081129-1252 EST
Mich:
I will suggest you always want to use twisted pair. If the environment is high vibration, then use stranded, otherwise probably use solid. If the data rate is high you want low capacitance cable. CAT-5 is about 15 pfd per foot, and Belden 8723 is about 60 to 70 pfd per ft. These values are rough from memory.
If you have a high current source and low source voltage, then use larger wire. For example 20 MA at 20 V = 1000 ohms. If the termination resistor is 100 ohms, then your loop line resistance must less than 900 by a reasonable amount. I would pick 500 ohms. In this case you could have a maximum cable length of 500/(16.14*2/1000) = 15,489 ft of #22 wire. About 3 miles.
Current loop will probably include optical isolation and no reference to earth or an ECG (Equipment Grounding Conductor), or it will be powered from the 20 MA loop and be isolated from ground, or requires no power because it is simply a passive resistor and isolated switch. In these cases there is no ground path problem. It is true that to have a difference in potential between two points I have to have a loop, except in the case of a static charge or voltage across a capacitor. but it is not the loop that is causing a problem, unless it is magnetic, it is the potential difference.
In a 422/485 application the typical terminating rsistance is probably in the range of 50 to 500 ohms. We use 249 ohms because it increases signal level without a major problem from mismatch reflections. If isolation is not at one or both ends then there is a major problem from ground path noise. The typical common mode capability of 422/485 is approximately +10/-7 V .
With a typical 422/485 4000 ft is possible at 115.2 kbaud. This is possible with CAT-5 cable. With our 249 ohm termination, CAT-5 cable, 115.2 kbaud we have sent error free data 8000 ft experimentally. But in an actual application I would not go over the 4000 ft range.
With respect to RS232 I created some photos of waveforms with CAT-5 cable at 100 and 1000 ft. These are at
http://www.beta-a2.com/cat-5e_photo.html
The reason you use twisted pair wiring is to reduce magnetically induced coupling. In CAT-5 cables the twist pitch is different for each of the 4 pairs to reduce crosstalk. Electrostatic shielding reduces capacitively coupled noise. Neither of these has any effect on the difference in ground potential between the ends of the communication path. Isolation is the ideal solution here, and fiber is by far the best method.
With our I232 Isolator System, which has 2000 V peak isolation capability, we have transmitted error free data at 115.2 kbaud while the interconnect wire had a 1000 V RMS 60 Hz signal added to it. This is 1414 V peak or 2828 V peak to peak.
Specifically to your application. Define your maximum length. Is it multidrop. What is the baud rate. Is there isolation. How is power obtained at each location, source and destination. Is it 485 or I-loop? In what environment does the wire run? Is it 4 ft from an arc welder? And probably more questions depending upon the application.
jrdsg:
RS422/485 is a signal level specification and has nothing to do with any formatting of the data. If you run under marginal conditions the data structure may be important for reliable communication.
Some communication data structures may allow operating in a more noisy environment than others. For example Xmodem. TCP/IP over Ethernet or some other methods can provide very low communication error rates.
.
