By 20 mA current loop, I'm referring to digital current loop communications, e.g. the standard used by old Model 33 teletypes.

ASR 33 Teletype
(Image credit: By Rama & Musée Bolo - Own work, CC BY-SA 2.0 fr, https://commons.wikimedia.org/w/index.php?curid=36769003)

It was my understanding that the distance limits of a 20 mA current loop were due primarily to the resistance of the wire in the loop, and that adding voltage to the current source input could be used to drive even loops in excess of several miles. Speed was obviously limited by the teletype electromechanical mechanism to about 10 characters per second, which worked out to 110 baud (bps.)

These days, the teletype electromechanical mechanism is only a museum piece, but the advantages of current loop communications remain, though it seems they are more frequently found in industrial analog sensor circuits.

I believe many modern high-speed communication mechanisms (like LVDS) are effectively current loop as well, but without the same long-haul wire goals a teletype had.

What are the speed/distance limit trade offs if a digital current loop is implemented with semiconductors over existing telephone wires (e.g. cat 3)and allowing open-circuit voltages up to 100 volts DC?

For example, if I want 1 Mb per second, how far will current loop drive a twisted pair point-to-point given the above limits? How does this compare with RS-422 and CAN with Flexible Data-Rate (CAN FD) networks?

I'm giving the circuit editor (and my rusty electronics education) a try to illustrate what I'm thinking of. Component values are estimates, including the 151 ohms for about a mile of cable. I presume the two communicating systems have a isolated local grounds, not connected to earth ground, and that any needed opto-isolators are on the other side of the SoC chip. I'm treating the comparitors as ideal, and thus ignoring the common mode DC offet the switch might produce, I chalk it up to lack of practice laying schematics.


simulate this circuit – Schematic created using CircuitLab

Rethinking regarding the "termination resistors" across the switches, perhaps both R2 and R6 should be about 20K ohms.

I expect I should be modeling the twisted pairs as a transmission line, rather than just R7, but I've forgotten how to do that.

  • \$\begingroup\$ What are the lumped characteristics of the type of cable you refer to? What are the consequences of poor terminations? How much noise might you expect to get over a certain distance? \$\endgroup\$
    – Andy aka
    Commented May 26, 2018 at 18:50
  • \$\begingroup\$ It's my understanding that CAT 3 cable uses 100 ohm balanced twisted pairs. But as far as I know teletype style current loop used no termination resistors. However limiting rise times with the semiconductors and passives might be possible. I believe that voltage mode noise is mostly irrelevant, and that current noise is quite limited by using differential receivers. \$\endgroup\$ Commented May 26, 2018 at 19:03
  • \$\begingroup\$ If we assume that anything less than 4 mA is a current mode space, I guess we could put a 5K resistor in the circuit as a pseudo-termination during spaces, and short it out for mark. Does that sound reasonable? \$\endgroup\$ Commented May 26, 2018 at 19:17
  • \$\begingroup\$ You might get some help here: bb-elec.com/Learning-Center/All-White-Papers/Current-Loop/… I seriously doubt that you could get higher than 38.4 kbps beyond about 2 kilometers. The best I've done is 38.4 kbps at 2km and +/-20 mA using an R/S latch on the optoisolators ouptuts, but in a very highly industrial and noisy environment. \$\endgroup\$ Commented May 26, 2018 at 19:18
  • \$\begingroup\$ Thanks @JackCreasey, that was helpful. The Simplex circuit they diagram is similar to what I was thinking of, but either stations should be able to transmit by breaking the loop, see new diagram. It seems like bb-elec is more conservative than you experience, they mention 19.2 Kbaud at 2000 feet. What I don't remember is how that speed limit comes about. \$\endgroup\$ Commented May 26, 2018 at 20:32


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.