| Author |
Message |
|
|
 Posted: May 07, 2012 - 03:52 PM |
|
Joined: Oct 18, 2001
Posts: 358
Location: Eastern England.
|
|
I have a system with one master and multiple slaves which communicate over RS-485. The slaves are powered from the master via the same cable as the RS-485. It's a hybrid cable with a screened twisted-pair 22 AWG data line and 2 x 14AWG power lines. The slaves are in metallic enclosures and the total cable run can be 200m or more. A slave enclosures can be in contact with metallic building services which may have a different mains earth/ground to the master and other slaves.
The cable runs from the master to the first slaves and then loops on to further slaves. Each cable core, plus the data screen carries on all the way down the system.
Where I can't make up my mind is...
a) What to do with the slave enclosures? They don't need grounding for safety purposes but may do for EM reasons. I feel they ought to connect to the data screen to form a contiguous screen but there is a danger that large currents could flow in the data screen if adjacent slaves see different building grounds. Maybe I connect the enclosure to the data screen with an RC network to keep RF out but to prevent any circulating 50/60Hz signals from flowing.
b) Whether (and where) to interconnect the data screen and the 14AWG 0v line. My instinct is only at the master end leaving the data screen as just that, a screen, relying on the 14AWG core to act as the systems 0v reference.
Or do I connect the data screen and the 0v line together at every slave on the basis that 'lots of ground is good' and float the slave metalwork? |
|
|
| |
|
|
|
|
|
|
Posted: May 07, 2012 - 04:24 PM |
|
Joined: Nov 22, 2002
Posts: 12053
Location: Tangent, OR, USA
|
|
Been there, done that (both questions), and its a challenge.
If EMI is an issue, then I would add a cap between the power ground and the enclosure. Say 1000pf.
On the data wire screen, its even harder to give a definitive answer. Certainly connect at the master. I think that I would provide a terminal on the slave, both in and out, for the screen. But, only provide a though connection for the screen, not connected to the power ground or the enclosure.
Hope this makes sense.
Jim |
_________________
Jim Wagner
Oregon Research Electronics, Consulting Div.
Tangent, OR, USA
"The only thing standing between us and victory is defeat" P.G.Wodhouse in Wooster & Jeeves series
|
| |
|
|
|
|
|
|
Posted: May 08, 2012 - 12:17 AM |
|
Joined: Mar 28, 2001
Posts: 20387
Location: Sydney, Australia (Gum trees, Koalas and Kangaroos, No Edelweiss)
|
|
Ideally one would put say a 100R from the main data ground to all the slave's ground, in this case it cannot be done as the data ground is also the power ground. (fed from same supply in the master)
If using shielded CAT5 then I would connect the shield at the master's ground and loop through all the slaves as shield only ie no connection to the power ground or the case. (edit ok Jim said that already  )
Can the slave cases be wired to mains earth at the remote locations? |
_________________
John Samperi
Ampertronics Pty. Ltd.
www.ampertronics.com.au
* Electronic Design * Custom Products * Contract Assembly
|
| |
|
|
|
|
|
|
Posted: May 08, 2012 - 04:04 AM |
|
Joined: Feb 19, 2010
Posts: 507
Location: Montreal, QC, CA
|
|
| The shield should be connected at both ends to the enclosures, but only at one end (the master) to ground would be my guess. Wouldn't connecting ground at both ends be a big fat current loop? |
|
|
| |
|
|
|
|
|
|
Posted: May 08, 2012 - 05:59 AM |
|
Joined: Mar 28, 2001
Posts: 20387
Location: Sydney, Australia (Gum trees, Koalas and Kangaroos, No Edelweiss)
|
|
|
Quote:
Wouldn't connecting ground at both ends
I don't think anyone is suggesting that??  Jim and myself are saying to connect the cables screen at one end, the master, and keep the screen connected all along the line.
The enclosure can just go to mains earth, if possible, and nothing else. |
_________________
John Samperi
Ampertronics Pty. Ltd.
www.ampertronics.com.au
* Electronic Design * Custom Products * Contract Assembly
|
| |
|
|
|
|
|
|
Posted: May 08, 2012 - 12:39 PM |
|
Joined: Oct 18, 2001
Posts: 358
Location: Eastern England.
|
|
Thanks for the thoughts guys.
This is one of those occasions where theory, supported by websites, books and application notes, doesn't quite work in the real world.
js wrote:
Can the slave cases be wired to mains earth at the remote locations?
Not reliably, there may or may not be a mains earth nearby. Likewise I can't guarantee that a connection won't be made.
It's interesting how many resources make mention of a 100R 1/2W resistor between a slave's local mains earth and the RS-485 0v reference conductor. A 'solution' guaranteed to fail if more than 15v exists between the earths of two slaves. |
|
|
| |
|
|
|
|
|
|
Posted: May 08, 2012 - 04:07 PM |
|
Joined: Dec 01, 2003
Posts: 2502
|
|
| FWIW, when using RS485, the Modbus spec requires that the two data wires *and* signal ground (which could include the shield) be connected at all the nodes. There could be ground loop current flowing in the ground/shield if the "grounds" differ in potential, but RS485 is "balanced" so it shouldn't matter. My observations (so far) in the field with industrial installations is that differences in ground potentials seem to be well under 100 mv. I'm curious what others have observed about ground potentials. |
_________________
Tom Pappano
Tulsa, Oklahoma
|
| |
|
|
|
|
|
|
Posted: May 08, 2012 - 04:33 PM |
|
Joined: Oct 18, 2001
Posts: 358
Location: Eastern England.
|
|
|
tpappano wrote:
I'm curious what others have observed about ground potentials.
20 or so volts with enough current to draw a spark!
This was in central London in a building complex which was originally 3 separate buildings and hence had three separate mains feeds from three separate substations. Add in two large generators and a number of large UPS systems and you'll appreciate that we had some 'interesting' interconnect issues. |
|
|
| |
|
|
|
|
|
|
Posted: May 08, 2012 - 05:49 PM |
|
Joined: Jul 19, 2011
Posts: 460
|
|
I concur with John & Jim that the screen (shield as we call it here in the USA) should be grounded only at the Master and passed thru the Slaves without connecting it to the Slave enclosures. I am assuming your cable's screen is surrounded by an insulating jacket, even though you didn't explicity state that.
But that still leaves open a question about the powering and grounding of the Slaves. Are the Slaves entirely powered from the two #14 supply wires that run in your cable? Or, do these #14 wires just supply power to the Slaves' RS-485 receivers/transceivers and there is also a local power supply for each Slave's main circuitry? |
|
|
| |
|
|
|
|
|
|
Posted: May 08, 2012 - 07:35 PM |
|
Joined: Oct 18, 2001
Posts: 358
Location: Eastern England.
|
|
|
Chuck-Rowst wrote:
I am assuming your cable's screen is surrounded by an insulating jacket, even though you didn't explicitly state that.
Yes, it's a fully insulated cable. The data pair, plus the drain wire, have an aluminium/mylar foil shield around them and the insulated 14AWG conductors wrap around those before a PVC jacket is put over the whole lot.
Chuck-Rowst wrote:
Are the Slaves entirely powered from the two #14 supply wires that run in your cable?
Yes, all power for the slaves comes up the 14AWG conductors. |
|
|
| |
|
|
|
|
|
FAQ
Search
Register
Log in to check your private messages
Log in
Maximize
RSS
