PDA

View Full Version : OT: Grounding DC Control systems



JoeFin
10-07-2010, 08:38 AM
I'm taking today off in protest of a situation at work and to get the ball joints replaced in my truck

I've been having HUGE problems integrating various PLC / DCS control systems at work and I believe ALL of the problems I've encountered have 1 simple root cause. The DC supplies powering the transmitters, and feedback loops is not grounded.

We have 4 Ovation DCS remote drops, each with 8 to 14 cabinets, each cabinet contains 12 to 20 I/O modules and a DC power source. The Mack 6 system controlling GE's turbines, and Allen Bradley's Control Logix controlling water treatment. Add into that mix various AC Drives with processor control and integral DC Loop power sources. ALL have an isolated (floating DC) power source. All the these I/Os, (with the exception of GE's Mach 6) are intermingled through providing feedback and control variable information via 4-20ma dc loop

I believe until the issue of "Floating DC" is addressed - every thing I throw at it is merely a "Band-Aid"

I come from this school of thought - http://www.idc-online.com/pdf/Papers/Joe_Zullo.pdf

J Tiers
10-07-2010, 08:57 AM
it can go either way.......

And it usually relates more to a question of "where" to ground as opposed to "if" to ground..... grounding usually is mandatory for shields, may not be for a 4-20 etc, as they may actually have the loop portion somewhat above ground although the circuit as a whole is referenced to ground.

The link, which I had not time to read completely, seems to relate to hazardous locations.... could include yours, with lots of hydrogen presumably around for generator cooling.

That's one reason.

Shielding and EMI is another.

The system is obviously very complex, and problems generally relate to crosstalk as much as interference in such cases. Crosstalk can be made a LOT WORSE by improperly grounding, where proper grounding fixes the lot.

The new guys must be on the job....... and not listening.... I am sure there is a consultant type who can fix it..... it's been done before.

No way we can give any clue from here:)

Evan
10-07-2010, 09:06 AM
This is the easiest answer other than chasing grounds. Now all you need to do is convince them to spend the money.

http://ixian.ca/pics7/fiberoptic.jpg

JoeFin
10-07-2010, 09:55 AM
This is the easiest answer other than chasing grounds. Now all you need to do is convince them to spend the money.


Oh they are spending the money

I ordered a slew of Allen Bradley's Isolated Analog Input and Isolated Output cards for the Contrologix system yesterday. However I still feel it is all a Bandaid until the ultimate question of bonding the DC Common is properly addressed. No matter which system's cards we are discussing, GE, Ovation, or Allen Bradley, they all eternally reference chassis ground in order to filter out noise. Already we have blown out a number of points on "Over-Voltage" not current

As outlined in the paper on grounding I come from the school of thought that you attach ALL DC commons and shield grounds to an isolated ground at the PLC/DCS cabinet. It then goes "Isolated" to the Delta Star ground system in physical earth which at that point attached to the "Common Ground Grid" servicing the entire facility.

To make matters worse the 480 volt distribution servicing the UPS system for the entire facility is on a "High Impedance Ground". I haven't yet checked all the single lines for the power distribution for the rest of the facility, but I'm sure I'll find additional conflicts there

MaxHeadRoom
10-07-2010, 11:15 AM
I am a bit apprehensive of posting in this site now as it seems that posters are liable to get Flamed unless they dumb down the answers.

But on the subject of grounding I have always used the common grounding and bonding method with all system P.S. commons connected to the common system star point connected to system earth ground.
Rather than the 'Isolate everything' method.
In all the years of installing Electronic equipment that includes PLC, CNC and PC based system it has served well.
Here is a nice online reference by Siemens that covers bonding and earthing.
http://www.automation.siemens.com/doconweb/pdf/840C_1101_E/emv_r.pdf?p=1
Max.

Evan
10-07-2010, 11:54 AM
Single point common ground is definitely the proper way. However, fiber optics for the comms will eliminate gound loops between the signal sources. Another item to look at is anything that travels by coax. That is remarkably easy to deal with if you have transients being picked up by the shielding. Just take ten feet of coax and wind it in a coil about 8 inches in diameter, tape it tight and insert that in the coax circuit. It will act as a choke for any signal on the braid from about 1 mhz and up. For other high bandwith circuits use either audio or video baluns and transmit by twisted pair. The balun eliminates DC and the TP has excellent rejection of common mode signals.

squirrel
10-07-2010, 12:48 PM
I am a bit apprehensive of posting in this site now as it seems that posters are liable to get Flamed unless they dumb down the answers.


Don't worry about getting flamed just ignore them, the ones doing it get all their knowledge from google searches and wikipedia. Those of us in the profession spot the half baked replies, its kinda like the drunk sitting on a bar stool all day and telling you how to make a million dollars.........

JoeFin
10-07-2010, 01:52 PM
Single point common ground is definitely the proper way. However, fiber optics for the comms will eliminate gound loops between the signal sources.

Evan - I couldn't agree with you more

Using an Analog Output from the DCS to transmit signal information via a 4-20ma loop to an Analog Input in the Allen Bradley Contrologix is purely the "Poor Man's Means" of communicating Data between PLCs. The kicker is they have existing "Prolink Fiber Links" between systems they are not using

I'm not the designer but merely the commisioner

garagemark
10-07-2010, 02:38 PM
We ground mainly as you prescribe in post #4. However, we do NOT ground to the plant power grounding grid. The DCS ground is it's own system, and is checked periodically for integrity.

gzig5
10-07-2010, 08:04 PM
Single point common ground is definitely the proper way. However, fiber optics for the comms will eliminate gound loops between the signal sources. Another item to look at is anything that travels by coax. That is remarkably easy to deal with if you have transients being picked up by the shielding. Just take ten feet of coax and wind it in a coil about 8 inches in diameter, tape it tight and insert that in the coax circuit. It will act as a choke for any signal on the braid from about 1 mhz and up. For other high bandwith circuits use either audio or video baluns and transmit by twisted pair. The balun eliminates DC and the TP has excellent rejection of common mode signals.

Fiber optics is not an infallable system. You have to be very careful about the electronics that are used on each end of the fiber. It is not uncommon for poor designs, especially in the recievers, to make the system as vulnerable to EMI and conducted noise as using unsheilded wire instead.

JoeFin
10-07-2010, 08:06 PM
We ground mainly as you prescribe in post #4. However, we do NOT ground to the plant power grounding grid. The DCS ground is it's own system, and is checked periodically for integrity.

Even thou Article 110 excludes Power Generating Facilities, we will attach "AFTER" the ground rod (or series of ground rods) to the rest of the ground grid making the facility on a "Single Point Ground"

Evan
10-07-2010, 08:35 PM
Fiber optics is not an infallable system.

It is guaranteed to get rid of ground loops.

The Artful Bodger
10-07-2010, 11:09 PM
It is guaranteed to get rid of ground loops.

Problems show when the devices are also 'earthed' to each other via routes other than the common point.

J Tiers
10-07-2010, 11:37 PM
Just about any transducer is "isolated", in that the sensor output is NOT in any way connected to the sensed item, and usually there is decent isolation from capacitive or inductive etc coupling.

Therefore signal wiring can exist inside of a shield without being a party to any grounding problems, at least theoretically. There are capacitive etc issues that can be a problem, even with coax. it isn't proof against interference, it's just pretty good.

Problems comes up when the shields are connected. If the transducers are set up to ground the shield locally (pressure and temperature transducers may do this, for instance, since they typically screw into piping or whatever) , then you have to deal with what to do with the shields..... Connecting them at both ends isn't usually a good option..... nor is NOT connecting them.

The transducers may have a local shield that is isolated, in which case there really should not be an issue.

Thermocouples should be connected with transmitters that isolate.... which usually solves the problem. Similar isolators are available for any type signal in most common systems, and it sounds like you are using or going to use them.

If the transmitters are shielded and floating at the pickup end, then the signals or DC power should be groundable without trouble at the other end, so long as they are intended to be grounded.

In the event that you still get trouble, then resorting to sensors that are differential, or sensors that directly output RS485, may solve the problem. RS485 can be sent easily over fiber and it usually is possible to locally power the transducers and isolators, but isn't always possible to use depending on the system.

yes, grounding everyone at the central ground at the receiver is pretty much obligatory, why would anyone not want that?

Your noise problems should be 30dB or so less of a problem than audio systems...... getting and keeping 100 dB of S/N is an issue in a complex system. Most transducers are good to tenths of a percent at best, so audio in a studio can be 50 dB better, and in a portable PA may still be 20 to 30 dB better (80-90 dB).

Studio systems use balanced circuits for a reason. Sending even 4-20ma signals, which are lowish impedance, but still unbalanced, long distances is asking for trouble in a noisy environment.

Quite a lot of power systems use a "high impedance ground", so you better just get used to that. It won't hurt you much as far as noise, a regular hard ground is not really very good at noise reduction.