Spanning two houses 1/3 mile apart

Instead of feedback, would you settle for some calculations? The missing factor is how dense and thick are the trees. If you had bothered to describe the obstructions better, it might be possible to grind the numbers with more accuracy.

What I know (or guess) is: Two 24dBi antennas. Two 15dBi transmitters mounted on or near the antennas. cable modem which implies at least 6Mbits/sec required thruput. 1/3 of a mile end to end. Commodity hardware. Undefined amount of foliage Undefined line of sight and fresnel zone clearance.

See the FAQ at:

for how to do a microwave link calculation.

TX power = +15dBm TX coax loss = 1dB TX ant gain = +24dBi Distance = 0.333 miles RX ant gain = +24dBi RX coax loss = 1dB RX sens = -75dBm (for 36Mbits/sec connection) Fade margin = unknown

The system requires a 20dB fade margin (SOM) in order to be reliable. See the table for reliability and fade margin at:

20dB is considered a good MINIMUM fade margin.

Usually, there aren't any trees in the way, so there's no need to throw that into the calculations. The easy way is to just tack it onto the required fade margin.

My guess(tm) is that a single medium size tree, of medium density, and of the broadleaf variety, is good for about 15dB of loss. If you're going through two trees, it's 30dB. If you're going through a thick redwood forest (like my house) forget it. If you're going through a thick tree trunk, also forget it. I'll be optimistic and use 15dB loss.

Plugging into:

I get a fade margin of 41.3dB. Subtracting off the 20dB required fade margin, you have 21.3dB left for foliage attenuation (and any other losses I didn't include). That's not too horrible and sufficient to drill your way through one medium size, medium density, tree. Any more and forget it.

Note that this assumes that you have nothing inside the Fresnel Zone. At 0.33333 miles, you need at midpoint about 10ft radius around the line of sight of clearance. Obviously, the trees are in the way. However, so might the ground. The access point on a "hill" is ideal but if you have some other junk along the line of sight, it might add additional losses.

Reply to
Jeff Liebermann
Loading thread data ...

I suggested that the ethernet switch *AT MIDPOINT* act as a repeater. That would be only 135 meters, which is over the 100 meter specification, but should work for light loads of a few watts. If you want to follow the rules exactly, then two repeaters, powered from each end via 802.3AF PoE arranged with:

100 meters PoE Ethernet 75 meters PoE Ethernet 100 meters CAT5 switch CAT5 switch CAT5

This way, the cable run between the PoE injector and the PoE Ethernet switch is 100 meters.

The problem is that I can't fine a cheap ethernet switch that is suitable for doing this. I found repeaters and media converters, but they're not PoE powered. I also found switches that will run PoE devices, but cannot be powered by PoE. I'm sure they exist, but I can't seem to find any.

As for calcs, I'm sure I can make this work without using the offical

802.3af specifications and hardware. You want calcs so I'll just run off a sample. However, I have to make an assumption on the power and dissipation of the ethernet switch at midpoint. I'll guess 12VDC at 0.5A.

CAT5 is 3 ohms per 100ft per conductor. 135 meters (443ft), is therefore 13.3 ohms per conductor. The extra four wires are usually doubled up, and the loop resistance is twice the per conductor resistance. Therefore, the loop resistance is 13.3 ohms.

So, the question is how much voltage is required to drive a load that wants to see 12VDC at 0.5A through a resitance of 13.3 ohms. The current is 0.5 through both the 13.3 ohms and the load. Therefore, the voltage drop across the 13.3 ohms is: E = 13.3ohms * 0.5A = 6.6 volts Add that to the required 12V at the load and we have an input voltage of: E = 6.6v + 12V = 18.6VDC That's not too outrageous. The problem is that the 13.3 ohm CAT5 resistance is going to vary somewhat. It makes sense to use a voltage regulator in this application (if the ethernet switch doesn't already have one inside). So, the input voltage will need to be increased for the required forward voltage drop in the analog regulator which is perhaps 5VDC. So, 24VDC should do the trick.

Of course, different switches will require different voltages and currents, so the numbers will vary substantially. This just demontrates that it is possible to run a cheap ethernet switch via CAT5 without PoE. Just remember to use a fuse.

Reply to
Jeff Liebermann

Yep. You just reminded me of Cisco Long Reach Ethernet (LRE).

5-15Mbits/sec ethernet to 5,000 ft. About $175.

That's the receiving end usually used with a Cisco Catalyst 29xx series LRE switch. I don't know if two 575/585 boxes can be used together as an ethernet extender or if a 29xx switch is required.

Reply to
Jeff Liebermann

For my money, considering also you want to use it for a phone connection also, the wifi solution in a borderline environment will end up a little short on your expectations. Even with reasonable investment in hardware, basic absorption physics will slow it down to a level that may be basically acceptable for idle internet browsing and email only provided you do not mind the number of retries that will occur on the wifi link.

Others have indicated a copper solution and my own experience has shown the

10 baseT being carried over some decent low loss coaxial cable to be extremely effective and reliable. Thin ethernet using RG58 type cable has been ok up to about 400 metres given good transceivers such as 3Com. Going to a higher grade coax such as Belden RG213 or RG214 was very reliable at a distance of almost a kilometre. These distances are outside the specification but have been shown to work well given good transceivers such as above. While 10 base ethernet cards are now not common, there are still plenty around that can usually be had for a song or less.

Peter

Reply to
Pierre

i think the cisco stuff is end of life. No replacement.

formatting link

Reply to
stephen

"stephen" hath wroth:

Cool. EOL means I can probably afford to buy one on eBay. Yep.... there's a bunch of 295x switches for $200 to $300 and one 585 CPE transceiver for $120. I can't tell if the 295x devices have LRE capeability, but there's one 2924 switch with LRE for about $300.

Let's see how wired compares to wireless:

Wireless: 2 cheap bridge radios $140 2 24dBi dish antennas $160 2 PoE system $120 Assorted conn and boxes $150 ==== Total $570

Cisco LRE: 1500ft gel filled CAT5 $200 1 Cisco 2924 LRE $300 1 Cisco 585 CPE (new) $140 ==== Total $640

Hmmm... not so good. Maybe if the LRE switch can be found for less at auction instead of Buy it Now.

Well, if all else fails, we're back to the ethernet extenders:

A pair for $400. 1 mile, 2 wires, 16Mbits/sec.

Reply to
Jeff Liebermann

Not sure why there'd be noise from 48V _DC_. I guess if its poorly rectified.

Reply to
Mark McIntyre

He has 550m to span doesn't he? Midpoint would be at 275...

You can run any low-voltage unit over PoE, no need to buy special hardware. I successfully ran a wireless router at the end of 30m of cat5 by just snipping the plug off its power cable, attaching this to the 4 spare cores, attaching the other end of the cores to the wall wart.

(snip interesting calcs)

Indeed, better than I expected.

Reply to
Mark McIntyre

Oops. Y'er right. 1/3 of a mile is 531 meters. At 100 meters per PoE run, that would be too far for 802.3af PoE from each end to work. Sorry.

Reply to
Jeff Liebermann

No, the noise is from cross-talk and reflections; where noise is defines as any unwanted signal.

Reply to
decaturtxcowboy

Ah, you mean noise in the signal rather than the power. Ok, I'm with you there.

Reply to
Mark McIntyre

I've noticed a lot of folks pushing copper and repeaters. Dunno why. Seems like a hell of a hassle. I'm gonna jump in and suggest fiber. Fiber to Ethernet transcievers can be purchased for $60/each (you'd need two) and 50/125 or

62.5/125 fiber can be had for pretty cheap. I bought 2000' of it about 18 months ago on ebay for $250. Might take some hunting to find a good deal, but I've been happy with my 830 foot double run (so I had a spare circut). I buried it in PVC and haven't had any issues at all.

You'd need about 2000'. (they usually sell it in round numbers)

50/125 OR 62.5/125 multimode fiber with the $60 transcievers will work up to 2000 meters or roughly 6562 feet.
Reply to
Johann Beretta

Cuz if it's a cabin, probably doesn't have power, so how you gonna power the fiber converter.. Even wireless, if there's no power, what you gonna do at the cabin end? If you do the direct burial power cable/utility direct burial cable, and networking over the power cable, you kill two/three/four birds with one stone (ie get both power network, and can even do phone and intercom etc) (he said there was a neighbors house about 1/3 mile away)

When I did mine I tried wireless first, but during the winter snow came (31 inches in the first storm, destroyed the fresnel zone), then spring came and trees/plants grew (wireless doesn't go well thru water/leaves), by summer I was looking for plan B, and the buried cable to do power/networking/phone/cable tv/intercom/etc made a bunch more sense..

Reply to
Peter Pan

Telephone lines for fax and voice, lines for an extension of your main phone system, dry pair alarm circuits, the list goes on of what needs a copper pair to work.

If you're going to run fiber, by all means run some copper while you are at it.

Reply to
DTC

I assume there is some power, since he wants it for his laptop and vongage phone. Yeah, a laptop has batteries, but that's good for what? A couple of hours? Besides, they have these awesome things called generators :)

Never said the other ideas sucked, just said they seemed a hassle. Didn't even see your suggestion. I was just pitching a suggestion myself.

Reply to
Johann Beretta

That'd work. No reason one can't lay down both. Fiber would just eliminate a lot of the hassle of ethernet repeaters or having to use a slower technology.

50/125 fiber can handle up to 1000mbits and it's cheap.
Reply to
Johann Beretta

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.