RS485 undriven line reads 0

Question

This is a follow-up to SN65176 RS485 contention .

I have solved the contention issue by waiting for at least 26us after the stop bit of a PC-connected XS201A RS485 driver before transmitting from the SN65176. However, avoiding contention is not enough.

Initially, my PIC-side termination network consisted of a single differential 100ohm terminator:

^{simulate this circuit – Schematic created using CircuitLab}

As soon as the XS201A is done transmitting, the differential voltage returns to an average value. However, this average value is not 0V differential as I would expect. Instead, the driver is strongly pulling the line halfway to the mark state. In the process, it thinks that that means a start bit and issues a "0" byte (or a corrupt partial byte from the PIC side, if it starts transmitting soon after).

I then made a bias network to mimic the "space" (opposite of mark) state, such that:

• the differential impedance is still 100ohm
• the differential voltage is biased to 500mV
• the bottom line is biased to 1.5V

^{simulate this circuit}

Adding this bias network made it such that over a few thousand bytes, my error rate is now about 0.4% (down from 100%). However, even with this improvement, there is no visible change in the idle-state voltages - the XS201A is essentially overriding my bias network and reversing the polarity of the middle resistor.

My questions are:

• Why does the XS201A not enter a high-impedance state when idle?
• Why does the XS201A have a strongly-maintained non-zero differential voltage when idle?
• Is there anything wrong with my new bias network?
• Should I not expect to be able to have a simple termination resistor instead of a bias network, especially since the XS201A claims to have a failsafe bias network built in?

Answer 1

RS-485 has a defined differential voltage at both mark and space. The idle state is when voltage V_diff is -0.2V < V_diff < 0.2V; this is done by putting the transmitter in a Hi-Z state, usually by controlling the RTS signal from the UART. RTS is then used to turn on/off the transmitter, from Hi-Z to the defined voltage level for mark/space.

Answer 2

Old question, but...

There is such a thing as failsafe biasing, where the lines are lightly held in one of only two valid states the twisted pair should be in.

If this is happening in your XS201A it could mean that it is, in fact, functioning as intended. The unexpected voltages could be explained by a hacky dc-dc conversion, like they did for the old PIC/AVR programmers. If that's a problem, I'd recommend getting a proper self powered adapter.

For a more in-depth explanation of failsafe biasing on RS485, observe this Ti application note.

Answer 3

To work around these issues, I ended up abandoning the XS201A altogether. It doesn't have proper hardware signalling, and its bias characteristics were problematic.