Where's the Noise Coming From?

Greeting Phys,

Was the first cable shield grounded at both ends or one end?

Was the second cable shield grounded at both ends or one end?

If the grounding terminations were the same on both cables than you have probably guessed correctly that the shield in the first cable was not continuous.

Shield grounded both ends even though the standard now is to ground the source end only to prevent ground loops. (quite frankly, I never saw this phenomena when we grounded both ends)

I think you just proved the thoughts you had previously, in that the shield is not continuous, so when you grounded at both ends, you made no ground loop, but you insulated the cable from noise, except for a small area somewhere probably in the middle.

When you only ground at one end, which is fully correct to prevent ground loops as you correctly mentioned, you left maybe half the cable with no ground shielding......

A resistance check of the shield MAY prove a missing continity.

But it is a faulty piece of cable......

Well posted and proved.

A further thought:

If the shield is discontinuous, and the 'ground' to which it is connected at one end is not a good ground, then the section of shield connected to that end will be actively coupling noise into the signal wires.

You could test this by

1. disconnecting the cable and measuring the resistance between the grounds, and then

2. connecting the faulty cable and, with the system operating, measuring the noise on the grounds at each end.

Additionally measuring the noise on each signal line of the faulty cable at each end would also be a useful exercise.

Probably either a discontinuous shield or drain wire.

Drain is continuous, but measuring the shield is a bit tough. I will slice that cable open someday and check the shield, when I have a free moment.

Closed loop? In other words do you have an encoder on the motor going to the VFD?

If you do check THAT cable and the Encoder, especially the mechanical coupling to the motor. Seen noise on that signal make a motor vibrate so bad it sounds like a bearing is going out soon.

No encoder - Pharma industry, which we build for, has concocted the notion that ground loops occur in signalling systems when shields are grounded at both ends. We are never allowed to ground both ends. This is general industrial and I grounded both ends, although it should not matter on a 12 foot run.

A ground loop forms a simple loop antenna......not to be recomended....

Many audio people have discovered to their cost that even a really short loop can destroy music quality. Usual is a mains hum.....but it could be even a local LW or MW radio signal.

Re: Pharma industry, which we build for, has concocted the notion that ground loops occur in signalling systems when shields are grounded at both ends.

Same thing happens in the steel industry--you get ground loops (ground current flowing in the shields) when you ground the shield at both ends of the cable.

Yes, but do you have different ground potentials to create the loops? You probably do in the steel industry. I am working with a process skid most of which have instrument runs of 15 feet or less, have ss frames and cabinets, and this one is fairly typical in that the motor, which is obviously metal and grounded is close to most instrumentation. How could there be different ground potentials on a small skid?

In fairness, I believe I do know the reasoning. It was originated for situations where the instruments connect back to a plant DCS, so there could be different grounds. To avoid problems the one ended shield is applied to all to be safe. Never know when someone might decide to skip the PLC on the skid and wire from the terminals in the panel back to a DCS.

Your comment:-

Yes, but do you have different ground potentials to create the loops?

makes no sense to me....potentials themselves do not make loops!!!

Grounding is often/usually needed to STOP different potentials on sperate but connected systems....

But grounding TWICE (or more) causes a loop!!! Do you understand now? Its two seperate items.....

Connecting twice (or more) MAY actually reduce the potential differences even more between units, but a ground loop is still not to be recommended, especially where low level signal accuracy is required.

Its often better to connect seperate grounds between equipment in a so called "star" formation (using relatively thick cables) and to not connect any shields of signal cables at the "far" end.

But signal cables still need to be shielded to stop/reduce "pickup" of unwanted signals.

It's a branch of electronics/signal propogation on its own, well almost!!

Some equipment is actually designed with two seperate grounds, that may or may not be linked in one place somewhere. Usually not in my (limited) experience.

The two grounds are often called "Signal" ground and "Safety" ground. The mames say it all!!

If I can help further just ask.

In an electrical system, a ground loop usually refers to a current, generally unwanted, in a conductor connecting two points that are supposed to be at the same potential, often ground, but are actually at different potentials.

From Wikipedia.

The shield, 2 grounding points, and return through the frame work/conduit/whatever creates a loop. My point was on a self contained skid with ss frame work, panel, metal grounded motor, etc. the 2 grounding points have to have equal potential. Without a potentail difference, there is no flow and no loop.

In a case with a DCS system, the ground for it could be a few millivolts different than the ground on my skid, hence the need for single ended shielding, as the shield in my panel is connected to the shield from the DCS panel and hence only to that ground point (potential).

You are (sad to say) misunderstanding the differences. I had thought I had explained it rather well, at least two kind people gave me a few GAs......thanks folks!

You are missing a little bit of electrical understanding about loops and potential differences.

A ground connection between two units, using a correctly sized cable, will cause both of the unit's frames to be at (almost!) the same potential.

Also no loop.

Adding a second cable in one way or another will form a loop, which it could be said has its "own" life, completely separate to any possible potential differences between the two units.....

The measured voltage difference may now be almost unmeasurable, 2 identical earth cables will effectively halve the resistance between the two units.

But if I brought another cable with for example 230 volt mains at say a few amps at least, close to it or running in the same conduit, due to the transformer effect, will induce a voltage in the ground loop that is very difficult to measure with conventional methods, because its flowing in a "loop". Rather like shorting out the secondary of a transformer, how could you measure the voltage in a shorted secondary???

The easiest way to measure something is to break one connection of the "loop" and place an ammeter in the gap made and measure the current flowing in the loop, then open the other connection and measure the current flowing through just one cable on its own. Take the second reading from the first reading, which will give you the "loop" current.

Under bad circumstances, you might measure several milliamps, thats a lot......basically this is noise.....and could potentially totally overwhelm any signals being measured....

There may be better more modern better ways/equipment to measure these loop effects, but as they are NOT referenced to ground in any way, so you cannot simply measure between the loop and ground......many simply do not understand or realise that!!!

Please don't feel I am getting at you, only engineers who have been in the business a long, long time know how to get rid of such problems. They are generally very well paid.

I used to do this for computer systems in the 70's right up to 2006.

I actually found noise in a computer that was fully and correctly installed in all respects. I was looking out the window wondering what the hell was going on as the failure happened sometimes for a whole day or so and sometimes a week went by with no errors....many parts had been already replaced.

I then noticed about 5 MILES away a D.E.W. Radar system. Only when the Radar was revolving and switched on did the computer error. I had to write a short subroutine to show the errors in a proper manner. They had to (Faraday) shield the room where the computer was installed.

A MW radio, placed on the memory, when off tuned on a remote station, demonstrated the noises that the radar generated in that computer.....it was the most sensitive way to "hear" the errors.....I did not have a micro/pico volt meter......

If I hadn't looked out the window, I would never have found the errors.....there were no ground loops either at that time!!! I had already removed around 20 (from memory!) ground loops and made the system far, far less sensitive to the Radar as the loops were one by one removed, but had not completely fixed it in this particular case.....

It was neither the first or the last time that DEW Radar systems caused processors of that time problems.....as I found out many years later!!!

I had a home HiFi system (35 years ago) and found that after replacing the amp, I had 50 Hz hum when the record (Vinyl) deck was plugged in to the new amp. I found out quite quickly that the record deck had a link between the shield of the Left and right channel, in the deck. Just cutting that link cleared the loud hum completely. No more loop!! It was a BSR stereo vinyl record deck.

I have also removed loops from between (many!) computer frames in computer rooms where the electrician and the computer installer never talked to each other, a very common occurence....

Loops are a science of their own making.......as far as I am aware (could be wrong), I have never seen a course/training on screen/earth loops and how to make them properly.....so many do not understand exactly what they are or how to fix them......

I hope this helps further a little.......

Are you having fun? I sure am. This is what I like about this site. I mention a problem with a bad cable which apparently has a portion of the shielding missing, and some how we get into ground loops, which I don't have a problem with. Original problem solution verified and a lesson on a phenomena that I really don't think I will ever deal with, but is good to know about. I only mentioned it because someone else brought it up and on most Pharmaceutical installations with 4-20ma 24VDC instrumentation where there is no local PLC there are explicit directions to NEVER ground the instruments and only connect the shield to the terminals in the box that are connected to the DCS system shield. One shield from DCS to instrument and no connection of the shield to the instrument case. No ground conductor from my system to the instrument, which is electrically grounded to the plant ground when wired for power which is almost always also present (motors, heated filters) run from 480VAC through a well grounded power panel. We are instructed by the specs to do this to eliminate ground loops. Does this sound logical to you? If the grounding of both ends would be desirable, as you imply, why do we not ground the instruments? (hate to tell the spec writers this, but with all ss tubing, framing, panels, boxes, and so on, the instrument is grounded to my system without the conductor being there)

Thank you for the informative responses.

There is nothing wrong with grounding the instruments through means other than a connection through the shield of the signal cable to the instrument.

In other words, connect only one end of that shield to ground. Grounding the shield (somewhere, one place) makes the shield effective to keep electrical noise from outside the signal cable getting to the signal cable by radiating through the shield.

Now, if you get a current flow through the shield, that current flow will couple to the signal cable, carrying noise by that route.

So, grounding good, except at both ends of the shield on a shielded signal cable.

And yes, significant currents can be carried in those other ground paths. You don't want those currents to flow in the cable shield.

GA and well explained.

Re: Yes, but do you have different ground potentials to create the loops?

And: makes no sense to me....potentials themselves do not make loops!!!

It is quite common to call it a ground loop when current flows in the shield of a cable. (I was going to say that maybe that is not fully accurate terminology, but, read on...)

And, that really is a ground loop, but the other part of the loop (not the shield part) is really current flow in the earth. So, thinking about it, it is exactly a "ground loop".

Am I just rambling? Probably ;-)

It could also be a broken or high resistance joint on one of the signal conductors in the cable

Yes, that is also a possibility. I checked the individual conductors with an ohm meter and they appear to be continuous.

20 mA is a huge signal to couple into another wire: I'd check for low impedances between all the signal (and other) wires.

or a bent pin if it is a male connector on the cable (obvious when you see it), or a spread socket if it a female connector on the cable (very difficult to identify) - a spread socket on one conductor will give a high resistance joint .....

It's not clear (to me) what type of connectors you have on the shielded cable. Sometimes a bad connector can give you a lot of noise too.

Bulk cable connected to terminal strips both ends.