A difference makes the bit

The receiver subtracts

An RS-485 receiver has two inputs, A and B, and ignores how high either one is. It looks at the voltage of A minus the voltage of B. If the difference is positive by at least 200 millivolts, the output is one value; if it is negative by at least 200 millivolts, the output is the other.

A driver pushes the pair at least 1.5 V apart into a 54 ohm load, well clear of that threshold.

The dead band

Between plus and minus 200 millivolts the datasheet guarantees nothing. The receiver may output either level, and it has a little hysteresis (about 70 millivolts typical on a MAX485), so it tends to hold the last one. Nobody drives the pair into that band on purpose. A bus with no driver active can end up there; lesson 6 returns to it.

Which wire is A

On the chips on our modules a 1 is A above B. That is the chip makers' lettering, and the bench uses it. The TIA-485 standard and the Modbus serial specification label it the other way: for a 1, B is more positive than A. A device that follows the standard can have its A terminal where our board has B.

Idle is the same in both: a line with nothing to say rests at 1, which is called mark. The only difference is which wire sits higher to say it.

Crossing the wires

Swap A and B and the difference changes sign. Every bit is inverted, and a resting line reads as a continuous 0. Nothing is damaged. The symptom is silence or nonsense, not occasional errors.

So the first thing to try when nothing answers is to swap the two wires. Follow the datasheet of the module in front of you. The RS-485 book shows the pins on our boards.

Common mistakes

• Trusting the labels. A, B, D+ and D- are not the same across vendors. The datasheet of the receiving device decides.
• Reading the height of one rail. The receiver only subtracts. A rail's own voltage, on its own, says nothing about the bit.
• Treating a swap as a fault to repair. Crossing the pair inverts the bits and breaks nothing. Swapping it back costs seconds.