More than 4 repeaters?

[kag]

Hi,
Me and a friend are debating why you shouldn't have more than 4 repeaters (the 5-4-3 rule). I agree that there is a reason why that rule exists, but I don't fully understand where it comes from.

I know I am wrong, but I don't know where. I say that if each repeater can clean up the signal at the other end of a 100M Cat5 cable, I don't see why there can't be say 10 repeaters in a row.

If we divide the problem into sub-problems (each sub-problem being from the "out" end of the Nth repeater, in a 100M Cat5 cable and in the "in" end of the Nth+1 repeater), and each sub-problem is correct, I don't see why the whole problem shouldn't work. Suppositing this is true, say the first repeater receives a signal without errors (because of the 100M standard is respected). So when it repeats the signal to its other end, it sends out the same error-free (but amplified) signal. If the next repeater is within 100M of the first one, it should receive an error-free signal also, etc etc.

To me, when a repeater receives a correct signal, it can send it to the next one without errors if the 100M standard is respected... no matter how many repeaters there were before the Nth one, if it has an error-free signal, it should be able to continue the communication...

They probably decided on 4 repeaters because of the latency time, or the time needed to re-send the frames if a tramission fails... but we can't agree or anything, hehehe. Tell me where I am wrong.

 

[CTho9305]

I was under the impression it had something to do with routing and wasted traffic, rather than signal integrity.

 

[kag]

You are probably right, but I am looking for actual facts
Thanks for stopping by though!

 

[zetter]

I was under the impression that the 5-4-3 rule only apples to 10Base5 and 10Base2 networks and not to 10BaseT.

 

[Pudgygiant]

Originally posted by: zetter
I was under the impression that the 5-4-3 rule only apples to 10Base5 and 10Base2 networks and not to 10BaseT.

Same for me. No hard facts though.

 

[zetter]

A quick google search will give you lots of links.

The whole 5-4-3 rule has to do with the time it takes a signal to propagate from one end of the segment to the other end. If a segment is populated, there is a chance of collisions on that segment, and the maximum back-off times for the transmitters involved in the collision will have to be added to the potential total transmission time. If there are more than 3 populated segments, the potential possible transmission time becomes too long, and what is called the 'collision horizon' becomes shorter than the segment, so a workstation at one end could potentially not see a collision at the other.

 

[AFB]

So it all comes down to latency.

 

[Wolfie]

Correct me if I am wrong, but doesn't a repeater also repeat noise?

Takes what it can "hear" and pumps it back up to full strength, including noise, and shoots it off to the next one. And so on. Wouldn't the noise be so bad at the other end after picking up the noise from EFI and other sources along the way that by the time it got to #3 or even before that, that there would be so much noise that it would cause other issues?

Wolfie

 

[AbsolutDealage]

Originally posted by: Wolfie
Correct me if I am wrong, but doesn't a repeater also repeat noise?

Takes what it can "hear" and pumps it back up to full strength, including noise, and shoots it off to the next one. And so on. Wouldn't the noise be so bad at the other end after picking up the noise from EFI and other sources along the way that by the time it got to #3 or even before that, that there would be so much noise that it would cause other issues?

Wolfie

In a typical system, collisions will be the primary problem, not signal noise. Unless the wires are passed through some severe interference, the signal should still be readable on the other end. It really is all about the latency for collision detection here.

 

[Pulsar]

Repeaters are not amplifiers. They do not take a signal, boost the amplitude, and send it on it's way. They receive the signal and *re*-broadcast it, I.e., create a new digital signal.

 

[kag]

Originally posted by: LsDPulsar
Repeaters are not amplifiers. They do not take a signal, boost the amplitude, and send it on it's way. They receive the signal and *re*-broadcast it, I.e., create a new digital signal.

Although I struggled to explain what I meant, this is what I thought was correct.

 

[sao123]

delay is significantly the problem. if a packet is no recieved on time it may be requested for retransmit or even lost. essentially timeout occurs causing deadlock.