RS485 grounding part 2

I've read some material (a Jan Axelson article), that further confuses me on this matter of proper RS485 grounding for the shield that surrounds our twisted pair of data.

Like I say, I have always just grounded just one end to a good EARTH ground, either my transmitter or receiver, depending on the "best" ground available at either.

The xmtr for example may be mounted closer to an electrical panel with a ground rod nearby etc....so, that would be the preference.

But now, after reading the article, it is recommended that we use the

100 ohm resistors at each end, and these tie to SIGNAL GROUND. ! ? (as in power supply ground?).

(My xmtr and rcvr both have a power supply ground that doesn't connect to earth ground).

for example....

xmtr signal ground > 100 ohm...........shield.............100 ohm < rcvr signal ground

She talks about how A and B need to referenced to ground but with long cable runs wouldn't you want to keep that doggone "antenna" isolated from the circuits?

Is she talking about the shield or another "third" wire?

Thanks very much -- I have to get this matter clear in my head.

Reply to
mkr5000
Loading thread data ...

Why? In addition to being RF silly, that sounds like a clear safety hazard.

The inner conductors are intimately coupled to the shield by capacitance. Any RF or lightning/ESD impulses that induce potential into the shield appear on the inners, too. If the shield stops short of the box, the path of ingress of all the crap on the shield is the inner conductors through the line driver/receiver ICs. If you ground the shield to the box ground, that relative potential is zero.

Not grounding the shield is like putting up a fence to keep the bears out, but not nailing the fence boards to anything.

If you can't find any 100 ohm resistors, garlic cloves or prayer beads work about as well.

John

Reply to
John Larkin

Yes John, I have always grounded the shield but to EARTH ground.

Like I say, her article says to ground the shield to SIGNAL ground, which would be circuit ground.

?

So, which is right?

Or should circuit ground be connected to earth ground?

And the second question is... I'm still fuzzy about the purpose of the

100 ohm resistors. What exactly are they isolating? Or are they acting as fuses?

THANKS.

Reply to
mkr5000

Also, I think you misunderstood me.

By saying keeping that "antenna" isolated, I meant giving it it a hard ground so it wouldn't feed the inner conductors.

Like I say, I've always grounded one end of the shield as close to "good" earth ground as possible.

Reply to
mkr5000

Cool. That way, you only have EMI/ESD/lightning problems in half of your boxes, which is better than all of them.

John

Reply to
John Larkin

I don't know. I would think she's talking about the shield, since no other wire is defined in the RS485 specification.(?)

ALso, not sure if it will help, but see Fig-10 note B as it applies to the RS485 cable.

formatting link

This is the way I learned it. That was many years ago. The TIA-485 Standard itself does not specifiy how to ground the shield.

-mpm

Reply to
mpm

Well which type of noise or transient are you trying to shield from? All those recommendations are correct - to solve different problems. However, you must first define which type of interference, current transient, voltage difference, radio wave, etc your sysetm needs to protect from.

Also not described (apparently) is necessary galvanic isolation.

Reply to
w_tom

I tried that, but the aircraft manufacturer objected to the practice.

Go figure. :)

--
Guy Macon
Reply to
Guy Macon

On Sep 21, 7:03=A0am, mkr5000 wrote:

You may want to consider reading the very useful National Semiconductor App Note 1057. This old, but still applicable, document entitled "Ten Ways to Bulletproof RS-485 Interfaces" has a sections on grounding and shielding and on galvanic isolation. In a nutshell, the bus cable must include a wire that connects the DC signal GNDs of all nodes together to control the common mode voltage between nodes to within specification range (and the range that the transceiver chips can handle). If there is going to be trouble with various nodes having their DC signal GNDs tied together then it is necessary to implement galvanic isolation within those nodes where the GND tie is problematic. (There are some systems where the galvanic isolation would be required in all nodes and others where the isolation may be needed in less than all nodes). Cable shielding is necessary in cases where there are EMI and RMI emissions and immunity requirements to meet. In general the cable shield should be applied around the RS485 bus pair wires and the common signal GND bus wiring. The system implementation should be done in a manner to prevent DC current flow in the cable shield. In most cases this means that the shield is direct connected to the equipment external metal box at one end and AC terminated to the metal box at the other end via a series RC network. Safety requirements demand that all the equipment metal boxes shall be tied to the Earth GND at each node. The currently valid web link for the National App Note is:

formatting link
The URL to this document has tended to shift around over the years :- (.

There are applications where the usage guideline for RS485 networks may actually suggest to the designer that there may be better ways to provide the interface than by using RS485. Some alternate choices may be optical fiber or a wireless RF connection. Selection of these is clearly up to the system implementer and would be made when the common mode specification cannot be assured via a GND bus wire within an RS485 cable despite the use of galvanic node isolation or when the physical environment can induce high currents into the cable shielding. The trade off of course is the higher cost of implementing these alternative interfaces.

-mkaras

Reply to
mkaras

Thanks.....great article.

John, I have a lot of respect for your opinion so you're saying a hard earth ground for both ends of the shield with no resistor, right?

If that's the case and it's a long run than why not even more than 2?

Why not ground in the middle also or every x amount of feet?

Reply to
mkr5000

The telephone system uses a twisted pair and has a similar problem. The ground for the telephone may be miles away from your own ground, so there may be large AC and DC voltage differences between them.

If you look at an old 56k dialup modem, you'll find a tiny transformer at the input. This isolates the telephone system from the computer by keeping the two grounds separate. DSL and Ethernet use a similar transformer for the same reason.

The telephone wires are normally not shielded above ground. DSL modems have a much greater bandwidth than voice, but they use the same unshielded wires. In most cases, Ethernet cables are not shielded either. If shielding is not necessary for these systems, why use it for RS485?

So if you are planning a short run on RS485, you can probably just use twisted pair. But plan on isolation transformers for longer runs, or one that goes between buildings than have their own ground system.

Shielding should not be necessary, unless you have a source of interference that is in the same frequency band as the RS485 data.

For example, if you have a lot of cables in a long duct that carry the same type of data, you can get crosstalk that might be strong enough to cause interference. One solution might be to reroute the vulnerable lines to keep them away form the source of interference.

If the interference is out of band, such as from the transmitter in a cell phone, a simple filter at the input of the RS485 receiver should solve the problem.

So in all these cases, you need to look at the situation and determine what is actually needed to make it work. One method is to compare it to other similar systems and find out what they use. If it works, why make it more complicated?

Best Regards,

Mike Monett

Reply to
Mike Monett

You are not going to get much satisfaction from shielding against nearby lightning strikes. That would take fibre.

Industrial RF garbage may be no concern. For example, does it interfere with DSL?

It could be in the ISM bands and well above the RS485 data rate. A simple low pass filter at the receiver should solve that.

So all you have left is the common mode voltages between system grounds, which can be solved with small isolation transformers.

Best Regards,

Mike Monett

Reply to
Mike Monett

Again, what type of noise or transient is being solved? When a wire interconnects two buildings, relevant earth ground is the building's only earth ground; service entrance ground that all incoming wires (including yours) must connect to before entering the building. Not just any ground. That RS-485 ground must be the same earthing used by telephone and AC electric.

Some use cables with two separate shields. Outside shield connects to each building's service entrance ground. Interior shield is for other noise that connects only at one end at receiving end chassis. Cable now addresses two type of problems with two separate shields.

Again, what are you trying to solve? Shield for noise must be terminated only at one end so that the shield does not conduct current that would only induce more noise on the signal wires. Outside shield simply connects external transients to earth where those terminated transients are made irrelevant - service entrance ground.

Adding intermediate grounds would accomplish nothing and may simply encourage more surge current to travel through the cable rather than through earth.

Again, before anyone can recommend any shielding strategy, first, define which of so many transients or noise is protected from.

Reply to
w_tom

Also, it was asked what type of noise I'm trying to reduce....

Induced crap from lightning strikes or nearby industrial rf garbage mainly.

Reply to
mkr5000

Gas tanker trucks used to drag a chain on the road to limit static buildup.

Reply to
Richard Henry

1950s Bell Telephone studies demonstrated effective protection from direct lightning strikes. They published only myths? Many somehow know protection from lightning does not exist only because they feel.

Telcos suffer, on average, 100 surges in a switching center during every thunderstorm without electronics damage. Why? Protection is installed to make lightning irrelevant - which also makes other smaller transients irrelevant.

According to Mike Monett, fiber optic cables were standard in the

1950s. Posts based on speculation - without reading professional papers, learning the science, and doing the work - is how such myths get generated.

No, "fibre" optics did not exist in the 1950s when lightning could never damage electronics including the new semiconductor switching stations.

Mike - read what Bodle and Gresh did in the late 1950s (Bell System Technical Journal) to quantify how lightning need not cause any damage. ("Lightning Surges in Paired Telephone Cable Facilities"):

Today, that protector gap is provided by a low capacitance, earthed protector also installed for free on every subscriber line where their wire connects to yours. Protection from direct lightning strikes is an old and well proven science. Still so many just *know* protection is impossible. "Fibre" optics were installed in the 1950s?

Reply to
w_tom

Content-Transfer-Encoding: 8Bit

Richard Henry wrote:

That's what I told them! But just *try* to get any major aircraft manufacturer to install a 50,000 foot long chain. They won't do it! So how am I supposed to do what mkr5000 says to do and "always ground the shield to EARTH ground?" Oh well, maybe I will have better luck with a satellite or space probe manufacturer... :)

--
Guy Macon
Reply to
Guy Macon

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.