Wave Interference, and bidirectinal 1 wire network

hinv

Hi,

I was watching this video: http://www.youtube.com/watch?v=lt8z9xLFBDA&feature=relmfu
when it hit me. Collisions as described to me when studying ethernet, really aren't collisions, but since the receivers on the line
aren't complicated to decipher the waves passing through, so a "collision" is detected, and both transmitting parties have to retransmit.
Why couldn't you make a point to point network with just 1 wire for both receive and transmit bi-directionally?
If waves really pass through each other, you could have a receiver that automatically subtracts the transmitted signal(if any) from the received signal?

I don't know if this is the right place to ask, but I know there are people here that would know the answer.

Thanks

LoopyByteloose

Dallas Semiconductor has a whole product line based on 1-wire serial communication. You might investigate their technology to find out the limitations.

Parallax forums are generally very open to any innovative thinking. But a lot of ideas just are not as good as existing solutions. Differential transmission via RS422 and RS485 protocols are very good at noise limitation and I am not sure that trying to push everything to one-wire has any real advantage.

The 1-wire technology that Dallas Semiconductor provides has limitations as to distance and problems with driving the lines with appropriate timing. In fact, certain products tend to violate their original timing standards as they later discovered that the specs were too tight to meet.

Mike G

I was watching this video: http://www.youtube.com/watch?v=lt8z9...feature=relmfu
when it hit me. Collisions as described to me when studying ethernet, really aren't collisions, but since the receivers on the line
aren't complicated to decipher the waves passing through, so a "collision" is detected, and both transmitting parties have to retransmit.
Why couldn't you make a point to point network with just 1 wire for both receive and transmit bi-directionally?
If waves really pass through each other, you could have a receiver that automatically subtracts the transmitted signal(if any) from the received signal?

In the context of the video, if two digital nodes are transmitting who is listening?

Phil Pilgrim (PhiPi)

I think the question really is: "Can you have full-duplex communication over a single wire (or wire pair)?" The answer is yes. You do it every time you talk on the phone. Both ends of the conversation are cairried simultaneously on a single twisted pair. Each telephone uses what's called a "hybrid circuit" to separate the incoming and outgoing signals at the handset. Hybrids used to be audio transformer affairs, but in more modern phones, the separation is done with active electronics. That's one reason they're so much lighter than their older cousins.

I see no reason that one could not make a hybrid circuit that would do the same for full-duplex digital data carried over a single twisted pair.

-Phil

kwinn

hinv wrote: »

Hi,

I was watching this video: http://www.youtube.com/watch?v=lt8z9xLFBDA&feature=relmfu
when it hit me. Collisions as described to me when studying ethernet, really aren't collisions, but since the receivers on the line
aren't complicated to decipher the waves passing through, so a "collision" is detected, and both transmitting parties have to retransmit.
Why couldn't you make a point to point network with just 1 wire for both receive and transmit bi-directionally?
If waves really pass through each other, you could have a receiver that automatically subtracts the transmitted signal(if any) from the received signal?

I don't know if this is the right place to ask, but I know there are people here that would know the answer.

Thanks

This has been done, and with a relatively simple single op amp circuit. The transmitted signal is subtracted from the summed signal received from the 1 wire connection using an op amp. This leaves only the signal from the other transmitter as input. This was required due to a mistake that left them one wire short in a long buried cable. Usually it would be simpler to add another wire.

NWCCTV

Phil Pilgrim (PhiPi) wrote: »

. Both ends of the conversation are cairried simultaneously on a single twisted pair.

-Phil

Phil, I am pretty sure you are right but I have one statement on this. When I was in the Military in the 1980's I was in communications. Field telephones are very similar to standard telephones. We had to set up 100 foot antennas and relay secure commo for the Generals using a KG-27 in between. We were taught that in the 2 wire system, one was transmit and one was receive. We use to mess around and disconnect the transmit wire in between calls. We could actually listen to conversations and the other end could hear also, but when they spoke it did not get through, which leads me to believe that theory.

Phil Pilgrim (PhiPi)

That may have been true with field telephones, although I'm skeptical. It's definitely not true with domestic landlines. The twisted pair carries both ends of the conversation differentially. Here's a Wikipedia article on telephone hybrids:

http://en.wikipedia.org/wiki/Telephone_hybrid

-Phil

Circuitsoft

Gigabit Ethernet does this. All four pairs are used for both directions of transmit, and at full duplex.