Communication method for long distance 3coreelectrical cable

[Mesuty]

Hello
In my last project , I have finf a reasonable way to communicate 2 PIC devices in a 3 km distance. Communication rate can be as low as 115 Baud. The communication will be performed over an classical 3 core , 2.5 mm2 electric cable that does not used to carry electricity anymore. Because of the geographical conditions , I have no chance to replace that cable with a usual comm cable.

Could someone suggest a proper way ?

[Guest]

[ljbeng]

Look at MAX485 or similar RS485 chips.

[Guest]

According to spec, RS485 transceivers can drive a twisted pair line as
long as 1.2Km, the reflections and signal losses in the classical 3 core in a
3Km distance will be an invitation to problems.
I would try using in the Tx side a MOSFET switch and in the Rx side
optocouplers (type 6N137) with Digital Outputs as front ends to overcome
noise problems.
The use of a Tx software handler to adjust the bit-timings to compensate
the low slew/rate in the other end to keep the original shape will be a
must. Of course there are more considerations to take in mind but this is
the general rule I will follow.

Humberto

[Humberto]

According to spec, RS485 transceivers can drive a twisted pair line as
long as 1.2Km, the reflections and signal losses in the classical 3 core in a
3Km distance will be an invitation to problems.
I would try using in the Tx side a MOSFET switch and in the Rx side
optocouplers (type 6N137) with Digital Outputs as front ends to overcome
noise problems.
The use of a Tx software handler to adjust the bit-timings to compensate
the low slew/rate in the other end to keep the original shape will be a
must. Of course there are more considerations to take in mind but this is
the general rule I will follow. My $0.02

Humberto

[SherpaDoug]

I would go to the site, short one end of the cable, and measure the noise at the other end. If you are going to use 1200 baud or slower a DVM will do for noise measurement, no need to drag out a scope.
I will bet that straight RS485 will work just fine. You may need some RC low pass filters if there is lots of noise. For a 100 Ohm cable impedance (a reasonable guess) and a corner frequency of 3kHz (to support 1200 baud) a capacitor of 3.3uF (nonpolar) should work. I would start with the usual 100 Ohm terminating resistor with a 1uF ceramic cap on each end of the cable. If it almost works try doubling or trippling the caps.
The delux version would be a low pass ladder network of series resistors and parallel caps, but I seriously doubt you need to get that complex.
Adding a checksum to the messages and a way to request a resend of any that are garbled would be nice as well.
_________________
The search for better is endless. Instead simply find very good and get the job done.

[Mark]

Run power back through it and use X10

[Kenny]

If noise is a problem and RS485 doesn't work out, consider current loop.
We usually use the Burr Brown (TI) XTR115/ XTR116 chips for 4 to 20mA,
but in this case the Agilent HCPL-4100 and HCPL-4200 4 to 20mA transmitter and receiver look more appropriate. See fig. 16a of the following.
There is a transmitter and receiver at each end of the loop.
There is also optical isolation.
http://www.dieelektronikerseite.de/Datasheets/Optokoppler/HCPL4100%20(Agilent).pdf

Of course, extensive error checking would need to be done.

Mark, re X10, how good is it and how expensive?

[Hans Wedemeyer]

I assume yo mean 115 Baud and not 115,200 Baud. BTW 110 Baud is a more "normal" Baud rate.

Electrical cable is typically not twisted, so RS485 will not work very well.

As you are only doing 115 Baud then Kennys suggestion 4-20mA will work.

115 Baud is close to second harmonic of power line 60Hz, if possible use a Baud rate that is further away from 60 Hz (I assume it's in a 60Hz country) So 110 Baud gets you further away.

I like the noise measurement idea using a simple AC voltmeter, it will tell you a lot about what to expect. Remember to put a load on each end of the cable equla to the driving source impedance, then the noise measurment will make more sense.

Hans W

If you do not want to use 4-20mA drivers, at that Baud rate you could do your own version of it.

[Mark]

[jbmiller]

For the past 30+ years(ugh,getting old) I've used a simple method using what can best be described as a true single wire tri-level,interlaced,proprietory(hacker proof) system to do remote energy control. It easily goes 20Km and back at 24 Baud.yes, twenty four baud. Drop it down to 16 baud and I can go at least 45Kms. Slow but reliable.Uses +-60 volt transmitter( homemade diff opamp) and optocouplers for rec. fuctions. has ability to detect broken wires anywhere in system. Was designed for 32 remotes and could isolate bad/open copper. Always impressed Bell guys that I knew where the trouble was long before they did.
Ah the good ol' days.
Jay

[Humberto]