This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

THVD1410: RS-485 Maximum Cable Length

Part Number: THVD1410
Other Parts Discussed in Thread: SN65HVD78

I am trying to understand where the 4000 foot (1200 m) maximum cable length comes from in the RS-485 standard. I have looked at various application notes and there is only reference to the maximum cable length. From The RS-485 Design Guide (SLLA272C), I see that the minimum required driver differential voltage is 1.5 V and the minimum receiver input differential voltage is 200 mV. In AN-1057, section 5, it states that the maximum cable length is the result of the voltage divider that the cable's DC loop resistance and the termination resistance create. I have analyzed this for a 4000 foot cable with 15 mOhm / foot resistance and the resulting differential voltage at the receiver is 750 mV, not 200 mV. 

Am I missing something or is this limitation imposed for margin?

  • Jim,

    If data rate is low, the major limitation on the communication distance is DC loss, like how you calculated. If data rate is high (in Mbps), the main loss comes from  high frequency. Please check out the 'Cable Length vs Data Rate' plot (Figure 21 page 18) of the datasheet of SN65HVD78.

    Back to DC loss evaluation, your calculation is correct. The standard requires minimum 1.5V output and maximum 200m input, therefore anything in between is the margin for cable loss. If 750mV is the amplitude you get at the receiver input, 550mV (750mV-200mV) is the room for noise.



  • Hi Hao,

    Let me ask some slightly different questions and then add in some detail about our application.

    My understanding is that the ±200 mV differential voltage (at the receiver) is the most critical aspect to recovering the data transmitted from the driver. Is this accurate? If so, and I can calculate and simulate that my differential receiver voltage never drops below ±600 mV after reflections are taken into account for any length of cable (either unterminated or matched termination), why does it matter how long the cable is with respect to the data rate? Is there a concern with non-monotonic signals even if they are always above the ±200 mV differential voltage threshold? My understanding is that the receiver is an analog comparator input with a digital output (RO) based on the differential voltage at the comparator input, thus monotonicity should not matter as long as the voltages are above some threshold (plus hysteresis), correct?

    As for the application, we're targeting a max cable length of 500 feet. It will have a master node and 1, 2 or 3 endpoint nodes (never more than 3). The target max data rate is 500 kbps. The goal is to have all endpoint nodes use the same termination (either unterminated or some value of termination) so as to minimize installation complexity. The master will have parallel termination and potentially external failsafe termination (depending on the minimum differential threshold of the receiver).

    The thought is that since we do not have a fully loaded bus (likely less than 3 UL since the transceivers we will choose will be fractional ULs) and that our maximum common mode voltage is -4 V, we won't have any issues (even with reflections) meeting the ±200 mV minimum differential voltage with margin.

    Am I missing something?

  • Jim,

    Your understanding about the receiver threshold is absolutely correct. If you can make the bus voltage always higher than 600mV, the receiver function should work. The reason that the cable length matter is 1) DC wise, the cable resistance increases with longer cable. This cable resistance and the termination will divide the driver signal to make the receiver input attenuated. 2) AC wise, the longer cable has more attenuation at high frequency, which generates more Inter-symbol Interference (ISI) jitter. Some lab data and discussion can be found in this app note.

    About termination, RS-485 network works best with a main bus (daisy chain) with two terminations at the farthest ends. I'm afraid if you want to keep the same structure for all nodes, you may have to leave it open or use a large value.