Advice on 2-wire system?

Originally Posted By: Greg Fretwell
This post was automatically imported from our archived forum.



Just a little gee whiz info. IBM stopped recomending Isolated Ground receptacles almost 2 decades ago. There wasn’t enough benefit to justify the extra cost. A house wired with Romex is basically IG anyway.


This really only comes into play when you have metal conduits and building steel that they are incidentally attached to.


Originally Posted By: jpeck
This post was automatically imported from our archived forum.



Paul, Joe F.,


Thanks, it's been 30-35 years since I worked with ground faults on electronic equipment (used to work in Standards Lab calibrating voltmeters, oscilloscopes, you name it, at a defense plant.)

Worked in a 'clean room' made of gold wire screen with the fine mesh screen (openings sized to eliminate unwanted signals in the airwaves for the frequencies we were working with). That screen also served as 'ground' and removed any ground loops and noise on the ground. Very sensitive equipment being used and tested.


--
Jerry Peck
South Florida

Originally Posted By: jfarsetta
This post was automatically imported from our archived forum.



Greg,


Not to dispute you, babe, but it depends upon the equipment.

For instance, when IBM was in the telecom business (ROLM), an isolated ground was a requirement.

For its mythical 6611 high speed router, an isolated ground was required.

For its high-end computing mainframes, Russel Stole electrical connectors were required, but the data center itself had to be layed out and powered in a specific configuration.

And, with its RS6000 RISC processors, isolated grounds are recommended.

It was only after someone realized that a PC, plugged into a conventional receptacle, but electrically bonded (via its communications cable) to a high speed switch which was powered via IG receptacle) , had inadvertantly created a ground loop, was the common practice re-examined.

I can assure you it was not decades since the practice was modified.


--
Joe Farsetta

Illigitimi Non Carborundum
"Dont let the bastards grind you down..."

Originally Posted By: Greg Fretwell
This post was automatically imported from our archived forum.



Joe I am not sure where you got your information but my last 5 years of my 30 year IBM life was as an Installation Planning Rep and I can guarantee the IG recomendation was not in the IPR manual for the IBM products. ROLM was not really an IBM product so I can’t speak for them. The ROLM systems were installed and serviced by the ROLM people, not IBM.


The whole concept of eliminating “ground loops” went out the window when we started connecting up STP LAN cables. You will always have a ground loop, IG or not. Current will go down branch circuit A’s EGC, through machine A, down the cable shield and out machine B’s EGC back to the service. This really becomes a problem when they are not on the same service. Your cable becomes a bonding jumper with all sorts of undesireable current. You create a worse problem (surge damage) if you don’t use a shielded LAN cable between 2 buildings. At that point the “lightning” fix is to create another ground loop with a fat bonding conductor between A and B.


Originally Posted By: jfarsetta
This post was automatically imported from our archived forum.



Greg,


You will not always have ground lops if the equipment is not on the same physical service of fed out of the same transformer, sorry. You speak of shielded cables, which has been replaced by unshielded twisted cable (category 3, 4, 5, 5E, and now 6). Unshielded cable is the defacto standard, and has ourstripped multi-mode fiber optic cable for speed.

If switches are isolated by optical fiber cabling providing the data pathway, there is no "bonding" between the backbone devices. And, as star topologies from the switches are limited to 100 meters, the chances are good that the switch and served PCs are fed from the same panel or transformer. When you say that there will be "all sorts of ground loops", this is where a good RCDD, design engineer, and electrical contractor or engineer collaborate to develop a real plan.

As to experience, you and I have something in common. While it is true that ROLM was not originally an IBM company, it is not exactly true that only ROLM folks servicesd the equipment. I worked for ROLM at the time. While IBM is certainly a world class company, they fell down with regard to ROLM and some of the technologies ROLM employed. IBM was a mainframe and PC company at the time. IBM sold ROLM to Seimens, which combined their existing switch with ROLM's CBX technologies. BTW, Seimens, Nortel, and AT&T still call for isolated grounds.

In the case of a data center I helped care for at Unilever, we fed the mainframes, and RISC machined via motor-generator, which produced a pure square-wave. Talk about clean! Again, grounding was critical...

But, back to some points... So, are you telling me that the 6611 router didnt call for an IG? Are you telling me that the likes of Cisco, Wellfleet, Nortel, AT&T didnt call for IG?

As to shielded cables between buildings, conventional communications cabling may or may not have included shielding. Multi-conductor communications cabling typically did, but it had little to do with ground loops, and more to do with grounding, period. Analog telecom equipment could care less about it. Dumb terminals with either baluns, or coax, or RS-232 had similar limitations. Again, low speed data.

Again, with high-speed data, copper was limited to 100 meters, so there went the idea of copper inter-building data cable schemes. Optical fiber itself isolated the equipment at each end from creating a ground loop between each backbone device.

Yes, IG receptacles are still called for in many, many applications and with equipment. Single point ground "windows" are also common. BTW, are you telling me that RS6000's dont call for IGs? Or are you saying that they merely call for a single-point ground?

As you mentioned your career at IBM, you should be aware that, I too, know something of what I speak. At one point I was a working Cisco engineer, a certifieed Wellfleet engineer, a ROLM support engineer, a communications distribution design engineer, a data center design engineer, was a Sr. support engineer for Unilever, was Director of Network Planning and Suport for NJ Transit, was Northeast Regional Director of Operations for Bell Atlantic Network Integration, was a Operations Director at Exodus Communications, was a Director of Sales Engineering at Exodus, and was VP or Operations and Professional Services at Generation 360. I am a published author, and used to have a monthly column in the WebSphere Developers Journal, where I regularly wrote about power, grounding, and data center design, business continuity, and operations.

I have been around the block more than once, and have watched much of today's technology be developed and grow up.

We'll have to agree to disagree on this one. But the validity of isolated grounding methodologies remains something we CAN agree on, which is what this post morphed into. ![icon_wink.gif](upload://ssT9V5t45yjlgXqiFRXL04eXtqw.gif)

I apologize to my compadres whom I've just lulled to sleep. Sorry for the thread drift (big time)...


--
Joe Farsetta

Illigitimi Non Carborundum
"Dont let the bastards grind you down..."

Originally Posted By: Greg Fretwell
This post was automatically imported from our archived forum.



I don’t want to start a pissing match but I probably still have the RISC physical planning manual on CD. The 6611 was really nothing but a reduced price 7011 sold as a router with a crippled BIOS. It is basically a PC with a RISC processor and they got installed in industrial bays at trucking companies all over SW Florida.


A few got IGs but only if the central site specified it. IBM never did. Nor did we require IGs on any of the other RISC boxes.


Most of my IPR time was spent mitigating lightning damage and line noise problems. 99% of the time we fixed it by adding additional grounds/bonds, not eliminating ground loops.


ROLM lived in my office but we never worked on their stuff. (although we did like their tool catalog). I did get 2 ROLM guys on a Burger King rollout but that was only because they were over-manned due to losing some accounts.


Originally Posted By: jfarsetta
This post was automatically imported from our archived forum.



No pissing necessary…


I'm just saying that there's more than one way to skin a cat...


--
Joe Farsetta

Illigitimi Non Carborundum
"Dont let the bastards grind you down..."

Originally Posted By: Greg Fretwell
This post was automatically imported from our archived forum.



IG has taken on a life of it’s own but IBM decided it simply was not worth asking the customer to spend the money sometime in the 80s. They made it very clear to the IPRs that we were not supposed to say it was required by IBM.


The only thing we required is that they used a conductor for grounding, not the raceway system but we still used the EGC as the machine ground.


In the grand scheme of things RISC was far from the most sensitive machines we made. The old 370, 308x and 309x stuff really needed lots of care … but the dirty little secret was that the big TRS power supplies in the CPUs was the source of the noise we were trying to mitigate.


Originally Posted By: jpeck
This post was automatically imported from our archived forum.



Greg Fretwell wrote:
... mitigating lightning damage and line noise problems. 99% of the time we fixed it by adding additional grounds/bonds, not eliminating ground loops.


By adding new ground rods as separate grounding points, or by adding new grounds / bonds back to a common ground / bonding grid?


--
Jerry Peck
South Florida

Originally Posted By: Greg Fretwell
This post was automatically imported from our archived forum.



Most of the lightning mitigation was paralleling communication cables with a “drain” wire (IBM carefully avoided any NEC terms like bonding and grounding to avoid union and inspector problems) We used anywhere from #16ga to #6 depending on where the problem was. Joe assumes as long as you stay under the “horizontal” rule of 100 meters you are in the same building and same service but when you are wiring in contiguous industrial bays you can have a new service every 60 feet or so.


Token Ring, an STP media, does not have the 100m limitation. If it does it was regularly violated.


We went straight between the frames of connected machines with the drain wire and added ferrites to the signal cables. The thought was you slow down the transient in the signal cable with the ferrite and give it a more attractive path with the drain.


There were arguments about what may have really been happening but it did work in a place where a$% kicking thunderstorms are a daily thing and customers are not going to turn anything off.


Ground related noise problems were attacked using the standard grounding prcedures, starting at the electrode and working back to the machines. We avoided asking for anything that wasn’t a normal article 250 issue but there was an emphasis on workmanship and quality connections.


I really suspect we may have had legal problems when we were demanding extraordinary practices although it may have just been that competitors were saying they didn’t need them.


Originally Posted By: jfarsetta
This post was automatically imported from our archived forum.



Token ring went out with garters…


Old, old technology blown away by ethernet, fast-e and gigabit ethernet. Sheesh... FDDI is newer than token ring...


--
Joe Farsetta

Illigitimi Non Carborundum
"Dont let the bastards grind you down..."

Originally Posted By: Greg Fretwell
This post was automatically imported from our archived forum.



I am moving thsi to a new topic


Originally Posted By: jpeck
This post was automatically imported from our archived forum.



And I’lll let you two discuss it to death.



Jerry Peck


South Florida

Originally Posted By: Bob Badger
This post was automatically imported from our archived forum.



The new thread


http://www.nachi.org/bbsystem/viewtopic.php?t=6133


--
Bob (AKA iwire)
ECN Discussion Forums
Mike Holt Code Forum