RS485 Biasing 2

[Morcego]

Hi experts,

I have the following RS485 network setup (2400bps):

Slave1 ---1.5Km----Master---300m---Slave2

I know that I must use 120 terminating resistors at
both ends. No problem, because the boards have jumpers
that can select the terminators.
But I must use bias as well (620 ohm resistor in A and B)
My quesyion is: Can I put this bias resistors in the master board (that is more acessible) or must be done also in one end?

Another question: This is a non isolated network with independent power supplies. Should I connect the 3rd wire (ground)? to take care of the common mode voltage? is is not necessary?

Thanks in advance for your help

 

[IRstuff]

If they are on different mains, then you should have a ground reference between the transceivers. Otherwise, you risk zapping the receiver inputs every time you power up the system.

TTFN

FAQ731-376

 

[itsmoked]

Yes add the ground.

Why the 620 ohm resistors??

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Morcego]

Thanks for the awnsers.

@Keith
The 620 ohm resistors are connected bettewen "A" and Vcc (one) and B and GND (the other). Because when the line is in idle mode (no transmit) is at unknow state. The 2 resistors are bias to make sure the line is at "mark" state.

I have read something about this, and the experts say that is better to add this 2 resistors. I just don't know is is better to put them at the end of the line or can be at the middle.

@IRStuff
The mains is the same. Only the power supplies are diferent. (They are industrial switching power supplies).

 

[IRstuff]

I would still recommend a ground reference.

Your comment about idle makes no sense to me. Unless you actually tristate the transmitter, its idle state is the "1" logic state. The purpose of that is a means of detecting a lost input.

TTFN

FAQ731-376

 

[IRstuff]

In fact, if you purposely bias the line, you then will be unable to detect a line break.

TTFN

FAQ731-376

 

[itsmoked]

Yes, that was my confusion. If your line is half-duplex then the master should be driving the line until it expects an immediate answer from the slaves - who are then driving the line.

But otherwise you can put the bias resistors where ever it's most convenient - probably the master.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[LionelHutz]

The power supplies are probably isolated from the mains so the 2 circuits are essentially floating unless you purposely tie them to the main safety ground. The common approaches seem to be either float the circuits and tie the commons with a wire or connect to the safety ground and hope it doesn't vary so much between locations that it messes up the communications.

 

[Noway2]

There is a voltage specifcation for RS-485, which it sounds like you have already familiarized yourself with. It states the signal swing and common mode voltage range. The signals must swing 0.2V differential to distinguish between a '1' and '0'. The common mode voltage between the transmitter and receiver may range between +12V and -7V. RS-485 is not an AC coupled, differential signal, like ethernet. It typically requires the common ground reference to ensure that the common mode voltage specifications are met.

The bias resistors are typically placed at the master, but can theoretically be placed anywhere. According to everything I have read on RS-485, your statement about holding the line in the mark state was correct; the bias resistors are not specifically for detecting a line breakage.

I was recently sent to a data center to diagnose a modbus (RS-485) communications problem between our equipment and the building management system. Debugging the receiver data showed that the first few bytes of the transmission would appear and then nothing else would be received until the line returned to a quiescent state. The problem turned out to be a floating power supply (ground) that did not create reference between the logic ground and the RS-485 'ground'. As soon as the transmission would start, the parasitic capacitance would charge up and push the lines outside of the common mode range. Adding the ground bond from the power supply to chassis, which also connected to the RS-485 ground solved the problem.

 

[Morcego]

Thanks all for the awnsers. Tomorrow I will implement this solution and post the results.

 

[Morcego]

Hi All
Terminating, biasing and add a ground reference solve completly the problems on my network. The switching power supplies are isolated, so I completly lost my reference (normally given by earth wire). Biasing was also important because the network was picking some noise.

I have learn a lot about the so called 2 wire networks, that in fact are 3 wire

Thanks for all the help

 

[itsmoked]

I have learn a lot about the so called 2 wire networks, that in fact are 3 wire

Yep! Complete BS. Just like the so-called "one wire" sensors.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[LionelHutz]

That was discussed either here or in one of the other forums. There are a lot of people who believe RS-485 is a true differential wiring sheme, hence it only requires 2 wires. It's not, it's 2 different wires that are referenced to common and the signal is transmitted as a difference in the voltages on those 2 wires.

Note 3 and 5 cover the wiring fairly well.

http://www.robustdc.com/techResources.htm

 

[John4EE]

Somewhere deep in the EIA 485 (and 422) specification is wording suggesting the DC resistance between transmitter return and receiver return not exceed some value. This value was somewhere close to 200 ohms maximum. This certainly implies your correct statements 2 wire really means 3.

Time and time again when explaining this I get the blank stare along with...

"I thought 485 is differential..."

Sigh.