Terminating Shielded twisted pair

difference between the grounds.

As always it is hard to say that it always applies. However it is usually the norm to only ground one end of the shield of a twisted pair.

========================

Grounding at the *receiving end* stops induced noise from RF and other EMI sources.

Both ends is fine, when the sending device is not itself grounded.

..... Phil

Ralph Morrison's book on grounding and shielding is very useful. The more recent issues pay more attention to terminating transmission lines that the older editions.

The termination resistor terminates the transmission line. Whether you connect one end to local ground or not is a grounding and shielding question - in an industrial setting you can expect to see potential differences between connections to local grounds if they are more than a few feet apart, and setting up ground loops which drive appreciable current through your ground return or screen can be a very bad idea .

Galvanic isolatlon can become necessary in extended systems, and it isn`t cheap, but nowhere near as expensive as the consequences of a ground loop under fault conditions.

With equipment drawing many amps through the power cabling, it is hard to assure all systems will have the same ground potential. If you connect a ground connection through the shield the current flow may only be limited by the resistance of the shield allowing many amps to flow through the shield, not a good thing.

The signals in the cables will have means of protection against differences in ground levels. That's why they use differential pairs. The differential signal will be independent of the common mode signal which can vary to some extent.

=========================

( snip Googled s**te)

You trolling, POS, stinking, code scribbling IDIOT

...... Phil

Generalizations don't mean much, so no rule applies. But I default to grounding shields as often as possible.

A little ground loop current in a shield seldom causes problems, but a lot of common-mode RF or spikes sure will.

People tend to have strong opinions about this.

It is amazing how Phil can pick out how carefully I crafted my previous post to strike fear into his very heart making him lash out blindly in the only way he knows how!

You have to give the guy his due. What a psycho!

snipped-for-privacy@gmail.com wrote: ===================== >

Ugly noise followed by a huge pong.

..... Phil

So Phil is rechristening himself as Pong Allison. Fits. Like Cursitor Doom, he seems to think that it is sunlight that pours out of his backside.

Which book is that?

Grounding at one end is not unheard of. It can sometimes prevent unwanted signals being coupled into sensitive circuits such as hum in an audio amplifier.

There isn't any general rule you can apply to everything. Some twisted pairs have transformer isolation at each end such as Ethernet.

Especially with TN-C and TN-C-S mains wiring conventions, different "grounds" my be at slightly different potentials. Connecting these grounds together through a cable shield will try to reduce this potential difference by flowing some of the load neutral current (50/60 Hz + harmonics) through the cable shield, causing EMC problems to the signal wires inside the shield.

It should be noted that with high loads with several amps of neutral current, an Ampere or more of this neutral current may flow through the cable shield ("ground loop current") if both ends are grounded and in extreme cases even cause a fire hazard.

To avoid these 50/60 Hz "ground loop" currents, the cable shields are usually grounded at one end only.

If high frequency EMC protection is needed, a capacitor is connected between the open cable shield and ground. The capacitor must be quite small to avoid passing too much 50/60 Hz.

If it is not known if the remote end is grounded, it may be a good idea to put a 100 ohm 2 W resistor between the open end of the shield and device ground. This will drain any static charge but still limit the "ground loop" current to a few tens of mA.

Of course long motor power cables from the VFD should also use shields to avoid EMC problems to the environment. The same rules about shield grounding also apply to power cable shields.

Ed, This book: "Electric Motor Drive Installation and Troubleshooting" Glen A. Mazur

While there are no shielding rules of thumb which apply in all cases, VFD's are a special case since, except when the large and expensive optional output filter is used, the motor is the reconstruction filter for the VFDs raw unfiltered switch output. The first few turns of the motor winding see the switching edges from the VFD, winding inductance blocking further penetration of the HF edge components while capacitance to frame ground shunts the HF edges into the ground system where it will take all available paths back to the VFD in inverse proportion to the impedance of each available path, one of them being the control cable shield if it is grounded at both ends.

I saw a nice example of this decades ago when I was sent to troubleshoot a turbogenerator control set trainer which used a VFD and motor to simulate the turbine. It was ready for training on schedule except for one little glitch, it tripped on overspeed at 6 RPM every time it was started. If the trip was bypassed through 6 RPM everything worked normally including overspeed trip but naturally the school was not accepting this and the original system designer was not able to figure it out in 2 weeks of trying. I fired the system up and observed a single

My first inclination was to disconnect the cable shield at the sensor, but it was a sealed waterproof assembly with sensor solidly clamped to frame ground, so I disconnected the shield at the instrument end and everything worked properly, the school signed off on the trainer and I was out in under an hour.

I would have liked to explain the issue to the original designer, but the customer was fond of secrecy and would not tell me who it was. His big mistake was not a sensor cable grounding error, it was VFD installation design. The installation manual for the particular VFD used specified a maximum motor cable length of 8 meters without either using the output filter or a VFD rated motor, where the first few motor windings get extra insulation to withstand the overvoltage from the superposition of incident and reflected edges at the large impedance discontinuity at the motor which occurs when the cable is electrically long compared to the switched edge rise time. The designer used a 7.5 meter cable with no shield and a separate ground cable substituting for the normal hull ground. Either a proper motor cable with shield and ground, or the output filter, or the most sensible approach - put the VFD as close as possible to the motor /and/ use a proper cable or metallic flexible conduit. Putting the VFD far from the motor is a lot like putting the inductor and cap of an SMPS on the opposite side of the PCB from the switch. I offered to fix these base problems, but the school was happy with the work-around fix.

It may depend on the VFD as to the distance. Where I worked we had many VFD units that were over 100 feet from the motor. They ranged form a horse power to over 100 HP. I don't recall any of them having a direct feed back tack sensor, but the tack sensor went back to a PLC. I was not in the design of the systems, just had to repair them. One problem we did have was the start/stop circuit. There was an interface board with some optical isolators and resistors. The resistors would get hot enough to melt out the solder. There seemed to be 2 problems with this. One was the resistors were not large enough to dissipate the heat and somtimes not high enough value to drop the voltage and reduce the current, and the other was that some wiring was so long in conduit with other wires enough voltage would be induced to over heat the resistors. None of the wires were shielded so induced voltage was always a problem at the plant.

I've been known to use 1 ohm // 100 nF between the shield and ground, with a diff amp (even for single-ended lines). That generally works very well.

Cheers

Phil Hobbs

The pulse risetime on VFDs and thus the maximum cable length without motor insulation reinforcement does vary and will be stated in the specific VFD manual, and it is often necessary to exceed that length which can be done without problem with VFD rated motors. I am pretty sure 100 feet is over the no-reinforcement length limit for any VFD you can buy, but using VFD rated motors is or should be standard practice in industrial plants - the extra insulation improves reliability at small cost and eliminates concern with cable length, but the long cables do increase current in the ground system and are best avoided when practical IMO. Generally industrial PLCs use I/O modules rated for at least 500V isolation, so common mode noise on differential signals is rarely a problem, but the problem controller I referred to was intended to be used in a steel Faraday cage with no significant noise sources present, where the single ended 5V tach signal through a short cable was not a problem, at least until everything was spread out in a classroom and subjected to VFD noise.

It is pretty much impossible to get a shielded cable between two pieces of equipment without the shield carrying ground current and/or approximating an antenna. You may have a choice between loop antenna (ground both ends) and half dipole (ground one end), or multiple small loops (ground many places), but unless you are operating in a good Faraday cage you will get current in the shield which will couple into the 'shielded' conductors and must be dealt with in come way, usually with differential signaling and good CMMR.

and when the resistor starts smoking you know there is something you need to look at ;)

Mil type connectors and things like twinax make it easy to use signal pairs and hard ground the shields. Some ethernet RJ45s have metal shells and ground the shield hard. I like that!

I suppose an ungrounded shield can be a safety hazard, one more reason why I'm right.

I will agree with that on new systems.

However if in a large plant like where I worked there are hundreds of old motors so the cost of replacing the motors is impractical.