RS485 polarization + common ground

Hi

I have Exactly a Same problem I use LT485 with PIC16F877A multi Slave

Hi Jakub,

When I run RS485 hundreds of feet, I run a system ground in the cable to keep all node power supplies referenced to the same ground. I don't know if it is correct, but it works.

Here is the normal practice for default states:

Failsafe Bias Resistors

When inputs are between -200mV and +200mV, receiver output is "undefined". There are four common fault conditions that result in the undefined receiver output that can cause erroneous data:

• All transmitters in a system are in shutdown.
• The receiver is not connected to the cable.
• The cable has an open.
• The cable has a short.

Fail-safe biasing is used to keep the receiver's output in a defined state when one of these conditions occurs. The fail-safe biasing consists of a pull-up resistor on the noninverting line and a pull-down resistor on the inverting line. With proper biasing, the receiver will output a valid high when any one of the fault conditions occurs. These fail-safe bias resistors should be placed in one place only - at the receiver end of the transmission line.

Doug

Jakub

True RS485 is designed to be floating.

It is a differential system which effectively gives opposite voltage transitions. This effectively cancels noise and induced voltage.

There has been concerns when long cable runs allow the induced voltage to effective raise the cable to much higher than can safely be touched, and clamp diodes with zeners are employed in these 'voltage sponges'  to minimize the potential above ground.

There are various devices to minimize it.

this article here does a good job of explaining it.

http://www.lvr.com/rs-485_circuits.htm

This is also a good reference

http://www.analog.com/static/imported-files/application_notes/AN-960.pdf

We have some gear that uses RS485 and we have issues from time to time.

The problem is usually the driver chip and either one of the outputs fails to hold the voltage at the correct level while no data flows, or one output dies.

Luckily the manufacturer also had issues and socketed these chips.

The problems increased when we terminated at the Tx end as the voltage was effectively the same on both legs, so by the time it reached the other end it was no different to induced noise..

My suggestion is to try to simulate it using resistors in series with the line (ours handle 1000 ohms) which may help identify where the problem is.

Mark

I have a same problem but is ok

the problem is a Transmission Bd whe you use a PIC to PIC there is no problem but whene you use a PC to pic then you have a problem to Write in PIC

you need to be in same BD, use a Quartz special

Hi,

Is it possible that you have more than one device trying to transmit at the same time?

Which protocol are you using? MODBUS? or is it just a point to point communication?

Just the termitation resistor should guarantee that the differential voltage between signal A and B is 0volts, unless any device is transmiting to the line.

About GND wiring, there are cable indicated for RS485, which has 1.5pairs, 1 pair for A and B, and the 0.5 pair(just 1 cable) for the GND, so I think that if you use a common ground it should be the same along the bus.

Tying to PE, is a technique using shielded cables, to avoid electromagnetic noise, and you connect the shield to PE, (there are different ways) to avoid such a noise, and this shield must be connected to PE not to GND. So you should consider GND as a one more signal and it is not connected to PE.

Hope that help,

Miguel

Hi Yassine,

You can also have the same problem, using half-duplex RS485, because not all the dedicated chips for USB-to-serial has an RTS signal able to manage the flow control.

In some cases RTS signal changes before the transmition is ended, so it is quite posible having corrupted data recived, when the data flows from PC to a PIC. I run into this issue when using a Bluetooth to serial module and interfacing it with a DS75176

Hope that helps,

Miguel

Don't understand what you're trying to say.

RS485 is just a method to transfer data.

Ethernet is a method to transfer data, along with serial and wireless.

The software at each end needs to be able to receive and transmit the data in accordance with the specifications.

While a uP will output serial data, it doesn't meet RS232 specifications, which may be where your system is failing.

Unless the RS485 converter contains a TTL to RS232 (or USB) it's likely that the PC won't understand as it won;t be getting the -12v signal.

Mark

Well the question here is pretty simple and short. Should I polarize the bus in one place (e.g. middle), where the terminators are (on ends in 2 places) or in each device?

 

Well for me its pretty simple, that the device should be polarized in one place (maybe two). I've read somewhere that all the devices should have those resistors witch I think is wrong thinking. From intuition the pull-ups/pull-downs will be to strong. This can be counted with a parallel resistors model. I really think that when there are more devices on the network the transition will be worser quality. What are you're thoughts on that?

 

See Doug's answer with the pull-up on the non-inverting and pull-down on the inverting.

@ mcb1, even if it's differential, it's not supposed to be floating. You'll have to be sure that the inputs are at defined  (differential) levels.

The second thing I've wanted to ask is what about the device grounding. Should the devices have common ground witch goes along with the whole bus? Or should they be grounded from PE cable from the sockets? Well I've read somewhere that if there is no common ground and the bus is long the communication could not work. This is caused by big potential differential between places where the devices are.

 

You'll need to add a reference wire, to connect the reference of one system to the others'. There's only a limited amount of common mode voltage that a RS485 receiver can suppress / survive. You need to take care that both drivers' levels do net exceed the maximum common mode voltage, so they'll have to have their local GNDs connected to each other. You might use PE for this, but be sure that you understand what you're doing. Is the PE also the local GND of both systems? If you're doing anything related to human safety, be very much aware of what connections you are making, and how currents can flow through your system.

vsluiter

I stand corrected.

Our RS485 used isolated supplies, and together with a document from the DOT about the issues they had re-enforced the 'floating' belief.

This document is worth reading and explains the various parts and problems

http://ftp1.digi.com/support/documentation/90000641_a.pdf

I noted they suggest the use of a 100ohm resistor to the PE to help prevent earth loops.

Thanks

mark