Very long transmission-line design tips

Question

I'm designing a very long transmission line between a master node a some slaves far away connected by a bus where only 3 wires are used: - A DC power supply (24vdc) - B Digital comm - C Ground

As there is only a communication wire, half-duplex is mandatory, so all slaves only answer when master asks them. The distance between nodes may by up to 100 m (110 yd), and up to 20 nodes may be installed. May be 2 Km long!!

By other designing reason, a relay may isolate the communication line from the rest of the slaves, they get inaccessible, this is ok. And the voltage on this line may be 5v or 10v depending on other conditions.

As I'm afraid of typical transmission problems (reflections, impedance-mismatch, etc) I have chosen a speed of 1200 baud.

My question is how may I estimate the impedance questions, how to overcome impedance variation. Is that speed low enough? Can be higher? Is this a realistic circuit?

This is the basic circuitry: And in the master and slave nodes, this is the communication diagram:

The X2 component is a microcontroller with the asynchronous terminals connected. It writes '0' when Q1 is on, otherwise a '1' is transmitted. The U1 comparator adapts the slightly high voltages to TTL levels. May be the plus-minus connection is reversed, doesn't matter at this design stage.

Resistors R3 and R4 (10k) have the function of polarize the diodes D1 and D2.

Please give me some light of how to save time of frustrating real tests. May be R3 and R4 have to be lower... But please take into account that Vb must be almost '0' when Q1 is activated. Current carried on line B is very low (as the 250 ohm resistors limit it), a low voltage drop is expected.

Thanks!

Answer 1

RS-485 is good for 100kbps at 1.2km, using two wires for data and two for power. You could probably stretch it to 2km at 1200bps.

Here is a TI application note for a 2-wire power + data solution. It couples the data transceivers through large capacitors, and connects the power source and sinks through normal-mode transformer filters.

1000BASE-LX gigabit Ethernet can run 5km over fiber.

A reflection may take 40us to travel 4km at 10^5 km/s. At 1200bps, you have 400us settling time from the leading edge of a bit to a sampling point a half-bit away.

In the comments, the OP mentions using untwisted, unshielded pairs. This will hurt noise immunity. Consider using parity bits, CRC's, retry protocols, or error correction.

Answer 2

One bit of advice. You'll probably find that transmitting unmodulated low speed data difficult to get the impedance right to avoid reflections and possible corruptions. Transmitting 100 Mbps is quite achievable on most coax cables over 100m these days without much hassle so please consider upping your data rate to push your transmit spectrum into the low MHz where the cable impedance is likely to be fixed at 50 to 150 ohms and purely resistive. You are also pushing your data spectrum away from power noise when feeding energy to your slaves.

One last bit of advise, consider using two wires for differential transmission instead of one and phantom power your slaves over the data wires. Cable selection will also be critical to this working effectively irrespective of my advise given above.