Node Ring Connections

[Dave]

For discussion . . . . . .


Recently, I have been trying to get a Node system running, unfortunately, the ROMs that I am trying to use have a problem and won't let the MTX boot. However, as part of the effort to get a ring running, I have tried to work out how the nodes should be physically connected.

My "best guess" on how the ring would be cabled is on this page :

http://www.primrosebank.net/computers/m ... txnode.htm

Having read the page, Bill (Brendling) made a few observations on the subject of resistors. I have included some of Bill's comments below. Bill suggested that we open this conversation to a wider audience, hence this thread. Hopefully, it will stimulate further discussion and we can make a better "guess" on how the system was originally connected.

" . . . What I think is more likely is some resistors for impedance matching to the coax cable. I would expect:

A series resister between the TX pin on the 25-D connector and the centre of the "Data Out" coax connector.
A parallel resister from the "Data In" connector to ground.

The value of these resistors would be either 50 ohm or 75 ohm depending upon the impedance of the cable used.
The effect of these resisters would be to halve the voltage seen on the RX pin, but more importantly it would reduce the occurrence of signal reflections at the ends of the coax cables. This would be important particularly for longer cable lengths."



"A resistor from the signal pin to ground (as I describe for the RX pin) is often called a "terminator" or "terminating resistor". At the transmit end you can either have another resistor to ground and then a constant current drive, or (as in the case of the MTX serial port) a constant voltage drive and a series resistor."



"If I remember the chip specs correctly, the TX line drives at +/-10V, while the RX line only requires +/-3V. More of a concern is the current consumption. If 75 ohm coax is used then each transmitter would see 150 ohms. and would have to supply 67 mA. With 50 ohm coax it is worse, the transmitter is driving 100 ohms and has to supply 100 mA."



Anyone else care to comment?

regards
Dave

[1024MAK]

Driver chips, designed for RS232 have to be current limited to meet the specification.

The distance a RS232 link can work reliably is inversely proportional to the baud rate (or more correctly, the bps rate). So at lower speeds, using twisted pair screened cable, it is possible to have a link 1/4 mile long.

Bill is correct, in that termination resisters should be used. However, normally higher values are used. Often industrial equipment has the correct value termination resistors already fitted.

Oh, and talking about resistors, cheap and nasty networks may use a resistor in place of your proposed relay contact. How well it works is another matter...

More later when my brain is awake

Mark

[Dave]

Hi Mark,

I agree with what you say about cable length vs. baud rate, I should have put the 9600 baud at 15m on the web page, but as Node is running at 9600, then the accepted norm of 15m is applicable in this case (without line drivers). As I intimated, the limiting factor is cable quality, the real point was to show why Memotech had chosen coax for the cable type

regards
Dave

[1024MAK]

@ Dave, I think you misunderstood my earlier post. I was just pointing out for the benefit of the people who do not know, that RS232 can go further than a lot of them think.

Right, now most RS232 driver chip outputs include a 300 ohm resistance. Also some have outputs that limit the short circuit output current to 12mA (for example the 1488 driver).
So for these chips, no extra resistor is needed at the transmitting end.

At the receiver, I think that 75 ohms would be too low a value. 1k is a more realistic value.

Some data sheets:
http://www.onsemi.com/pub_link/Collateral/MC1489-D.PDF

http://www.ti.com/lit/ds/symlink/ds1488.pdf

http://www.onsemi.com/pub_link/Collateral/MC1488-D.PDF

http://www.ti.com.cn/cn/lit/ds/symlink/mc1488.pdf

Somewhere in a file, I have a lot more information. Alas, I currently do not have easy access to it

While searching, I came across these sites which provide interesting cable information (although most of this is nothing to do with RS232):
http://www.newark.com/pdfs/techarticles ... _Cable.pdf
http://www.epanorama.net/documents/wiri ... dpair.html


Mark

[Dave]

Hi Mark,

Aah, OK, thanks - I see.

As you say, there's a lot more to designing an RS232 link than one might first think - cable considerations alone is a topic in itself.

Although we can speculate and could even design a modern day serial port to coax interface that would probably work better than the Memotech original, it is shame that it is unlikely that we will ever see an original Memotech "dongle" to see how they did it

regards
Dave

[Bill B]

Mark,

If you want to avoid reflections in the coax, the source and destination impedances must match the characteristic impedance of the cable. See, for example, http://en.wikipedia.org/wiki/Reflection ... ting_lines.

It may be that for 9.6kb or even 19.2kb serial transmission over reasonable lengths it is not important and there is time for sufficient reflections for the output voltage to stabilise before the data is sampled. Correct impedance matching certainly matters for radio applications. It is also important for higher speed data transmissions.

Bill.

[1024MAK]

Bill, I agree 100%. At every impedance mis-match, reflections of the signal will occur including at the combined termination of the receiver chip and 1k resistor that I suggested.
However, if the driver chip includes a 300 ohm resistance in the output, and the output is current limited to 10mA (short circuit current is 12mA max), it's not possible to properly match it to 50 ohm or 75 ohm coax.

The 1488 driver and 1489 receiver is shown on the schematic for the communications board, hence why I have only considered this type of RS232 line driver and receiver. Some of the other line drivers have higher output short circuit currents and therefore could be operated with a lower value termination resistor that better matches the cable impedance.

Now it maybe that what Memotech meant was screened data cable. The impedance of this cable may be somewhere between 100 ohms to 600 ohms.

Also as you say, it may be that the design of RS232 systems ignores proper impedance matching. RS232 was developed from 20mA current loop teletype systems operating at baud rates of less than 300 baud. I think people just tried moving further up the frequency divider for the clock, and that's how the higher speeds came about.

Returning to actual practical RS232 systems, I don't remember seeing any discrete resistors in the output circuit of any RS232 equipment. I'm sure that I have seen some resistors in the receiver circuitry for some RS232 equipment, but the rest of the RS232 receivers don't have any termination resistor that I remember.

Actually thinking about it, the Acorn BBC RS423 interface has termination resistors as an option.

Mark