" house, I want to put a directional
>antenna pointing back to the base, connected to a repeater, so that the
>computers in the client houses have wireless access via the repeater.
15 degrees is not impossible but difficult. A 14dBi panel antenna
will have about 30 degree -3dB beamwidth which should work. You don't
want to get a 15 degree wide antenna as the stations at the edge will
probably be marginal. Instead of a repeater, I suggest either of the
following approaches.
1. Use WDS (wireless dispribution service) to build to store and
forward network system. I has all the advantages of a repeater, plus
you can plug computahs into the WDS radio to "tap" off a connection.
WRT54G is perfect for such a system. Essentially, each house forwards
to the next house. The problem is that every time you store and
forward repeat the data, the maximum bandwidth gets cut in half. See
guesswork on bandwidth below.
2. Install the WRT54 in an outdoor box on top of a pole and run CAT5
to the computers. No need for power over ethernet at the WRT54G will
run on just about any DC voltage from 4.0 to 18VDC. Just connect
12VDC, accept the copper losses, and it will work. This is generally
simpler, easier, and probably more reliable.
>The real problem is that I'm confused by much of the terminology used and
>I'm not sure what I'm really looking for. Using D-Link as an example, just
>because it was easy to find, not because I know enough to want to buy
>them :-), it would seem I should be able to use:
>
>- DI-524/624 at the base station, connected to the Cable and external
>antenna
>- either a DWLG710 range extender or a DWL-G700AP as a repeater. Seems like
>they'd both work, but the AP has more capabilities and costs less!
No comment on range extenders and repeaters. Well, maybe a small
comment. I think they suck, work only in about half the systems, are
incompatible with much of the available hardware, and chop the
available bandwidth in half.
>Am I on the right track? And is that fog going to be a problem?
Fog will not be a problem at 1km. However condensation on any outdoor
electronics will be. I've had to coat my boards with various Humiseal
products to keep them water tolerant. Perhaps a heater inside the box
to raise the dew point will be necessary.
My guess(tm) is that at 1km (0.62miles), you'll end up with a 19dBi
dish antenna at the client end and a 14dBi panel at the cable modem
end. With +15dBm tx power output, about -85dBm rx sensitivity, and
4dB cable losses at each end, you'll get a fade margin of about 25dB,
which is more than adequate for the link. My guess(tm) is that you'll
end up with a fixed connection rate of about 12Mbits/sec OFDM. That
will give you about half that in thruput. The first WDS bridge will
get the full 6Mbit/sec thruput. However, the 2nd wireless hop will
only get about half that, and so on. Methinks that's adequate. If
you want more bandwidth, get a 24dBi dish instead of the 19dBi, and
increase the connection rate. http://www.terabeam.com/support/calculations/som.php
On Wed, 10 Aug 2005 10:26:36 -0300, Derek Broughton
<news@pointerstop.ca> wrote:
>frank wrote:
>
>>
>> Derek,
>> I don't think you will get 15 degree coverage over 1km, As I learned
>> the hard way a 15db / 8 degree cantenna goes a lot further than a 20db
>> / 30 degree panel antenna.
>
>Well, we'll see that when my test equipment arrives :-) I've got a couple
>of 14dBi/20degree antennas to test with. My brother-in-law thinks it'll be
>no problem, but that's one of the things I want to be sure of.
Fire your brother in law. A 14dBi antenna with a 20 degree -3dB
beamwidth would be a yagi. In my never humble opinion, this is a
terrible choice. The signal strength falls over VERY rapidly at the
-3dB points. Even though you allegedly lose only -3dB at the end
points, the slope is so steep that even the slightest misalignment
will be fatal.
Here's some sample beamwidths extracted from random datasheets.
Type gain -3dB Beamwidth
dBI degress
patch 13 38
yagi 14 30
planar 13 25
dish 15 19
This chart for a yagi may also be of interest:
gain -3db beamwidth
dBi degrees
8 60
10 40
14 30
16 25
>> it might be
>> cheaper to use a pair of dish antenna to bridge the lake and then
>> redistribute from your side.
See my previous posting. Methinks that would be best. A solid dish
to dish link across the lake, and then redistribute using WDS
repeaters.
You may have one potential problem with the cross lake link that's
unique to microwave over water. There will be two paths. The direct
path between antennas and the bounce path off the lake. They will
arrive at slightly differnet times at the destination resulting in
self-interference also known as multipath. OFDM is very good at
eliminating these effects but is not perfect. There will be times
when the RF signal cancels and there's no signal to work with. The
ugly solution is to setup one end of the link so that the antenna can
be raised or lowered. I use cast aluminium hand rail connectors for
the purpose over a pole/pipe to raise and lower the antenna assembly.
After a bit of tinkering, the optimium, or least disgusting location
can easily be found.
>It might even be necessary. My two neighbors are very close and I'm not
>certain the one (who _is_ conveniently line of site to all the buildings)
>doesn't block the other from the cross-lake line. If a single antenna at
>the internet end could serve all three clients, the centremost house would
>probably be in the Fresnel zone of the other house.
Ok, time to get irate.... Would you kindly replace "very close" with
some real numbers? I'm a big fan of calculating before I build and
find it difficult to insert "very close" into my spreadsheet.
It might be possible to run CAT5 or coax between houses. I've broken
the cable length rules many times and know most of the tricks. You
can go about 1000ft of CAT5 or RG-6/u (yeah, I know it's 75 ohms) at
10BaseT-HDX speeds without any difficulty. You'll get about
6Mbits/sec thruput. I've posted details previously. Please advise if
interested.
> On Tue, 09 Aug 2005 16:35:13 -0300, Derek Broughton
> <news@pointerstop.ca> wrote:
>
>>I have an off-grid home.
>
> Nice. Do you read Home Power Magazine?
> http://www.homepower.com
On and off. I wish I'd read it _at all_ before I went off-grid :-)
> Look into the Linksys WRT54G. It has some big advantages for what
> you're trying to do. It can run on anything between about 4VDC and
> 18VDC (negative ground).
Cool, but not necessary. I'm off-grid, but I do have AC - and plenty to
handle a wireless device.
> Using altenative firmware from Sveasoft, you
> can configure it as a router, access point, or client device. It will
> do WDS repeater bridging.
> http://www.sveasoft.com/content/view/3/1/
However, that's nice.
> DirecWay isn't all that fast.
Tell me about it (oh, you did :-) ). I tell everyone who asks that it's
only a solution for people who have _no_ other solutions. At the time I
installed it, I didn't
> 15 degrees is not impossible but difficult. A 14dBi panel antenna
> will have about 30 degree -3dB beamwidth which should work. You don't
> want to get a 15 degree wide antenna as the stations at the edge will
> probably be marginal. Instead of a repeater, I suggest either of the
> following approaches.
>
> 1. Use WDS (wireless dispribution service) to build to store and
> forward network system.
The difference is what? That's my problem. I don't have a clue what the
wireless manufacturers are selling when they say "repeater". Can you point
me to some links? It's especially difficult when I go to sveasoft.com and
they tell me their firmware has "WDS/Repeater mode"!
> I has all the advantages of a repeater, plus
> you can plug computahs into the WDS radio to "tap" off a connection.
I thought that's what the repeater was supposed to do.
> WRT54G is perfect for such a system. Essentially, each house forwards
> to the next house. The problem is that every time you store and
> forward repeat the data, the maximum bandwidth gets cut in half. See
> guesswork on bandwidth below.
That's not a huge issue. I want to be able to promise better speed for less
cost than Direcway for each of three sites. I don't think that's hard :-)
(for those who don't know DirecWay, it's usually considered to be
approximately ISDN speed once everything's taken into account - it can be
much faster for large downloads, but latency kills www access).
>
> No comment on range extenders and repeaters. Well, maybe a small
> comment. I think they suck, work only in about half the systems, are
> incompatible with much of the available hardware, and chop the
> available bandwidth in half.
As would the WDS, right? Half the bandwidth is just fine :-)
Jeff Liebermann <jeffl@comix.santa-cruz.ca.us> wrote:
>It might be possible to run CAT5 or coax between houses. I've broken
>the cable length rules many times and know most of the tricks. You
>can go about 1000ft of CAT5 or RG-6/u (yeah, I know it's 75 ohms) at
>10BaseT-HDX speeds without any difficulty. You'll get about
>6Mbits/sec thruput. I've posted details previously. Please advise if
>interested.
Very interested, though I don't remember the discussions, please fill
us in!
On Thu, 11 Aug 2005 17:02:23 -0400, William P. N. Smith <> wrote:
>Jeff Liebermann <jeffl@comix.santa-cruz.ca.us> wrote:
>>It might be possible to run CAT5 or coax between houses. I've broken
>>the cable length rules many times and know most of the tricks. You
>>can go about 1000ft of CAT5 or RG-6/u (yeah, I know it's 75 ohms) at
>>10BaseT-HDX speeds without any difficulty. You'll get about
>>6Mbits/sec thruput. I've posted details previously. Please advise if
>>interested.
>Very interested, though I don't remember the discussions, please fill
>us in!
Basically, I've gone about 900ft with CAT5 and 1200ft with RG-6/u.
The article above explains some of the tricks and calcas involved.
The rules and limitations for coax are:
1. Must be 10baseT-HDX (half duplex)
2. No "Tee" connector in the middle of the coax run.
3. To short a run with RG-6/u are a problem.
4. Ground only one end of the coax or you will have a ground loop.
5. 50 ohm terminators, not 75 ohms.
6. Squirrels and mice love to eat coax cable. Conduit is nice.
>>Well, we'll see that when my test equipment arrives :-) I've got a couple
>>of 14dBi/20degree antennas to test with. My brother-in-law thinks it'll
>>be no problem, but that's one of the things I want to be sure of.
>
> Fire your brother in law.
Hey, I don't have a problem with that - but my wife's attached to him, so I
have to have cause :-)
> In my never humble opinion, this is a terrible choice.
LOL. I always hated "IMHO"!
> Here's some sample beamwidths extracted from random datasheets.
>
> Type gain -3dB Beamwidth
> dBI degress
> yagi 14 30
OK, you've lost me again. You're saying 30degrees, and I only have to worry
about half that. Where's the problem?
> You may have one potential problem with the cross lake link that's
> unique to microwave over water. There will be two paths. The direct
> path between antennas and the bounce path off the lake. They will
Yeah. I was afraid of that. I get a whole lot of electromagnetic radiation
off that pond - I use photovoltaic panels for power, so I know there's
reflection, I just have no idea how that works at wifi wavelengths.
>>It might even be necessary. My two neighbors are very close and I'm not
>>certain the one (who _is_ conveniently line of site to all the buildings)
>>doesn't block the other from the cross-lake line. If a single antenna at
>>the internet end could serve all three clients, the centremost house would
>>probably be in the Fresnel zone of the other house.
>
> Ok, time to get irate.... Would you kindly replace "very close" with
> some real numbers? I'm a big fan of calculating before I build and
> find it difficult to insert "very close" into my spreadsheet.
otoh, I'm a fan of experiment, rather than spreadsheet. Tomorrow (I hope),
I'll take my two antennas, an AP and a computer, and try them out from the
base station in a direct line to my house. Then I'll try aiming the base
antenna midway between the client houses. And if that works, I'll see what
I get at the problematic house.
"Very close" is as close as I can say. I don't have a measurement for angle
and I don't _know_ how close is too close, either in raw distance or angle.
> It might be possible to run CAT5 or coax between houses.
If it comes to that sort of thing, it just isn't even worth trying to help
the neighbors out. If I can't work a wireless solution that includes them,
I'll do it point-to-point for myself.
--
derek
Yeah, Google is my friend, thanks for the pointer, I'll try not to be
so lazy in the future...
>Basically, I've gone about 900ft with CAT5
I discovered the 'problem' with a run I had installed was that it was
500 feet long, and the 10BaseT was pretty good! Now I've gotta add
some Cat5 lightning arresters, as the D-Link router keeps blowing up.
8*}
On Thu, 11 Aug 2005 21:15:02 -0400, William P. N. Smith <> wrote:
>Jeff Liebermann <jeffl@comix.santa-cruz.ca.us> wrote:
>>http://groups-beta.google.com/group/...40b8ec03d59cfe
>
>Yeah, Google is my friend, thanks for the pointer, I'll try not to be
>so lazy in the future...
>
>>Basically, I've gone about 900ft with CAT5
>I discovered the 'problem' with a run I had installed was that it was
>500 feet long, and the 10BaseT was pretty good! Now I've gotta add
>some Cat5 lightning arresters, as the D-Link router keeps blowing up.
>8*}
Been there. I had a 200ft aerial run of RG-6/u on top of a mountain
at a radio site. Worked just fine until mother nature delivered a
nearby lightning jolt. Blew up a DLink DI-604 and two Milan/Digi
10baseT to 10base2 media converters. So, I replaced the router,
repaired the Milan converters, and promised to "do something" about
the lightning problem, some day, maybe. A week later, another storm
blew up the replacement router. At that point, I buried the coax
underground, inside some black plastic pipe and haven't had a problem
in the last 2 years.
On Thu, 11 Aug 2005 21:21:16 -0300, Derek Broughton
<news@pointerstop.ca> wrote:
>> Here's some sample beamwidths extracted from random datasheets.
>>
>> Type gain -3dB Beamwidth
>> dBI degress
>> yagi 14 30
>
>OK, you've lost me again. You're saying 30degrees, and I only have to worry
>about half that. Where's the problem?
You mumbled someting about a 14dBi gain antenna with a 20 degree
beamwidth in your 2nd posting. If you're trying to connect to two
stations 15 degrees apart, you cannot use a 15degree beamwidth
antenna. That's because the gain fall off at the -3dB points is
rather sharp. Even the slightest misalignment will be fatal. I think
20 degrees is not enough. 25 would be about right but 30 is the
closest available beamwidth.
>Yeah. I was afraid of that. I get a whole lot of electromagnetic radiation
>off that pond - I use photovoltaic panels for power, so I know there's
>reflection, I just have no idea how that works at wifi wavelengths.
It's a bit different. You don't get cancellations and reinforcements
with ordinary non-coherent light. If the sun generated only one
frequency of light, you would see interference bands (interferometry)
and have cancellations and reinforcements. Well, 2.4Ghz is basically
coherent spectra and does have the cancellation and reinforcement
effects.
>otoh, I'm a fan of experiment, rather than spreadsheet.
Did you notice my domain? LearnByDestroying.com
You don't really understand anything until you've broken it.
>Tomorrow (I hope),
>I'll take my two antennas, an AP and a computer, and try them out from the
>base station in a direct line to my house. Then I'll try aiming the base
>antenna midway between the client houses. And if that works, I'll see what
>I get at the problematic house.
Well, I can calculate what you will get if I knew the details on your
equipment. The acid test for a fixed installation is how much
attenuation you can insert in one of the coax lines and still have
things function. That's roughly your fade margin. 10dB is the
absolute minimum. 20dB is doing good.
>"Very close" is as close as I can say. I don't have a measurement for angle
>and I don't _know_ how close is too close, either in raw distance or angle.
Sigh. 50cents will buy you a protractor. I have a marine sextant,
but that's kinda cheating. Piece of paper and some drafting supplies
will also work.
On Thu, 11 Aug 2005 16:26:18 -0300, Derek Broughton
<news@pointerstop.ca> wrote:
>The difference is what? That's my problem. I don't have a clue what the
>wireless manufacturers are selling when they say "repeater". Can you point
>me to some links? It's especially difficult when I go to sveasoft.com and
>they tell me their firmware has "WDS/Repeater mode"!
A WDS repeater is a combination access point and repeater. It has to
be the same chipset at each point in the repeater chain so don't try
to mix hardware. The "range extender" type of repeater is quite
different, but is also chipset specific. There's quite a bit of
configuration involved in setting up the WDS repeater with the
advantage of having some control over which data gets repeated and to
where. The range extender just retransmits everything it hears with a
specified SSID.
>> I has all the advantages of a repeater, plus
>> you can plug computahs into the WDS radio to "tap" off a connection.
>
>I thought that's what the repeater was supposed to do.
No. The "range extender" type of repeater cannot handle a directly
connected CAT5 cable to the device. The WDS type of repeater can do
that.
>> No comment on range extenders and repeaters. Well, maybe a small
>> comment. I think they suck, work only in about half the systems, are
>> incompatible with much of the available hardware, and chop the
>> available bandwidth in half.
>
>As would the WDS, right? Half the bandwidth is just fine :-)
Yep. All these repeaters (and for that matter, all the wireless
hardware) is simplex type of operation as opposed to full duplex.
They can only talk or receive, one at a time. The result is that the
repeater has to wait until one end of the link does its thing before
it can retransmit the packet. That results in half the thruput.
Jeff Liebermann <jeffl@comix.santa-cruz.ca.us> wrote:
>A week later, another storm
>blew up the replacement router. At that point, I buried the coax
>underground, inside some black plastic pipe and haven't had a problem
>in the last 2 years.
Unfortunately my Cat5 is direct burial, and already in the ground. The
other house isn't currently in use, so it's not clear anyone cares if
I fix it, so I get some time to <ahem> contemplate my options. 8*)
> On Thu, 11 Aug 2005 21:21:16 -0300, Derek Broughton
> <news@pointerstop.ca> wrote:
>
>>> Here's some sample beamwidths extracted from random datasheets.
>>>
>>> Type gain -3dB Beamwidth
>>> dBI degress
>>> yagi 14 30
>>
>>OK, you've lost me again. You're saying 30degrees, and I only have to
>>worry
>>about half that. Where's the problem?
>
> You mumbled someting about a 14dBi gain antenna with a 20 degree
> beamwidth in your 2nd posting.
Right, that's what I have available right now. If I can get a good link
across the pond point-to-point, with those, then it would seem that I
should be able to get links to all three houses if I have a 14dBi / 30
degree antenna, then.
> If you're trying to connect to two
> stations 15 degrees apart, you cannot use a 15degree beamwidth
> antenna. That's because the gain fall off at the -3dB points is
> rather sharp. Even the slightest misalignment will be fatal. I think
> 20 degrees is not enough. 25 would be about right but 30 is the
> closest available beamwidth.
OK. That's working for me :-)
> It's a bit different. You don't get cancellations and reinforcements
> with ordinary non-coherent light. If the sun generated only one
> frequency of light, you would see interference bands (interferometry)
> and have cancellations and reinforcements. Well, 2.4Ghz is basically
> coherent spectra and does have the cancellation and reinforcement
> effects.
Got it. Thanks.
>
>>otoh, I'm a fan of experiment, rather than spreadsheet.
>
> Did you notice my domain? LearnByDestroying.com
> You don't really understand anything until you've broken it.
LOL. I hadn't noticed.
> Sigh. 50cents will buy you a protractor. I have a marine sextant,
> but that's kinda cheating. Piece of paper and some drafting supplies
> will also work.
Knowing the actual angles and distances does nothing for me if I don't have
a clue how close something has to be to be in the fresnel zone.
--
derek
> On Thu, 11 Aug 2005 16:26:18 -0300, Derek Broughton
> <news@pointerstop.ca> wrote:
>
>>The difference is what? That's my problem. I don't have a clue what the
>>wireless manufacturers are selling when they say "repeater". Can you
>>point
>>me to some links? It's especially difficult when I go to sveasoft.com and
>>they tell me their firmware has "WDS/Repeater mode"!
>
> A WDS repeater is a combination access point and repeater.
That's exactly what the D-Link DWL-G700AP I suggested in my first post is
supposed to be. So I _think_ I was on the right track.
>>> I has all the advantages of a repeater, plus
>>> you can plug computahs into the WDS radio to "tap" off a connection.
>>
>>I thought that's what the repeater was supposed to do.
>
> No. The "range extender" type of repeater cannot handle a directly
> connected CAT5 cable to the device. The WDS type of repeater can do
> that.
Ah, well, you said "into the WDS radio" so I didn't think you meant "plug"
literally.
--
derek
Critique my RAN (rural area network) 
Linear Mode
