Terminating Shielded twisted pair

Use good old 20 mA current loops for serial communication. The optoisolators have often 2500 V isolation.

At least old VFDs had all the electronics at the midpoint of the DC bus. In a three phase six pulse rectifier without neutral connection, the DC midpoint potential varies some tens of volts several times each AC cycle around the ground potential. Current loops had no problems with this common mode voltage.

snipped-for-privacy@highlandsniptechnology.com wrote in news: snipped-for-privacy@4ax.com:

snip

Individual load line conditioning.

I had a 17kW set of line conditioners in my boss's house in Virginia.

The dang UPS's would go off whenever the Generac fired up.

And they were name brand UPS too.

Roasted input resistors are usually good evidence for rejecting a warranty claim. ;)

Cheers

Phil Hobbs

Did they put up the skyscraper on that property yet?

Cheers

Phil Hobbs

I think it's under construction now. I haven't been there lately. Street View is almost a year old.

The intersection is going to be a nightmare.

That resistor sounds quite small. Can your cable shield carry 1-10 A without overheating ? Usually a much higher resistors are recommended such as 10 or 100 ohms. The often recommended 100 ohm 2 W resistor can carry 140 mA and cause a 14 V potential difference. Such potential differences will cause problems to non-isolated RS-485, since it above the common mode range.

Regarding RF, the 100 ohm resistor will have a similar impedance as the cable shield acting as an antenna, so the resistor will dissipate (and not reflecting back) some of the RF "antenna" current.

This is shielded signal leads we're talking about, not the power supply. What's in view is breaking ground loops without trashing the EMI rejection. A 1-m diameter loop has an inductance of about 3 uH, so at

60 Hz its reactance is about a milliohm. That's on the order of the contact resistance of BNC connectors, which is why wiggling the cable will sometimes make the ground loop signal change by 10 dB or more. (We aren't usually thinking about millivolt signals with circulating currents of an amp.)

For RG-58, the shield resistance is about 0.015 ohms per metre, but better coax or electrical conduit can be much lower. In either case, the 1-ohm resistor gets rid of the 60-Hz resistive voltage drop along the coax shield, so that the magnetic pickup remains balanced and the ground loop problem more or less goes away.

The cap takes care of the RF. It isn't a complete solution, of course, but it reduces the incidence of hum loop problems by probably 90%.

Cheers

Phil Hobbs

Not if it's a 100 watt part.

I was trying to address different mains wiring conventions used in different countries or at different times.

In the US and in most parts of countries in Europe use these days the TN-S wiring convention consisting of one, two or three phase wires, a Neutral wire (N) and a separate Protective Earth (PE) ground wire. The current flowing in the phase conductor(s) as well as in the neutral wire will cause a voltage drop in the conductors. Measuring the voltages between two points of the same conductor will show up to a few voltage difference.

In normal situations, no current flows in the PE conductor and hence there is no voltage drops in the PE network. Connecting the chassis of two equipment at different locations does not cause problems with signal cable shields. Only when there is a ground fault from a phase to PE and hopefully burns the phase fuse.

Fixed pitch font;

Dev1 Dev2 Big load " cable shield " +---------------------+ " ! ! " ! ! " =======+=====================+=== PE conductor " " ================================================+ N conductor

However, if the TN-C wiring convention is used, in addition to 1-3 phase wires, there is only a single combined PEN conductor which carries the normal neutral current but also handles the ground fault current. In normal operation, the neutral current in the PEN conductor causes a voltage drop of a few volts.

Connecting the chassis of two devices are connected at different points on the PEN conductor. When measuring the potential between the chassis of the two devices also show a potential difference of a few volts.

If you connect a cable shield between the two devices, it will be in parallel with the PEN conductor and part of the neutral current will take a tour through the cable shield. This can be several amperes, unless the cable shield resistance is increased,

Fixed pitch font;

Dev1 Dev2 Big load " cable shield " +---------------------+ " ! ! " ! ! " =======+=====================+===============+ PEN conductor

The return current from "Big load" causes a voltage drop between Dev1 and Dev2 attach point and some of the PEN current will flow in the cable shield. The current depends of the ratio between PEN resistance between Dev1 and Dev2 compared to the shield resistance.

What all of this ignores is that the shield works perfectly well if only grounded at one end. So virtually no downside.

Then there is also the issue of connecting equipment that is not on the same power distribution. Even in TN-S system with a separate protective earth conductor the protective earth is grounded at the point of entry to the facility. It can also be grounded at other points. The result is there can be current flowing in the protective earth conductor as a result of the two ground points being different, even if only at times of fault.

The point is why create a conductive path that can not sustain the currents that could flow if that conductive path is not useful?

I don't build a lot of AC-powered devices. Usually wall-wart power, very occasionally a built-in AC-DC switcher. The safety ground may or may not be connected to the chassis, but never to the neutral.

But that's only relevant in some gross fault condition, where beaucoup current is flowing in the ground rather than the neutral.

Not highly relevant to lab equipment.

Cheers

Phil Hobbs

Even when your device has a three pole (L, N, PE) mains cable and plug but you plug it to a socket with TN-C fixed wiring, the socket N and PE connectors are connected together and then connected to fixed PEN wiring. Thus you are going to be ground (chassis) potential differences if you connect your two devices into different mains sockets. This is actually a TN-C-S configuration in which the TN-C is split into TN-S within each mains socket.

In other cases the split point might be in the apartment distribution panel, thus no potential differences within the apartment but problems if you connect the signal cable to a different apartment, Moving further the TN-C-S split point towards the distribution transformer and the equipotential area becomes larger, Finally we get a pure TN.S system if both N and PE are wired separately to the distribution transformer secondary mid/start point.

Sure. Ground loops are usually millivolts in labs and light industrial settings. Steel mills and ships are another matter of course.

Pretty unlikely for my stuff!

Cheers

Phil Hobbs

>

I thought there was supposed to be a ground connection at the point of the split? So every outlet has an earth ground?

Nope.

Originally in TN-C the PEN conductor was connect to some kind of ground electrode only at the star point of the distribution transformer to drain off e.g. some atmospheric charges.

Later it was discovered that if only the PEN conductor was broken, the phase voltage would penetrate trough the appliance loads to their neutral pole and form there to the metallic chassis, causing a serious shock risk from all metallic applications.

The risk of a PEN failure is quite real if open wire distribution is used. A falling tree might take out the PEN wire but not all phase wires. For this reason, each building often has a local grounding electrode connected to the PEN bar in the main distribution panel. If the distribution PEN wire fails, there is an alternate current path from the house grounding electrode to the distribution transformer grounding electrode through the hopefully wet soil.

In genera, the risk with TN-C is the risk of a PEN fault anywhere in the wiring and for this reason TN-C is often depreciated these days, but old installations are still plenty.

Even though you explain the problem with PEN use and also describe the reason for using a direct earth connection at the point where the PEN wire ends and becomes a separated neutral and earth, you say... "nope". If the PEN is carried directly to the outlet with no protective earth at that point, it is a dangerous situation. While an open PEN may be less common for wiring inside a home, there is no reason to think it can't happen.

That's why a protective ground connection is required at the end of the PEN where it is split to neutral and earth wires.

It works ok for electrostatic shielding, but won't help with RF getting induced into the wiring. Grounding the shield at both ends helps to prevent RF getting coupled into the signal pairs as common-mode interference, and prevents common mode currents in the signal wires from getting radiated also.

Of course if your electrical distribution system is such that the grounds have a large mains frequency voltage difference between them then that may caue problems. My preferred solution would be to not be putting large currents through your protective earth all the time, so that there is little current in the shield, and failing that just don't interconnect those pieces of equipment except via isolated interfaces (optical, transformer etc.). If you must connect them, then run a fat earth wire in parallel with the shield that can take whatever mains current you might see, and that can continue to do so for longer than it would take for your mains fuses to all blow in the event of a fault.

Wrapping the shielded lead around a ferrite toroid helps.

It creates an appreciable common mode inductance, and you end up with a lot less RF current flowing the shield and correspondingly lower induced RF currents in the signal wires.

Not a cure-all, but one more useful trick.

<snip>

Common overhead urban power distribution in the US has the neutral and an earthing electrode connected at the utility pole transformer. Supply to buildings has no ground wire. The neutral is connected to system ground and earthing electrode(s) at the building. (Inside the building ground is never connected to neutral again.)

Cable TV main-line has the shield earthed and thus connected to the neutral at the utility poles. Cable drops to the building have the shield connected to the building ground and earthing system (and thus neutral) at the building entrance/service. The cable shield would appear to be in parallel to neutrals between poles, and in parallel to the neutral to the building.

There should be a lot of 60 Hz noise on the cable. Why isn't that a problem?.