Full Duplex on a single wire

[originator]

http://circuits.iwarp.com/fulldupidea.jpg

I am just getting into serial communication and was discussing Full Duplex on a single wire with someone, neither of us had a good solution, so no better place that Eng tips to find out whats the story!

Q. Is it possible to have a single wire(not including shield if req'd) to talk simultaneously back and forth, on top of each others signal? I was thinking that if the two signals "added" on the bus to increase the current when both were High, the state could be sensed and decoded on each end. Obviously this would require two pins at each end, Tx and Rx. I googled quite a bit and didn't see exactly what I had in mind. The other gentleman got it to work using a comparator on each end to sense when both were on, but said it didnt work fast enough.

Here is an idea I was kicking around(link above). First, convert the logic signal to sine waves (say what?), then treat the data bus as an audio signal with a psuedo summing bus on each end to cancel its own sides signal out at its own Rx input. If this has been done please let me know where to look. If not, is this idea even a slight possibility?

The concept is to have each sides output ride on top of each other. Phones do something similar right? If the decoded sine waves can get squared back off at each side thenmaybe this is doable?

Thanks for any ideas

 

[IRstuff]

Sure, and look at how fast a phone modem is.

TTFN

 

[itsmoked]

Do both directions have to operate at the same data rate? If the rates are different enough you can just pump in the signals and low/high pass them. This is what the fiber people do constantly. They frequency multiplex with different colors of light then just look at the color of interest.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[originator]

I thought that the modem limitation was more of a wire length/noise problem.

The speeds can vary, this is just for experimental purposes at this stage. Your idea is interesting Keith. I assume you are referring to full cycle sine waves. Would a half cycle sine wave function as well in this case? If so, it would eliminate the need to convert a 0>1>0 square wave into a full cycle sine, and would require minimal external parts to filter it. I think it wouldn't make any difference if it were fiber or wire either. I'll do some reading on it.

Thanks

 

[benta]

Yes you can.
This is basically what an analog (POTS) phone does. Originally with a transformer, nowadays with a socalled SLIC (subscriber loop interface circuit).
The circuit is also called a four-to-two wire converter. Basically both ends modulate onto the cable, and one's own signal (which is known, of course) is subtracted locally. Thus you can decipher the signal from the other end.

Your signalling speeds are unimportant, they can be the same or quite different.

Regards,

Benta.

 

[VisiGoth]

Yes. You can do it in analog or digital.
Using a standard that has been thought through debvugeed and has hardware and tutorials is your best best (IMO).

In analog the phone example is a good one. But POTS has hybrids, you may not have those. Today, you might want to consider a digital approach of multiplexing. Time division multiplexing is the most common and easiest to find parts and examples for. Frequency division is more expensive and relies more on what is considered analog. Code division is interesting, but if you are not versed inthis you might not want to do it at first. In code division there is always the problem with the patents to worry about. Hedy Lamarr's might have expired, but the rest have not.

jsolar

 

[originator]

itsmoked got me thinking about looking into the idea of running a fiber with two trancievers at each end. I am searching for hardware that could filter the colors right now. Most likely that route will require major hardware and cost, but is worth a slight departure to get some basic understanding in any case. SLIC ic's look inexpensive and not too difficult to implement, that route should be worth some experimentation as well. Thanks for the tips.

 

[logbook]

Well the key things you haven't mentioned are just how fast you want to communicate and how many you ever want to make. If it is only one then design time is critical.

Fibre would give fantastic bandwidth, albeit at fantastic prices (hundreds of pounds).

 

[lighterup]

There is a FO device called a wave division multiplexer (WDM) which accomplishes the color (wavelength) discrimination for you. These work well, and are used widely in the industry. The only drawback, at least from the days when I used to do a little FO work, is that they are difficult to manufacture, and are a bit pricey for a low budget commercial operation. Here are a couple of links for your reading enjoyment:

http://www.ofsoptics.com/resources/SplittersBrochure.pdf

http://www.canarycom.com/products/cwdm/p1.htm

 

[itsmoked]

originator;

What kind of application is this for?
What data rate are you interested in?
What distance?
What environment?

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[originator]

Keith

Applications will vary, but the concept I build will be based around the Propeller multiprocessor. Others will have their own interests as well. I have several uses in mind, but nothing nailed down yet. Possibly a continuation of the small business and hotel scenerio we discussed before. Data rates would be in the RS232 standards, maybe cat5. A few hundred feet would be nice using non-fiber. Home, office, and industrial environment typically.

The fiber route looks way to costly and bulky it seems, but I appreciate the links guys, very informative.

I wouldn't think too much on it, I was just wondering what the current options were, and I have gotten some great places to look at already from the post. The learning is the fun part.

 

[itsmoked]

Okay, I see, thanks.

Not sure how far this is good for.

http://www.i-fiberoptics.com/

If you're using potent processors on each end you could use straight time division. Just break the messages into periods and let each side talk half the time. Yes I know half duplex. A lot of apps work fine with it.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[originator]

That link has some interesting stuff. Check this out:

http://www.fiberdyne.com/products/cwdmintro.html

Precisely what you brought up first.

 

[originator]

http://www.fiberdyne.com/products/multiplex.html#top

also this one,

 

[itsmoked]

Oh hey I've never heard of "Coarse" interesting.

I did a temp control for a bleeding edge WDM system. It was nasty. The light filtering was done by mechanical means and so was dependant on physical positioning which was of course strictly tied to temperature. "Please make box stay at +/- 0.02C." Gag.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[originator]

"Oh hey I've never heard of "Coarse" interesting. "

Hey man anytime you need to brush up on electronics give me a shout. I am always there to help out. hehe

Light filtering, as in a prism?

 

[itsmoked]

Light filtering as in bandpass filters which are doped glass or coated glass or interferometers (mirrors) or gratings(narrow grooved etched glass) and yes in some cases prisms.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com