full duplex or half duplex

Full

They are full duplex, both parties can talk at the same time, data is transmitted in both directions. Modems are full duplex, they transmit data in both directions but at different frequencies. Here's my webbook that discusses data flow. There's a few pages on modems in there too...

Most are, some aren't. Mass-market consumer modems have pretty always been full-duplex. But, there are still a lot (as in millions) of half-duplex modems used in industrial process-control settings.

The full duplex ones do. The half-duplex ones both transmit using the same carrier frequency and therefore have to take turns.

Are that many left that are actually half duplex? I was never involved in "industrial process-control settings", so I don't have a good feel for how many used to exist, much less how many are still being used.

The telecommunications industry used to see relatively few to begin with, and that quickly diminished to virtually none when asymmetrical full duplex modems became available.

Most full duplex modems in use today use the full VF spectrum in both directions, and separate the two using echo canceling circuits. V.32 and v.34, for example, utilize all of the bandwidth in both directions. Earlier modems using 2400 bps or less generally use the different transmit frequencies for the two directions in order to get full duplex.

The original problem was that technology was not available to obtain more than about 1800 bps in one direction using the full bandwidth. The initial answer to that was to run a low speed channel in one direction (e.g., 75 bps) and a high speed channel in the other direction, which worked quite well for a terminal used for manual data entry from a keyboard... but wasn't worth much for anything else.

Some manufacturers then came up with modems that would switch the high speed channel from one direction to the other as needed. That worked very well for file transfers. In fact Telebit modems emulated certain protocols in order to reduce the number of times that a channel switch was necessary, thus making for some relatively very high speed data transfers if those particular protocols were used.

When technology made v.32 (and later, v.34) modems available at prices anyone could afford, the use of anything else rapidly diminished.

How do full duplex phones know which wire is send and which is receive? Is a normal phone cable some type of crossover cable?

It doesn't work like that. Both wires are used as what is known as a "balanced transmission line". The signal is applied as a voltage difference between the two wires (as opposed to being a voltage difference between a single wire and ground, which is what you are thinking of).

There is no real problem with having signals going both ways on the same pair of wires. Each end can hear each end. (There are a few very simple tricks used to make them sound about the same volume, but that gets a bit technical.)

I dunno about process control, but every CLID box has a 202 simplex modem.

It would surprise me at all if there werem't a lot of 202 and V.23 modems still in use in process control and SCADA. If your data rate is under 1200 bps, they're cheap and reliable.

For example, our water system has two tanks and three pumps at a total of three locations, and I'm virtually sure that the modems are 202's. The data rate is very low, just a tank level every few seconds, and for the pumps, basically one bit in every few seconds telling it to turn on or off, and one bit out telling whether it's currently on or off.

A disadvantage of the fancy signalling that modem V.92 modems do is that there is a very long setup time as both ends sniff the connection between them and decide how much data they can send in what bands. The 1200 bps modems, whether 202 or the full duplex 212's, which are basically a pair of simplex modems running an octave apart, have no training time at all, just connect and start talking. Credit card terminals all use 212 modems because the amount of data is small and the savings in transmission time for a faster modem would be swamped by the setup time.

R's, John

What if two phones keep trying to send a very positive signal?

Won't they keep stepping up the voltage?

Sure. One company I worked for is shipping over a million units a year that include half-duplex Bell-202 modems. I'm currently working on designs for two new products that will include half-duplex Bell-202 modems.

My ex-employer has been shipping between 500K and 1M units a year for 10+ years, and the average lifetime of these products is prbably somehwere around 15 years. There are other companies shipping similar quantities. IOW, there are tens of millions of half-duplex modems out there embedded in various process control devices. They're very rarely used over POTS lines (though I have seen it done).

Here's a web site describing the particular protocl used on the devices I'm talking about:

The Bell-202 modems I'm referring to are dead-simple electronically (no 75 baud back channel as allowed by the spec). Trying to use the same bandwidth in both diretions would have 1) required more analog circuitry, 2) used more power, 3) been pretty pointless, since the protocols used are pretty much half-duplex.

I remeber having Bell-202 desktop modems that implemented the

75 baud back-channel, but none Bell-202 implemntations used in process control used it.

That's true for IT usage on POTS lines.

In the process control area the switch from analog signal loops to digital transmission via FSK or manchester or whatever is just starting. Bell-202 1200-buad half-duplex is still somewhere between mainstream and cutting-edge.

Then they both send a very positive signal.

Eh?

The instantaneous voltage at any point on the line is the sum of the two instantaneous signal voltages.

If they're both sending 1V P-P signals, then you'll just see the sum of those two signals on the line (2V P-P if the signal are uncorrelated).

IIRC, most all of the water meters in Minneapolis have Bell 202 modems (I don't know if they're simplex or half-duplex), and those _are_ used over POTS lines. So that's several hundred thousand non-full-duplex POTS modems right there. I'd be surprised if a lot of other cities aren't doing the same thing.

Ideally, all that happens is the two signals add, and whatever the sum is, is what each receiver produces for sound output. That would mean that if you and I are talking at the same time on the phone, we may or may not be able to understand what the other person said... :-) (No different than if we are talking to each other across a lunch table, and do the same thing.)

However, there is an interesting catch that you might enjoy to idea of. If the two phones are a *long* distance apart, the signals can do some strange things. That is because the signal from your voice can actually be reflected from the distant end, and come back over the circuit. If the time it takes to do that is very short, your ears can't hear it. But if it is long enough to affect your ears, it starts sounding as if you are talking into a drainage ditch pipe! If the distance is *really* long (for example when a geostationary satellite system is used, and the distance is 23,000 miles each way to the satellite), you can actually say "boo", and then hear the entire word echoed back at you.

As long as the echo is *weaker* than the original, what you'll hear is not just one "boo", but a series of progressively weaker ones as it echoes back and forth. On the other hand, if the echo is actually *stronger* than the original, you'll hear maybe one boo, and then it will start squealing with probably a single tone (but possibly it will do strange things, with a tone that changes pitch, and one common description is "motor boating" because that is what it sounds like).

That brings up some interesting points too! Different modem technologies may seem "obsolete", but on the other hand may not be at all in certain circumstances.

Grant described the use of Bell 202 modems in industrial control processing, but can you imagine Bell 103 modems (300 baud FSK dialup modems) being useful? And can you imagine it being a

*step backwards* when those credit card verification machines went from 1200 bps modems (Bell 212A) to v.32bis modems (14.4 kbps)?

I don't know if they are still using these systems, but at least up to 10 years ago there was a "hotel" billing system for long distance companies which allowed a hotel/motel customer to get an immediate printout of all toll charges to any room phone. This was being done with Bell 103 modems! I'd expect it still is... because those modems use simple FSK, and will work over the most obnoxiously impaired telephone lines possible. They "lock on" within milliseconds and since the amount of data being sent is very small there is no disadvantage to using a 300 bps link. The advantage is that it works *every* time, no matter how bad the connection.

And those credit card verification machines, which some years ago were all using Bell 212A protocols. That isn't nearly as good as Bell 103 as far as being impervious to bad lines, but it will work on some lines that v.32/v.34 modems cannot handle. And in fact in the mid-1990's when they were switched to v.32 protocols there were a lot of customers in strange places (who were lucky to have phone service at all) that could no longer get credit cards to verify! Again, the amount of data made using a faster modem unnecessary; they apparently didn't have a clue about reliability, and made the wrong decision. They'd have actually been better off going in the opposite direction and putting out 300 bps modems! :-)

I used to work someplace that made ISA and PCI cards with 6 or

8 modems per card.

Just a few years ago, we had a customer complaining about some newly purchased multi-modem cards ) because Bell 103 mode didn't work in the last batch they bought. IIRC it had something to do with hotels or set-top-boxes or something like that.

IIRC the modem chips were from Rockwell/Conexant, and they just plain forgot to test Bell 103 mode when they rev'ed the firmware in the modem chips. R/C finally got us a new version of firmware, and we had to set up a production station to reprogram all of the boards we had in stock with the "new and improved" modem chips that didn't work at 300 baud.

A couple of things. Isn't CLID deivered via half duplex Bell 212A?

Secondly, the v.x modems can walk all the way back to the Bell 103 standard if you get a decent modem.

No, it's simplex Bell 202.

Those last three words are key.

R's, John