Why do we have cross-over cables.

It doesn't matter if one stops speaking or not. The data still goes both directions. Both *can* communicate simultaneously. You have obviously not watched women chat. ;-) ...or their husbands trying to talk to them on a cell phone. :-(

Data modems also use hybrids to separate transmit and receive data,

That's what a hybrid is for. The transmitter can subtract his signal from the combined signal. This doesn't require more modulation (bandwidth), instead can use "another" bit of dynamic range (S/N).

If you snuggle up a V-shaped piece of wire next to a transmission line, the capacitive coupling between the two will induce a voltage at starts traveling (as in, current will begin flowing) "left and right" from the apex of the V. On the other hand, the inductive coupling between the two will induce a current that flows in just one direction (opposite of that in the line you "snuggled up to," as you'd expect).

Hence, on one side of the apex you have currents that are heading the same direction, whereas on the other side they're heading in opposite directions and (some) cancellation occurs. With the right geometrical arrangement, at some frequency you'll have complete cancellation on that side (aka, "port"), whereas the other side still has all the currents in phase and gives you signal proportional to what's happening in the original transmission line. That's how you build a (crude) directional coupler using transmission lines.

You can do the same thing with transformers and resistors, where the resistors effectively sample voltage (the same as what the capacitive coupling did) and the transformer samples the current (as the inductive coupling did). This is what you find in telephones, known as a "hybrid" (since it's a hybrid of magnetic components and resistive components). It's also perhaps easier to understand if you're not fond of transmission line theory :-).

Finally, these days you can just use an op-amp to behave like a transformer and make it all solid state with no magnetics whatsoever.

"How does a directional coupler or hybrid work, at a qualitative level?" is probably a good interview question!

---Joel

A 'pin diode'.

EHHHH! Wrong ASSumption AGAIN, dumbfuckKK!

I NEVER once said that it didn't or couldn't, you RETARDED, PRESUMPTUOUS F*CK!

Good for you, idiot. The remark was about what is in place and in use NOW!

No s*it. I knew what OC-96 was before you even knew what an optical fiber was.

A San Diego company just recently got 'near wire speed' (unheard of) on the new 10Gb/s standards being developed.

They got 8.5Gb/s out of it. That is better than anything current. Most of the current stuff has far too much overhead. We are lucky if Gigabit Ethernet can push 600Mb/s on a good day.

They did it with a Cell Broadband Engine CPU.

There will be Cell processors in our future.

Hell, if we make it past 12-20-2012 things will actually look good.

Maybe we'll close off the borders and kick some illegal immigrant asses, and then clean up the gang problems too!

Maybe the horses will ride after all!

While the wire does not know or care, the end equipment is where all this is done. Between that and the properties of the wire/fiber connection, which can limit what can be done, the end equipment is king. .

For two counter-propagating signals superimposed on a transmission line, voltages add, but currents subtract. Knowing that for each signal, voltage and current are related by Z0, you can work out the individual signals going each way.

The design of a circuit doing so in real-time is left as an exercise for the student. ;-)

Practical implementations yield directivity in the 25dB ballpark, plenty good for simultaneous two-way traffic of binary signals over useful distances.

Jeroen Belleman

Hi Rich, It is all phase! Since the frequency band is know, you use some phase transormers, like 1/4 wavelength segments, so that your transmitted feed comes back 180 degrees out of phase, but the received signal is only 90 degrees. This effectively kills most of the transmitted energy (that would otherwise swamp your input stage) and gives you a better receive signal.

Of course, if loosing that phase information is important, your SOL...

Charlie

ITUT V.34 mananages 8 times that in a 1200Hz band

IF it actually gets a full 7 tier handshake, which is rare.

Depends on what you mean by "band". Sure, the signal is 1200 baud, but the sampling is done at 8kHz so it is a 4kHz band, at least in my book.

ISTR seeing it often quoted as 3kHz

I mean the signal level is -20dB at +/- 600hz from the carrier frequency

Maybe "spec'd" over 3kHz. The sampling is 8kHz; 4kHz Nyquist limit. To get full data rate V.90 needs every bit of that.

That's the spectrum of the "Info sequences", not data.

"INFO sequences are used to exchange modem capabilities, results of line probing, and data mode modulation parameters."

- and -

"All INFO sequences are transmitted using binary DPSK modulation at 600 bit/s ± 0.01%."