Is there a relationship between noise error and distance for 4-20mA?

This noise error could be expressed as S/N ratio, percent from full range, dynamic range, noise density, resolution, etc.

Digital Bus standards traditionally come with "baud ratio - distance" tables. In a digital bus, the baud rate is the resolution, and in that case is often published (i.e. HART, CAN, etc.)

For example, if i have a 0.01% range error sensor, how much cable distance should be admissible?.


  1. If you are confused on how both distance and resolution can be related, and how one can estimate figures for real application, read:

    • This article at pcb.com, for voltage signals,
    • This article at hartcomm.org, for the HART protocol (Digital Bus over 4-20mA),
    • This article at ni.com for the CAN protocol.

This question was voted to be closed (XD!?). A mirror of this question, can be found here.

Thanks for your time.


closed as unclear what you're asking by uint128_t, Daniel Grillo, PeterJ, Peter Smith, Bimpelrekkie Mar 31 '16 at 17:14

Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • 1
    \$\begingroup\$ I'm sorry but your question makes no sense to me. 0.01% range error sensor, how much cable distance should be admissible Are you suggesting that an error can directly be related to a cable length without even specifying what cable is used and that it is independent of the circuits connected to that cable ? You will have to explain to what situation your question applies because like this it is unanswerable. \$\endgroup\$ – Bimpelrekkie Mar 30 '16 at 19:38
  • \$\begingroup\$ I appreciate your answer. I am not implying this is unrelated to the cable specs, "of course". The situation is quite simple. I have sensors and i wish to measure them -under some estimated- error for 4-20mA. And there is no circuits between. A sensor, a DAQ/PLC/etc., and the cable run. I need to estimate "how much" the 4-20mA circuit wil degrade the signal. I cannot design a loop for a 0.01% or better sensor, if the loop will degrade it at 10m. You can calculate ("estimate") that for digital buses. This question is for current loops. \$\endgroup\$ – Brethlosze Mar 30 '16 at 20:02
  • \$\begingroup\$ This is analog, not digital, so it will depend on many things. I do not mean circuits in between, I mean the circuits at begin and end of the cable. There is no "general" answer as it depends. On many things, like: How good is the sensor's current source, what is the input impedance at the receiver end. Specs of the ADC (if any). Speed, type of cable (shielded or not, series resistance). Is it a differential signal or not. What speed ? Are there any disturbing devices in the area ? You see, much much more than only cable and distance. \$\endgroup\$ – Bimpelrekkie Mar 30 '16 at 20:10
  • \$\begingroup\$ I agreed on those design variables. Check this link: digital.ni.com/public.nsf/allkb/…. That is a standard design table for CAN Bus. Agreed that all specs are relevant, but considering all the years the 4-20mA is in the industry, im truly unclear why i cannot find a similar "conservative reccomendation". Or a design guide for that... \$\endgroup\$ – Brethlosze Mar 30 '16 at 20:13

There are only a couple things you can calculate and you'll probably miss something.

If the cable resistance is too high then the compliance range of the transmitter can be exceeded, meaning it can no longer regulate. You can calculate this from the datasheets of the transmitter, the receiver(s) and wire tables and minimum power supply voltage. Remember to calculate for the round trip at minimum supply, maximum current, and maximum wire resistance (taking temperature into account).

If there is leakage in the cable it will affect the reading. 1uA of leakage represents 0.0063% FS error. Easy to calculate if you know the leakage number, but that's seldom specified.

With a reasonable length (tens of m) of clean dry cable you can usually safely assume the error will be determined by the errors in the transmitter and receiver. Noise will be more difficult to determine and will depend a lot on the details. Signal isolators can help, but often introduce a lot of error.

  • \$\begingroup\$ Really appreciate yor answer!... Now you are understanding my problem. That 0.0063% would indeed come from the cable specs -for example, capacitance per meter, just as like for the HART protocol-. And from there i can estimate some figures. But that is only a base idea i am having, unverified up to date. I am impressed why so many years having the 4-20mA standard, and lacking of these kind of reccomendations. Unfortunately most of my applications exceed several hundred of meters, and now i have an application in which the quality of the sensors (0.001% - 1e-5 errors) is critical.... \$\endgroup\$ – Brethlosze Mar 30 '16 at 20:26
  • \$\begingroup\$ There are many fine standards to pick from for a digital bus. Maybe Ethernet would work for you. \$\endgroup\$ – Spehro Pefhany Mar 30 '16 at 20:32
  • \$\begingroup\$ That is my second choice actually. But for this i would require converters for every sensor. 4-20mA is too convenient and simple for 200m cable runs, and works fine for high data rates such as 1kS/s/ch..... \$\endgroup\$ – Brethlosze Mar 30 '16 at 20:35
  • \$\begingroup\$ @hypfco: You didn't accept the answer and you said you would. \$\endgroup\$ – Transistor Sep 1 '16 at 18:48
  • \$\begingroup\$ I am sorry @Transistor. I am here for technical answers, and for solving concrete issues. Spehro gave me a great great clue, but now that clue is not enough and my concern is still strong on this communication design issue. This question is closed, so please check for alternatives above if like the discussion. \$\endgroup\$ – Brethlosze Sep 1 '16 at 19:02

It isn't simply a matter of distance. It is a matter of how much noise your current loop picks up along the way. In a very "quiet" environment, you might get away with a kilometer. But in a very noisy environment, even 1m may pick up so much ambient noise, you signal will be unusable.

The current-loop interface was designed specifically to remove line resistance from the operating equation (within the limits of the power supply). Current loop interfaces are typically used on instrumentation applications where levels CHANGE pretty slowly. So filtering can be implemented in hardware or software at the receiving end depending on the level of the signal and the level of the noise.

Discussing this as an abstract concept can take you only a limited distance. Discussion of practical cases could bet much more practical.

  • \$\begingroup\$ I positively know that voltage line sensors -with coaxial lines, such for IEPE accelerometers- are affected by other equipment EMI -such as motors-. But it is the same for 4-20mA?. Are 4-20mA also sensitive to EMI -i understand it isn't- ?. What is a noisy environment in a current loop?. I know about caring about ground loops, power supply ripples and sensor transmissor stability in 4-20mA loops -all at "DAQ" and "Sensor" side. But how could i ruin the signal from "Cable" side? \$\endgroup\$ – Brethlosze Mar 30 '16 at 21:50
  • \$\begingroup\$ It is true that a proper 4-20mA current-loop is a true "balanced line" where you can cancel common-mode signals as presumed noise. And that makes a current-loop less inherently sensitive to environmental noise. But it is a very big and wild world out there and Murphy's Law applies here. \$\endgroup\$ – Richard Crowley Mar 30 '16 at 22:01
  • \$\begingroup\$ Yeah.. i know murphy is there, and even more if you are trying to do something non-standard. I cant figure a noise scenario when 4-20mA is getting degraded. Or conversely, which is the sensor side transmitter spec guiding this degradation. If i know which should be that scenario, i could try to avoid it. Or if that is unavoidable, just move to another choice and let 4-20mA rest in peace... \$\endgroup\$ – Brethlosze Mar 30 '16 at 22:05

Not the answer you're looking for? Browse other questions tagged or ask your own question.