ISOLATED GROUND circuits for computer

ISOLATED GROUND circuits for computer
This for the Gurus of Grounding…
We are doing a remodel and the owner wants four isolated ground circuits in his office.
Does that mean four home runs for four outlets or can one daisy chain the grounds between the four boxes (box to box, recpt. to recpt).?
Cries out outrage? I suspect this would negate any preceived advantages, but it’s worth asking.
Four more home runs…three AFCI runs, four GFCI, etc. Better get that 40 space sub panel.
The ToolBear
“Never met a man who couldn’t teach me something.” Anon.
Replies
Fur getaboutit.
First if this is not off a sub-panel is does not buy you anything.
The idea is that in an heavy industrial location you migh have noice on the ground as it goes through several levels of sub-panels.
But from what I have read the the use of isolated ground to try and fix noise problems from other equipment have not worked. If there is that much noise it is everywhere. And that needs to be fixed.
Slight different problem, but I work the software to control damping system on ofset printing presses. It consists of a master controller and individual controllers at each printing tower.
Whenever the presses would run the communications would go down. Each indiviual unit worked ok, but the master could not communicate with any of them.
They checked and rechecked the ground. I think that they even ran isolated grounds for the electronics and made sure everything else was bonded together.
Still the same problem.
So I was sent out to look at it.
The com was twisted pair differnetial signalling. At one of the terminals a few strands of the standed wire did not get under the screw. The presses running vibrated the wire enough that it would short to the case.
.
A-holes. Hey every group has to have one. And I have been elected to be the one. I should make that my tagline.
Fur getaboutit.
First if this is not off a sub-panel is does not buy you anything.
The idea is that in an heavy industrial location you migh have noice on the ground as it goes through several levels of sub-panels.
These would land on a subpanel in this sector of the house. What methods do you use to identify noise on a system?
(Not that I could on this system at this point. Most of the old circuits are cut off and more go tomorrow as a french door is going through several runs from the panel.)
The ToolBear
"Never met a man who couldn't teach me something." Anon.
These would land on a subpanel in this sector of the house.
Whups, there goes your isolated ground.
What methods do you use to identify noise on a system?
Today, they've probably got "Noise Meters" just for that, but back in the day, one would use an oscilloscope across the outlet. Any noise over about 3% of the supply voltage of the equipment being effected would be suspect. In the case of todays computors, that would be 10% of 5 volts or .5 volts in the frequencies used inside the computor.
Most computors are relatively proof against supply line noise simply because of the huge capacitors in the power supply bridging from the output voltage leads to ground. This puts any noise to equal potential on the supplied voltages and ground inside the case.
Open the case on one computor with another running beside it and all bets are off.SamT
Anyone who doesn't take truth seriously in small matters cannot be trusted in large ones either. [Einstein] Tks, BossHogg.
If you are on T&M and worried about noise install some power conditioners.http://www.solaheviduty.com/products/powerconditioning/Index.htmUnless he is runnin something like a punch press in the basement or maybe a plasma cutter in the garage he is not going to have any noise the bothers computers.But if you are going to put in orange receptacles then I would run the isolated ground wire daisy chained through the 4 boxes. The best thing is to get all of the equipment at the same ground reference..
.
A-holes. Hey every group has to have one. And I have been elected to be the one. I should make that my tagline.
But if you are going to put in orange receptacles then I would run the isolated ground wire daisy chained through the 4 boxes. The best thing is to get all of the equipment at the same ground reference.
I second the emotion.SamT
Anyone who doesn't take truth seriously in small matters cannot be trusted in large ones either. [Einstein] Tks, BossHogg.
If you are on T&M and worried about noise install some power conditioners.
I don't see any heavy equipment. I think the orange outlet and cover sounded neat, so ... I keep thinking of a paraphrase of the Wizard of Oz: You don't need an isloated ground, Sam. You need an orange cover plate.
I like the thought of daisy-chaining the lot - cuts down three home runs. The panel is a old Crouse H. 100A meter-main and it needs to go. Two thirds half height breakers. No room. Very cramped wiring gutter. Saw a big one at HD, but it was a Murray. Have to find a decent 200a meter main.
According to the appraiser, this project ('50s) has had a number of electrical fires. One of the previous owners did a lot of wiring and had some interesting practices.
Can't trust anything in the walls that I didn't run. Out goes the proto-Romex. Out goes the Romex with nails in it here and there. Just found a 12/3 with butt splices sitting on a top plate today. They used one of those El Cheapo crimper/mashers and I could pull one leg apart. (FWIW - that Harbor Freight crimper does a nice job. Can't put those splices apart.)
The ToolBear
"Never met a man who couldn't teach me something." Anon.
How about Red or Turquoise receptacles.http://www.lutron.com/products/ColorsOfLutron.aspx?pid=Satin&cid=0http://www.lutron.com/products/ColorsOfLutron.aspx?pid=Matte&cid=0http://www.lutron.com/products/ColorsOfLutron.aspx?pid=Metal&cid=0
.
.
A-holes. Hey every group has to have one. And I have been elected to be the one. I should make that my tagline.
How about Red or Turquoise receptacles.
Saw a merlot there. Curiously, I had been looking in my Leviton catalog and lusting after the hospital grade toys. Bright red, custom wall plate: "Do not touch this switch!"
They have black marble in the Decora series. Or walnut. White or ivory is so dull. Blue surge supressor recpt. - I'll look into those. The ToolBear
"Never met a man who couldn't teach me something." Anon.
There is a very very slim chance that the seperate ground might help in a SOHO situation, but it means running the computor outlets to a seperate ground rod and then bonding that rod to the main ground rod.
Even more effective is to slip an RF choke over the romex as it enters the oultlet box.
This presumes, of course, that there is a large enough "noisy" load on the same transformer as the building the computors are in. Unless somebody crossed wires inside a microwave and ran its' romex thru the computor outlets themselves, that ain't gonna that much "noise" in a house.
SamT
Anyone who doesn't take truth seriously in small matters cannot be trusted in large ones either. [Einstein] Tks, BossHogg.
"that ain't gonna that much "noise" in a house."Need to qualify it, only ELECTRICAL noise.None of these will have any affect on spousal unit noise..
.
A-holes. Hey every group has to have one. And I have been elected to be the one. I should make that my tagline.
ran its' romex thru the computor outlets themselves, that ain't gonna that much "noise" in a house.
In my experience, it's primarily electrical noise caused by flourescent lights, other electronic equipment (such as computers) and ground loops. You wouldn't think it would be much, but it can be enough to cause problems for sensitive electronics such as computers, radio equipment, and X10 modules. Often the problem is that the offending equipment is generating more noise than permitted - for example, removing and reinstalling the cover of a computer is often sufficient to significantly increase the noise it radiates, both through the air (i.e. RF) and through the power line (coupled noise). All those gaskets, filters, and seals are critical to achieving the rated noise performance, and almost insignificant changes to them can have significant consequences.
For those interested in learning more about these issues, here are a couple of references:
High-Speed Digital Design: A Handbook of Black Magic, by Howard Johnson
Noise Reduction in Electronic Systems, by Henry Ott
Ott is the standard reference in the field, and is probably more readable for most folks, but if you want the detailed nitty-gritty, Johnson will give you that.
We usually use 4 guage bare copper. To reduce the likelihood of ground loops, it's best to route the ground with the cable.
I was Roaring at that line. You're worried about ground loops, but you're using bare copper. LOL
BTW, I've got 9+ years maintaining and repairing military communications equipment. RFI and excess noise is of prime concern in that field.SamT
Anyone who doesn't take truth seriously in small matters cannot be trusted in large ones either. [Einstein] Tks, BossHogg.
You're worried about ground loops, but you're using bare copper.
Why would you think insulation on a wire would affect ground loops?
Noise is caused when the ground path and the current path in a loop are different, so that inducing sources affect one path more than another. That imbalance creates the noise.
If you only have 9 years experience in the field, Ott may be a little before your time, but it's still the standard in the field. He explains it in a lot more detail that I can get into here.
woodturner9
30+ years experience designing military and commercial communication systems, include ship comm, manpack, DS and FH spread spectrum, telephony......
:-)
Sometimes real world experience has value. Not that I'm claiming any - I'm just another design engineer.
But the problem with the bare wire makes sense to me anyway. To avoid ground loops, make sure that the wire is connected to ground in one and only one place - which is a lot easier to do with an insulated wire.
Yep, that would be my take -- the ground for an isolated ground outlet should be insulated all the way to the panel.
So convenient a thing it is to be a reasonable Creature, since it enables one to find or make a Reason for everything one has a mind to do. --Benjamin Franklin
But the problem with the bare wire makes sense to me anyway. To avoid ground loops, make sure that the wire is connected to ground in one and only one place - which is a lot easier to do with an insulated wire.
Thanks for your comments.
Make sense, and probably more of an issue in buildings with metal studs or other sources of metal that could short out. My residential experience is almost exclusively with conventional wood construction, so that is less of an issue. Still, there are water pipes, stray nails, and other sources of metal, so insulated wire or zipper tube probably would be a better choice.
Certainly the bare ground in standard romex (or plastic conduit) will count as "insulated" from this perspective (not code/standards-wise, but in terms of effectiveness), PROVIDED that it's not allowed to contact other metal items (including other ground wires) along the way.
So convenient a thing it is to be a reasonable Creature, since it enables one to find or make a Reason for everything one has a mind to do. --Benjamin Franklin
Certainly the bare ground in standard romex (or plastic conduit) will count as "insulated" from this perspective (not code/standards-wise, but in terms of effectiveness), PROVIDED that it's not allowed to contact other metal items (including other ground wires) along the way.
Agreed. The only issue would be if you want or need the lower inductance of the larger wire - but I can't imagine that would be an issue for residential applications.
Others have covered my concerns.
Almost thirty years since I got out of the field of correcting electrical engineers theoretically incomplete work.
Such incomplete theorys as not including the possiblity that a bare wire may contact another conductor causing its' reactance/length to be different than the parallel conductor it is supposed to be matching.SamT
Anyone who doesn't take truth seriously in small matters cannot be trusted in large ones either. [Einstein] Tks, BossHogg.
Almost thirty years since I got out of the field of correcting electrical engineers theoretically incomplete work.
Ah yes, the typical technician response - "I don't understand, so the engineer must be wrong".
There is a reason technicians are technicians, and engineers are engineers. Techs usually have more "hands on" experience implementing circuits, and can be invaluable in solving issues related to implementations, such as manufacturing issues. Unfortunately, they don't usually have the theoretical understanding or training, even if they are EETs.
Engineers, on the other hand, have the training to understand and apply the theory.
Put another way, engineers understand the "why", techs work on the "what".
But many techs will always believe they are "correcting" the engineer's "errors". Most engineers have had to deal with those "corrections" and fix them. I suppose it's just the way of things, and not likely to change.
Even more effective is to slip an RF choke over the romex as it enters the oultlet box.
Any particular type of choke? Off to Radio Shack for one?
This presumes, of course, that there is a large enough "noisy" load on the same transformer as the building the computors are in. Unless somebody crossed wires inside a microwave and ran its' romex thru the computor outlets themselves, that ain't gonna that much "noise" in a house.
Then those isolated ground circuits should work just fine. <g> A bit of witchdoctoring never hurts.
The ToolBear
"Never met a man who couldn't teach me something." Anon.
Any particular type of choke? Off to Radio Shack for one?
What I'ld do. Actually, since I dismantle computors, I'ld just take 'em out of the junk pile. Every computor power supply I've taken apart has one.
A bit of witchdoctoring never hurts.
Whaddid I say.
It will, in fact, knock 3 to 6 decibels off any noise that might be there.
Add a 0.47 picofarad capacitor across the white and black terminals of the outlet and take another 3 -6 DB off off of any high-freq noise. Radio Shack :)
None of this "noise reduction" will prevent damage from a voltage spike; they last more than a millisecond! Only a surge protector can do that. Make sure your client understands this before he sues you for a new computor system.
SamT
Anyone who doesn't take truth seriously in small matters cannot be trusted in large ones either. [Einstein] Tks, BossHogg.
Add a 0.47 picofarad capacitor across the white and black terminals of the outlet and take another 3 -6 DB off off of any high-freq noise. Radio Shack :)
None of this "noise reduction" will prevent damage from a voltage spike; they last more than a millisecond!
Just sticking in any old capacitor or inductor is NOT safe design practice. At least make sure the voltage and current ratings are adequate. You won't find a suitable inductor in a PC.
An inductor (i.e. your "choke") CAN reduce the magnitude of a spike - an inductor "slows down" current changes, so the voltage spike can be attenuated by the inductor because the current can't surge (V=IR). However, it has to be properly sized for that application to be effective.
A surge supressor in it's simplest form is simply a capacitor between hot and ground and neutral to ground. A slightly better supressor will use a Metal Oxide Varistor (MOV) - because the response time is better than a capacitor. A medium to high end surge suppressor will use a pi circuit - a combination of capacitors and inductors - to form a low pass filter to reduce the spike.
Edited 3/28/2007 8:50 am ET by woodturner9
At least you are a good copy editor. I left out the descriptor "ceramic" as in 0.47 pf ceramic capacitor. And in case you haven't picked up on it yet, I have been talking with ToolBear about that which would impress his client, both of us being aware that none of it is needed.
Unfortunately, they don't usually have the theoretical understanding or training, even if they are EETs
Fortunately, this one does. Or did, I literally have forgotten more theory than you been taught.
I have designed radios and radars, even designed transistors and vacuum tubes. Had to in order to graduate. Then I went into the field and saw what school left out.
There is a reason technicians are technicians, and engineers are engineers. Techs usually have more "hands on" experience implementing circuits, and can be invaluable in solving issues related to implementations, such as manufacturing issues. Unfortunately, they don't usually have the theoretical understanding or training, even if they are EETs.
Engineers, on the other hand, have the training to understand and apply the theory.
Sounds like an architect talking about builders.
But, I'll make you a bet, if I lose, I'll rejoin the Tavern and buy you a beer, if you lose, you go to the tavern and put one in the chiller for me foe the next time I visit there.
The bet is that I have more hours in electrical theory classes than you do, incuding your lab time and your CE time.
Let's see 97 weeks times 40 hours is 3880 hrs theory and design, plus 71 x 40, or 2840 hours of specific equipment class, but I'm not incuding the equipment classes in our bet. Just the 3880 theory hours. Or we could bet on final GPAs, kick out the highest and lowest and average the rest. But I warn, of the roughly 90 scores in my record, only one was below 95%.
Did I mention that I also taught some of those classes?
The Navy schools of 40 years ago are a whole 'nother story than today.SamT
Anyone who doesn't take truth seriously in small matters cannot be trusted in large ones either. [Einstein] Tks, BossHogg.
Let's see 97 weeks times 40 hours is 3880 hrs theory and design, plus 71 x 40, or 2840 hours of specific equipment class, but I'm not incuding the equipment classes in our bet. Just the 3880 theory hours. Or we could bet on final GPAs, kick out the highest and lowest and average the rest. But I warn, of the roughly 90 scores in my record, only one was below 95%.
In my experience, your level of training and experience is very unusual for technicians. Good techs such as yourself are VERY hard to find in this day and age. Just finding a new grad tech who knows how the pins are numbered on an SOIC can be a challenge. Even with a few years experience, the "kids" don't seem to "get it". It's the same issues that plague engineering (and many other professions) - the financial rewards aren't there to attract and retain those who do their job well.
Haven't thought about training hours in a long time. One could argue that contact hours are different than "training" hours, but I'll stipulate to that.
Undergrad: 245 credit hours x 15 class hours per credit = 3675 hours
Masters: 45 credit hours x 20 class hours per credit = 900 hours (plus the thesis, which is counted as 15 credit hours)
Ph.D: 45 credit hours x 20 class hours per credit = 900 hours (plus the dissertation, which is counted as 45 credit hours)
So I guess that totals out to 5475 theory hours in EE.
I'm not sure grades are a good indicator of ability - I've worked with a lot of 4.0 idiots and far more really sharp B students.
But FWIW, I did get one B as an undergrad, which dropped my GPA to 3.93. For masters and Ph.D, grades don't mean so much - everyone gets A's or they don't remain in the program.
But what the heck - I'll buy anyway.
But what the heck - I'll buy anyway
Then I guess I'll just have to buy the house a round. When I get there (|:>)
In my experience, your level of training and experience is very unusual for technicians. Good techs such as yourself are VERY hard to find in this day and age.
Yeah, I went in the Navy in the "Golden Age" of electronic discovery and had an existing background education from a HAM BIL and a pushy mother.
Just finding a new grad tech who knows how the pins are numbered on an SOIC can be a challenge. Even with a few years experience, the "kids" don't seem to "get it".
I had two techs working for me on a satcom system. One was so smart, he aced all his schools, but couldn't find a blown fuse or change it out without blowing something up. The other barely made it thru school and would never advance in grade, but he could tell you to the day two weeks in advance when something on that contraption would break and exactly what it would be.
I've known engineers that I would pay to work for and others who thought wrapping ribbon cables in foil tape and stuffing them back into (Intel BLC) computor cases was a good idea. True story.
Every field of endeavor has people who fit any possible combination of genious and idiocy in their hands and heads. God happened to have blessed me with geniosity in both.
Anybody who can get a masters and use it in practical ways has my respect.
BTW, make it a Bud, I've got real plebian tastes.SamT
Anyone who doesn't take truth seriously in small matters cannot be trusted in large ones either. [Einstein] Tks, BossHogg.
The old engineer/tech, doctor/nurse, captain/sergeant argument has been going on for millenia. Practical knowledge vs book learnin'.In my experience, though, the real argument is about arrogance. Both parties tend to arrogantly believe that the other has nothing to offer and discount their opinions. In some cases this is amplified by an "imposter complex" -- the feeling that one is living a lie -- while in other cases (especially medical and military) folks are actually trained to pretend to be sure of themselves when they aren't.IMO, it's important to show a little humility and try to find out what the other party has to offer. This seems to be an ability that skilled leaders have.
So convenient a thing it is to be a reasonable Creature, since it enables one to find or make a Reason for everything one has a mind to do. --Benjamin Franklin
It does sound like it would be helpful to learn more from the client about what they want to accomplish with the isolated grounds.
If they are primarily concerned about power line transients causing damage, there are "whole house surge supressors" that are installed at the panel to protect the whole house. There are some pretty expensive units out there, but in our testing, the $20 units that look like a large cylindrical capacitor work as well. They mount on the side of the box, and if they take a large enough hit, they absorb it, but the case ruptures and drops "sand" on the floor - which is the indication they need to be replaced. However, it's rare to get that big a transient in residential applications (I have never seen it), so normally they just work fine forever.
It does sound like it would be helpful to learn more from the client about what they want to accomplish with the isolated grounds.
It would, but he is off to London on business - and we hope to get a few facts on the ground in his absence to slow the changes. Or put the bathroom complex on casters. It's moved around enough on the plans.
Foolish moi - asked if he wanted a isloated ground for the computer. He figured one on each side of the office would work just fine.
We'll chain them together and see about running it to the panel rather than the sub.The ToolBear
"Never met a man who couldn't teach me something." Anon.
We'll chain them together and see about running it to the panel rather than the sub.
If you are going to daisy chain them, there is probably not much benefit to running them back to the panel instead of the subpanel.
The engineer in me cringes when we talk about daisy chaining them. Daisy chaining increases the inductance and more significantly, the number of connections. From a pure engineering standpoint, it's much less effective.
FWIW, that's one reason why the NEC requires the system ground to be a single, unspliced wire - because in the case of a lighting strike or other fast, transient ground fault, the impedance of a connection becomes significant.
However, that really means less effective at higher frequency. So if you are seeing 1 pS transients, maybe even 1 nS transients, it might make a difference. In practice, transients are normally 1 mS or longer - 1,000,000 times slower.
So, in practice, daisy chaining them should be fine, unless the house is subject to some very unusual conditions (like sitting in the near field of a broadcast antenna, or something like that).
So, in practice, daisy chaining them should be fine, unless the house is subject to some very unusual conditions (like sitting in the near field of a broadcast antenna, or something like that).
Nothing in the area like "Radio Noventa - Cinquenta Milliones de Potencia" (sic) - kids around there probably had three arms. No lighting the fluor. tubes without wiring them up.The ToolBear
"Never met a man who couldn't teach me something." Anon.
We are doing a remodel and the owner wants four isolated ground circuits in his office.
Does that mean four home runs for four outlets or can one daisy chain the grounds between the four boxes (box to box, recpt. to recpt).?
Yes, that's what it means - a separate, distinct ground wire from each circuit run directly back to the common ground point (usually the service panel).
It's helpful to make these large wire - you want the lower impedance to get the maximum benefit. We usually use 4 guage bare copper. To reduce the likelihood of ground loops, it's best to route the ground with the cable. We zip tie the ground wire to the NM-B every foot or so, to ensure that the ground stays physically close to the hot so we don't get ground loops (which introduce noise). Alternatively, "Zippertube" may be a more practical option if the wallboard is up and you have to pull cable. This is a "sleeve" that opens up to install the wire, then zips up to essentially for a single cable. It's also available with a good braided ground in the sleeving. Down side is that it is kind of expensive - $5 to $15 a foot (and that is WITHOUT the cable).
This will function as a safety ground (and thus meets NEC), but the primary benefit is for equipment protection and reduced electrical noise.
Edited 3/27/2007 12:08 pm ET by woodturner9
ROARSamT
Anyone who doesn't take truth seriously in small matters cannot be trusted in large ones either. [Einstein] Tks, BossHogg.
Yes, that's what it means - a separate, distinct ground wire from each circuit run directly back to the common ground point (usually the service panel).
I had best update my panel plan to show four more home runs. The ToolBear
"Never met a man who couldn't teach me something." Anon.
What does he/she mean by isolated ground? Ask the owner just what they want to accomplish and then you'll ahead of the game. I've run into more than one computer savvy owner, that dug deep into some obscure computer magazine or serious conversation with an electrical engineering friend, only to come away with completely harebrain grounding/bonding ideas.
A point with regard to computer or entertainment equipment: With any grouping of interconnected equipment it's vitally important that all equipment be connected to the SAME ground. More important than having a "clean" ground is not having ground potential between the units.
So convenient a thing it is to be a reasonable Creature, since it enables one to find or make a Reason for everything one has a mind to do. --Benjamin Franklin
What does he/she mean by isolated ground?
I think it is a style points issue. Those orange outlets are rather dashing, and it might work. I was expecting one for the office, but he got the pen and one became four.
He also wants phone drops by each can. I though that was what cell phones were for. I am looking for a large structured wiring panel for all that.
He is off to London and we want to get stuff in before it gets moved around, again.
Whatever, it's T & M.
The ToolBear
"Never met a man who couldn't teach me something." Anon.
"A little knowledge is a dangerous thing."
"Isolated ground" means separate from the ground path provided by the conduit. When you are using romex and plastic boxes, ALL of your circuits are "isolated grounds."
There have been various reasons given for IG circuits in the past; most of these ideas have failed to accomplish anything. IBM, who are mainly the ones guilty of starting this, has since backed away from the concept.
Simply put, a number of electronic things (surge suppressors are but one) use the ground wire in their operation. What matters is that there is a good, low impedance ground path back to the panel.
All the replies seem to be assuming that there is an existing problem with noisy power circuits.....not that I'm any kind of expert, but I've never heard of a little AC noise affecting a typical office computer setup. Typically the spec. is for dedicated circuits, which makes sense-you don't want some bozo blowing the breaker & loosing all your data on a shared circuit. Isolated ground seems like gold plated specification to me....
The isolated ground seems gold plated to me as well.
Just add in a whole house surge suppressor and use a decent UPS - problems solved.
A UPS is likely to be a far more satisfactory solution to the client's concerns, based upon what we all speculate are his "real" issues.