ADPankow

I need to setup a network between two places about 1200 ft. apart, I am wondering what would be the most economical and easy setup? I am open to any options be it wired or wireless...

MRK

You have several optoins.
Since ethernet cable (cat5) can only run 300 feet, (i think...) you cant use cat5.

If the two places are in line of sight, i would build to directional antenna's and buy 2 wireless ap's and point the antenna's at each other and boom. you have a connection.

You could also get a special circut from your local phone company.... lol. but that's not economical.

I'm sure there are other ways, but i'm to tired to remember.

Good Luck!!

-Mr.K

ADPankow

Ok so I will probally do a LOS wireless network, but one more question about it... what would be the lowest gain antenna I could use to go the distance? Thanks for your replies!

agentgrn

You're right in that it can only run 300' (actually, 100m per spec), but it's got nothing to do with cat5 per se.

The spec is specifically tied to the properties of the electrical signal. Without too much technical explanation, after 100m, the attenuation gets to be too great which sets the ends up for errors over the cable. It doesn't matter if the cable is Cat3, 5, or greater.

__________________
-A.G.-

agentgrn

There was something I didn't ask...

What's between the 1200'? Even though the expertise here is geared towards wireless, your application might be better suited to another medium.

__________________
-A.G.-

ADPankow

I do not believe there are any obstructions between...

netrambler

I don't mean to be nit-picky. but as I understand it, the limitation on Ethernet really doesn't have as much to do with the attenuation per se (i.e. the increased resistance that comes with longer pieces of copper cable) as it does more with timing issues and round trip times.

And I believe that the Ethernet spec states a limitation of 328 feet (100 m) between segments.

__________________
RTFM - Always a good idea

agentgrn

I was hoping not to get too technical...but you leave me no choice!

It has everything to do with attenuation. A lot of you youngin's didn't have the pleasure of working with Ethernet when it was coaxial...

Ethernet

The Ethernet specification deals with a 10mbps manchester-encoded signal in a bus network over a variety of mediums, particularly 10base5 (50-ohm, thicknet), 10base2 (50-ohm, thinnet), and 10baseT (unshielded twisted pair).

At the absolute maximum, the total length of a single collision domain is 2.500 meters in Ethernet, over a maximum of 5 segments, which would take four repeaters, three segments of which are allowed to be populated...hence what is generally known as the 3-4-5 rule.

A note on 10-base T...is that even though it is physically wired in a star configuration, it functions as a logical bus.

The absolute maximum distance was based on the propogation time from a station at one end of the bus to reach the other end in time for a collision to be detected within a limit of 512 bit times...or roughly the time it takes to transmit 48 bytes...essentially the MAC header. This is called a collision domain.

If the distance between the hosts was over the maximum limitation within a collision domain, the possibility of a collision after the first 512 bit times would result in a more serious error called a Late Collision. In the case of a late collision, transmission is not even reattempted and it is up to the higher-layer protocols to retransmit the packet, since an LC should never occur in a properly designed and functioning system. LCs are always a sign of trouble somewhere. (with the advent of Ethernet switching, it's almost always related to duplex settings)

Collision detection circuitry listened on the medium during transmission, and if the channel was clear, would allow transmission to proceed. If another station started to transmit, it would observe an electrical signal that was stronger than the signal being transmitted (transmit signal x + recevied signal y > threshold), record a collsion...perform a backoff delay, then attempt retransmission again (up to 16 times).

However, for collision detection to work correctly, the signal strength of another station has to be strong enough to exceed the threshold of the circuitry in order for it to work properly. This is what cable attenuation was extremely important.

10base5 cable, or thicknet was large and bulky, and subject to much less attenuation than the smaller 10base2 cabling. You could liken this to comparing LMR-400 and LMR-100. Even worse was 10baseT twisted pair since there was no shielding and even higher attenuation over a longer distance.

Fast Ethernet

When Fast Ethernet was introduced, in the interests of retaining compatibility with Ethernet, the same encoding and collision detection methods were employed as in Ethernet. However, due to the faster requirements needed for collision detection, the maximum segment length dropped from a 2,500 meter maximum to a roughly 205 meter maximum length from station to station.

Fortunately, the IEEE threw out coax in the FastE spec, which left us with mostly twisted pair, but now on Cat5, which specifies 3 twists per inch instead of 3 twists per foot like Cat3...thus reducing the effects of outside interference and crosstalk.

Since the attenuation is roughly the same as Cat3, the 100 meter limitation remained the same. However, it also allowed for two different classes of repeaters, a Class 1 and a Class 2 repeater.

A class 1 repeater allows the integration of the three different types of FastEthernet to be used within a collision domain. 100baseTX, 100baseFX, and 100baseT4. Since this is translating between mediums, there is some delay introduced which affects the round-trip times. Only one is allowed per collision domain, with up 100m from the station to the repeater.

Class 2 repeaters are akin to hubs, and introduce very little delay, thus two are allowed per collision domain. Thiis allows for a maximum of 100m from each station to the repeater, and up to 5m between repeaters.

This is why "stackable" hubs became important in Fast Ethernet networks, as they would function within the spec as a single repeater, even though there was more than one device involved.

GigE

Since GigE does make some changes to the rules as far as timing and collisions and stuff like that, I'm not going to describe that in detail here.

Switching

When switching is utilized between stations, practically all the propogation delay stuff goes away, and all that's important is the loss to attenuation. Removing collision detection in the case of a full-duplex link can theoretically allow the length of an ethernet segment to be longer than the maximums that were previously defined.

Vendors such as Cisco Systems have products that support long-range Ethernet or LRE, that can easily exceed 5,000 feet over Cat3. Since these devices are often full-duplex and switched, collision detection isn't even required.

Conclusion

So, who's the fucking ethernet geek now?

__________________
-A.G.-

netrambler

Thank you for your very technical response. And you're right, I never have had the pleasure to work with Ethernet when it was still used over coax. And as you pointed out in your response, attenuation does matter (you learn a little something every day).

Having never dealt with thicknet (only twisted pair), that is what I am most familiar with. As I understand it (and correct me again if I'm wrong), you have to have a minimum ethernet frame size that equals the amount of data that can be transmitted in the time required for a signal to propogate from one end of the network to the other and back. Otherwise, you could have a sender think that it sucsessfully sent a frame when it actually collided with another frame. So you're stuck with a design condition: as the network speed goes up (i.e. faster tranmission rates), the minimum ethernet frame length must either go up or the maximum cable length must go down. That's what I based my response on. :-)

In any case, thanks again for your info. And BTW, I never considered myself the "fucking ethernet geek".

__________________
RTFM - Always a good idea

agentgrn

LOL...the wink at the end of my reply was supposed to indicate a touch of humor.

Yeah, coax sucked...but it can still be found in some places, scary enough. Then again, back in the day, it was (relatively) cheap and very fast...compared to the file sizes we're used to running around and about nowadays.

For the mostpart, the faster the speed of the system, the shorter the maximum length of the cable system needs to be. GigE tosses the conventional rules out, however it would be kinda silly to deploy a GigE solution with a maximum usable distance of 20m. Yes...GigE has a specification for a half-duplex, collision detection segment. Thankfully, nobody ever made any GigE half-duplex equipment.

__________________
-A.G.-

netrambler

Thanks

__________________
RTFM - Always a good idea

ADPankow

Ummm don't suppose you can answer my question, on the minimum requirement for an antenna (In dBi) to reach 1200', could you? Thanks.

agentgrn

A couple low-gain yagis should do the trick...1200' isn't really that far away.

__________________
-A.G.-