Slave Device Lacks Connection to RS-485 Common Connection

H

Thread Starter

Harvard

How does a slave device that lacks access to the RS-485 COMMON manage to maintain a Receiver common mode voltage between -7 and +12 ?
 
L

Lynn August Linse

A trick question? It of course cannot, and it risks damage, although the most common symptom is 'outages', where the unit talks for a few days, then stops for some hours, then starts again. Power-cycling the device while it's in the "not talking" period often causes RS-485 to start working again, leading to lots of diagnostic confusion where people think the device is faulty.

Many vendors make such products, and if ALL RS-485 devices are from the same vendor, they often have the same power/ground design and appear to work. But mixing vendors causes problems.

In the above situation, if the slave is DC powered, likely one can assume the 0vdc/ground can be tied to the RS-485 common ... of course if your questions really is "Some idiot pulled 800 feet of direct-bury armored cable ... and it is a single pair ... where the heck is my common?" then sometimes one can use any metal shielding as a backup, perhaps with some 1K resisters in series to dampen surge noise.

Been-there, done-that for nearly 25 years now! The single-pair cable is probably for 4-20mA loops, not RS-485 too ...
 
It can't. It may work, it may not....but you're basically playing with fire. We come across many products that are designed with lack of a common reference: it's a terrible design, and no it doesn't make a difference if the product is optically isolated or not (optical isolation can mitigate some of the common/ground differentials that can occur, but does nothing in terms of ensuring that transceiver common-mode voltages remain within specifications).

Here is a link to an online webinar (from Mike Fahrion @ B&B) regarding bulletproofing/grounding industrial networks. In my opinion it is one of the best-explained sources around for this type of information, and requires less than an hour of time investment:
www.automationworld.com/how-bulletproof-serial-communications-industrial-plants

I think this webinar does an excellent job of dispelling the unfortunately very common myth that "RS-485 is a differential network, so I don't need a common/reference wire".

Darrin
 
C

Curt Wuollet

Common mode voltage can be construed as the average of the differential signals, the more apt question is voltage in relation to what. It works in many cases because the analog ground eventually connects to earth ground, but there's no guarantee. So if the analog ground isn't returned to earth on one machine and it floats out of the common mode voltage range, it quits working. That's why you should always connect a common. Even if it's not explicitly grounded it keeps the analog (reference) grounds at the same potential. Consider the case of battery operated equipment.

Regards
cww
 
Thank you for the responses and no this is not a trick question.

I am dealing with a "slave" device which has NO RS485 COMMON terminal that provides access for a 3rd wire reference to the A and B signals. I have confirmed there is no AC or DC coupling of A or B to any other likely terminal such as the 0VDC power input.

Any yes the symptoms are intermittent communication lock outs to this slave device that require warm starts or power cycles to recover.

There is only one slave and the Master does have a RS485 COMMON on the RS485 terminal block. So, it looks like that is all we have to play with.
 
C

Curt Wuollet

It's possible that the slave device is broke. That is, the common terminal is unintentionally open. There is another possibility, that the input is optoisolated, which would work regardless of common mode voltage. In that case, all you need is a failsafe termination at the master end that biases the lines between the power rails. To save research, this uses three resistors one between Vcc and + and one between + and - and one between - and ground. Resistances vary with voltage but profibus uses 390, 220, and 390. The center resistor is not the nominal 120 because the AC load would be the parallel combination, the power supply rail being a virtual ground at signal frequencies. If they are indeed isolated, whether intentionally or not, that should keep the line voltages within the common mode voltage range. Do let us know, I like solving weird problems.

Regards
cww
 
I too gave consideration to the external fail safe biasing as being a ways and means of trapping the common mode but concluded that would only happen if the slave device were transmitting. Problem with a slave is that it has to receive before it is going to transmit. It is a Catch 22 issue.
 
C

Curt Wuollet

No, it keeps the bus from drifting off when neither is transmitting. My thought is that the bus is drifting out of the range where the open collector outputs can operate. It also prevents an ambiguous state when no one is driving the bus. But in any case, it prevents common mode problems by keeping the wires referenced to the masters power supply, That and having a common between the two is about as much as you can do. And since you need to terminate both ends anyways.......................

Regards
cww
 
L

Lynn August Linse

I guess one work around would be - if you have the common wire available in the bus cable - is use a 1-watt, 1K resister between the common and the (-) or 'normally lower voltage' terminal. So you'd be adding a negative bias externally, but it should keep the RS-485 driver chips referenced to this external ground. Other values for the resister could be used: 390, 480, and so on.
 
This would still be a chicken and egg issue. If this Slave has a valid receive then it would transmit. At the time it transmits then we know the Slave RS485 A to B to COMMON would be within the COMMON MODE voltage range (-7 to +12). But there is no guarantee (from noise etc) it would remain stable to be ready to receive the next valid poll. IMO.

However, your feedback is worth the try to do something constructive, thank you.

> I guess one work around would be - if you have the common wire available in the bus cable - is use a 1-watt, 1K
> resister between the common and the (-) or 'normally lower voltage' terminal. So you'd be adding a negative bias
> externally, but it should keep the RS-485 driver chips referenced to this external ground. Other values for the
> resister could be used: 390, 480, and so on.
 
Many times 2 wire devices are referenced to frame ground and do not supply a terminal with + and -. Connecting common to remote frame can be a problem with intermittent coms and damage to equipment if you have a long run and connect each side. Use a RS485 port isolator near the device with no common terminal, in order to protect and correct for different commons.
 
It has been a long time since I have interfaced to a device with a metal frame, every thing in my world seems to be plastic, except for the wall panels some components are mounted. Other then that, for the most part the end devices are in plastic cases mounted in explosion proof housing and the likes.
 
Top