0
\$\begingroup\$

I'm designing a system where we are communicating with a device (ultrasonic anemometer) which is at the end of a cable which could vary between 80-200m.

The sensor communicates via RS-485 half-duplex. There is mention of terminating bus resistors possibly being required. On the sensor, there is a 100R terminating resistor between TXD- and TXD+ which can be enabled or disabled.

Can this system be deployed without a terminating resistor and not have any problems? I read that 'long' cable lengths increase the requirement, however in the context of the guidance, long is relative.

I have used this calculator but there are some parameters which I do not know and I have made the assumption that the 'network cable' is similar to that of which is being used, such as: Bit Rate of Transmission Technique (bits per baud), UART Controller Sampling Rate (Clock cycles per baud).

My questions are these:

  • Is it standard rule of thumb for terminating resitors to match cable impedance?
  • If yes, how do I calculate cable impedance? (Example cable)
  • How do I determine if terminating resistor is required?
  • What is the tolerance of the terminating resistor for it to serve a purpose? (i.e. if the cable length varies, do I need to calculate a new resistance each time or is there +/- metres leeway)
  • Is 200m considered long in this context
  • Is 115k baud considered high-frequency in this context?

Thank you for your help in advance.

\$\endgroup\$
  • 2
    \$\begingroup\$ Do you have any reason at all to not use a resistor that is already included in the device..? \$\endgroup\$ – Wesley Lee Nov 16 '20 at 16:27
  • \$\begingroup\$ No, there's no reason. Although I thought it was significant that the manufacturer had chosen 100R and couldn't reason if that was because by and large 100R would typically match the cable which should be used with the instrument. We also have these deployed currently where the default setting is no resistor - with no apparent issues so I'd like to determine if it's working with or without the item (we've had problems with these sensors in the past but I had never considered that this may be the issue) or a separate issue completely has been identified. \$\endgroup\$ – user325107 Nov 19 '20 at 14:42
  • \$\begingroup\$ I've seen unterminated RS-485 busses that would work with one master but not with another due to the "strength" of the driver IC on the master (i.e. masters with higher rise times created issues with unterminated busses). Adding the resistor solved the issue. \$\endgroup\$ – Wesley Lee Nov 19 '20 at 16:03
4
\$\begingroup\$

All your questions seem to be about how to calculate if termination should be used or not. This is an XY problem. Simply always terminate it so there is no need to worry if it works without termination.

  1. Yes, standard transmission line rule is to terminate with cable impedance. On the other hand, some systems that use RS485 always specify to always use a certain termination regardless of cable impedance. This is simply because default infrastructure (devices and termination plugs) come with default termination value and it would be unrealistic to assume that everyone knows the cable impedance and has matching termination resistances in their pockets when installing something.

  2. Usually you buy cable that is meant for RS485 or has specified the impedance. Usually the datasheets do not contain enough parameters to calculate the impedance at the frequency you want.

  3. If you are unsure, why even bother worrying about it. Always terminate the bus on both ends so you don't have to worry if it is needed or not.

  4. Tolerance is not specified by the standard. Interfaces that use RS485 over 120 ohm cables define it as 120 ohms 5%. In fact, the termination is there so the cable length does not matter at all - when a cable is terminated at the characteristic impedance, it looks like an infinitely long transmission line.

  5. You would have to know the signal edge rate to know if 200 m is long or not. It depends on the driver slew rate. Note that you don't have to think this at all, if you use termination.

  6. 115200 baud is do-able for 200 m. But you don't have to calculate the flight time or be worried about slew rate, if you terminate.

\$\endgroup\$
3
\$\begingroup\$

You can answer a lot by putting an oscilloscope probe at each end of 200m of cable (several turns around the lab) and seeing for yourself.

My guess is : yes you'll need termination.

Or,

  • yes;
  • see cable datasheet;
  • test as above;
  • some leeway (characteristic Z is broadly independent of line length;
  • yes
  • and yes.
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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