MWC on Computers

[Mr. Tuttle]

Would it be a bad design to run a 12-3 MWC to a room for computers on 1 circuit and large printers/plotter machines (motors) on the other circuit? I'm concerned that this may cause some sort of unwanted noise on the computer circuits. Mr. T

 

[roger]

Although I am a regular user and supporter of MWBC's, in your scenario I would run individual two wire circuits. printers and plotters can cause problems on nuetrals.

Roger

 

[stickboy1375]

What actually happens that causes these problems?

 

[dkidd]

I have seen two installations over the years where the third harmonic currents caused a neutral to ground voltage of just over one volt, which caused the equipment to malfunction erratically. One was a plotter, and the manufacturer's instructions stated that the neutral to ground voltage must not exceed one volt. Pulling a dedicated neutral corrected the problems. But that was quite a while ago.

 

[kwired]

Consider the fact that the service or feeder supplying the branch circuits or the multiwire branch circuit is itself a multiwire circuit. Higher amperage design is likely but is still multiwire circuit.

 

[don_resqcapt19]

I have seen two installations over the years where the third harmonic currents caused a neutral to ground voltage of just over one volt, which caused the equipment to malfunction erratically. One was a plotter, and the manufacturer's instructions stated that the neutral to ground voltage must not exceed one volt. Pulling a dedicated neutral corrected the problems. But that was quite a while ago.

How do you know that it was harmonic current that caused the neutral to ground voltage? The normal voltage drop due to load on the grounded conductor between the main bonding jumper and the point of measurement results in a neutral to ground voltage.
It is also a very rare case where an elevated neutral to ground voltage causes problems. I think Brian posted a case where the wiring error had resulted in a neutral to ground voltage of 120 volts and the electronic equipment was working fine.

 

[gar]

120709-2127 EDT

Consider the computer by itself. Assume it has a 3 prong cord and you have intact EGC circuits. The computer chassis will be connected to the EGC. In general this means all the electronics in the computer is tied to the EGC circuit. There also may be an input filter to the computer power supply. This filter might have input capacitors between all three of the 120 V power cord wires.

Some output signal wires that might connect to other external devices may be referenced to the computer chassis (its internal common or often times called ground whether connected to ground or not). Only if the in/output signal was DC isolated (often described as galvanic isolation) would it not be referenced to the EGC.

Outside of any effects from the power supply AC input filter I don't much care what you do with the AC hot and neutral lines. The EGC is the problem.

Suppose we have a printer connected to the computer by a parallel port cable. The printer also has its signal referenced to its chassis and therefore its EGC.

Use two different EGCs from the main panel to the two different pieces of equipment and you have the possibility of a substantial difference in potential between the two pieces of equipment. The wire in the parallel cable is possibly #24 and the EGCs #12. Place a current on either EGC or both and there will be a voltage difference between the two pieces of equipment. This voltage can cause data errors or in the case of lightning or faults to the EGC large voltages that destroy equipment.

I would want both the computer and printer to have a heavy ground wire between the two chassis and only one EGC source. Or provide DC isolation in the signal path between the two pieces of equipment.

.

 

[don_resqcapt19]

...
Use two different EGCs from the main panel to the two different pieces of equipment and you have the possibility of a substantial difference in potential between the two pieces of equipment. The wire in the parallel cable is possibly #24 and the EGCs #12. Place a current on either EGC or both and there will be a voltage difference between the two pieces of equipment. This voltage can cause data errors or in the case of lightning or faults to the EGC large voltages that destroy equipment. ...

Other than fault conditions, there should not be any voltage difference between two EGCs that originate at the same panel. There may be substantial differences in the neutral to grounding conductor voltages, but there should be no voltage between the two EGCs.

 

[gar]

120709-2249 EDT

don:

Theoretically there should be no difference between EGCs except under fault conditions. But in the real world there are voltages.

I have an extension cord from my main panel to my work bench to provide a path for a TED PLC (power line carrier) signal. The bench is supplied by its own circuit. No TED units are on at the moment, but various computers and associated equipment are on in various locations. Reading between the two EGC paths is 0.7 V RMS on a Fluke 27.

.

 

[gar]

120710-0734 EDT

I need to emphasize that the fault condition on separate EGCs back to the main panel is the most serious problem because this likely causes component failures. Probably a bigger cost than data errors to a printer.

Ethernet connections likely are transformer coupled and thus DC isolation exists. Should not be a problem up to a few thousand volts.

.

 

[iwire]

120710-0734 EDT

I need to emphasize that the fault condition on separate EGCs back to the main panel is the most serious problem because this likely causes component failures.
.

Isn't that only an issue with legacy systems that bring the ground between components on the data cable such as a serial port connection?

 

[gar]

120710-0950 EDT

iwire:

USB is not normally an isolated connection.

.

 

[iwire]

120710-0950 EDT

iwire:

USB is not normally an isolated connection.

.

Thanks I did not know that.

I can't think of a typical application of a USB cable that would be tied to the building EGC on each end.

 

[gar]

120710-1033 EDT

iwire:

Consider this:

From a main or sub-panel run two different 20 A circuits to a room with a computer and a printer. The printer is on one circuit and the computer is on the other circuit. For simplicity consider a main panel with the power company transformer at the main panel.

This means neutral, EGC, and the transformer center tap are all at almost exactly the same voltage even under very heavy fault conditions.

With no faults or leakage current to the EGCs in the computer room the difference in voltage between the two EGCs is essentially zero. Assume all the wires in the two circuits are the same size. Thus, the same resistance for each wire.

On one of the circuits apply a bolted (very low resistance) fault between the hot wire and its EGC.

What is an approximate peak voltage between the two EGCs?

.

 

[don_resqcapt19]

120709-2249 EDT

don:

Theoretically there should be no difference between EGCs except under fault conditions. But in the real world there are voltages.

I have an extension cord from my main panel to my work bench to provide a path for a TED PLC (power line carrier) signal. The bench is supplied by its own circuit. No TED units are on at the moment, but various computers and associated equipment are on in various locations. Reading between the two EGC paths is 0.7 V RMS on a Fluke 27.

.

What is the source of that voltage? 0.7 volts implies a significant current flow on the EGC.

 

[gar]

120710-1324 EDT

don:

I do not know, but switching power supplies are a possibility. Also maybe my fluorescent lamp noise filters.

The Fluke 27 is fairly good up to about 30 kHz. A Simpson 260 is much better, maybe to 500 kHz. When you get to the higher frequencies the impedance is much higher than at 60 Hz. Data flow to a printer by a parallel port might be in the 100 kHz range, and a USB is probably at least 1 MHz.

Use of an Ethernet connection to a printer is better because it provides DC isolation.

When we discuss power line shorts to the EGC then it is 60 Hz, but if it originates from lightning, we are in the MHz range.

.

 

[kwired]

Isn't USB limited to fairly short distance? In most cases a computer and a USB device will be very close in proximity and the chance of being connected to same AC circuit is usually pretty high. There would likley be more noticeable problems if one were able to run hundreds of feet of USB from the computer to the other device.

 

[gar]

120712-0953 EDT

kwired:

Yes USB is a short range interface and for many devices the AC connections will be from the same circuit, and likely from the same socket strip.

The original post was relative to multiple branch circuits to the same computer room.

Also a large laser printer might be put on its own circuit even though it is a few feet from the computer.

All my printers and plotters use parallel or serial connections. One printer is 25 ft from one computer and I use an extension cord from the computer to the printer so as to share the same EGC.

A number of years ago I had a plotter with an RS232 connection to a computer with about a 50 ft spacing between them. One night with no equipment turned on and a nearby lightning strike the RS232 components at both ends were destroyed. Nothing else in the building was damaged.

.

 

[kwired]

120712-0953 EDT

kwired:

Yes USB is a short range interface and for many devices the AC connections will be from the same circuit, and likely from the same socket strip.

The original post was relative to multiple branch circuits to the same computer room.

Also a large laser printer might be put on its own circuit even though it is a few feet from the computer.

All my printers and plotters use parallel or serial connections. One printer is 25 ft from one computer and I use an extension cord from the computer to the printer so as to share the same EGC.

A number of years ago I had a plotter with an RS232 connection to a computer with about a 50 ft spacing between them. One night with no equipment turned on and a nearby lightning strike the RS232 components at both ends were destroyed. Nothing else in the building was damaged.

.

Actually the OP is about multiwire branch circuits - so you are likely looking at a single equipment grounding conductor vs multiple EGC's in the OP situation. I think the only issue with that would be if the shared neutral introduces any problems, other than the possiblilty of an open neutral problem I think harmonics on the neutral is the only other thing that needs looked into, the potential of problems (if there is any problem) with equipment grounding are the same in either case. Just my thoughts.

 

[iwire]

120710-1033 EDT

iwire:

Consider this:

From a main or sub-panel run two different 20 A circuits to a room with a computer and a printer. The printer is on one circuit and the computer is on the other circuit.

As I said, typical.

USB is usually used for short runs so the chances are low that the circuits would come from different panels / sources. But yes it could happen.


Printers are often two wire devices. But yes it could have an EGC. I have ho idea if that EGC is connected to a conductor in the incoming USB cable.

This means neutral, EGC, and the transformer center tap are all at almost exactly the same voltage even under very heavy fault conditions.

With no faults or leakage current to the EGCs in the computer room the difference in voltage between the two EGCs is essentially zero. Assume all the wires in the two circuits are the same size. Thus, the same resistance for each wire.

On one of the circuits apply a bolted (very low resistance) fault between the hot wire and its EGC.

What is an approximate peak voltage between the two EGCs?

Has this proven to be the issue that serial port connections have had with current loops on the data ground signal?