DC Transient voltage

[JMK 0]

Hi all - I have an issue with a magnetic reed switch flow meter application. The flow meter is connected to +24 VDC and the output is connected to a high speed counter card on a PLC. Over the past few months, operations have seen inconsistent batch levels associated with said flow meter. In the near past, it was brought to my attention that after a batch is complete the batch count keeps increasing on the HMI screen. However, the valve is closed and there isn't any oil flowing in the pipe. Well, I got the O-Scope and noticed that there was noise on the signal (return from reed sw.) wire. OK - First of all, the wires were 14 awg THHN running out to the flow meter. This past weekend, I pulled in a Belden 8770 18/3 shielded cable. I terminated the drain to ground on the PLC chassis per the AB spec. Today, operations called with the same problem. The noise was still present. Again, I got out the O-Scope. This time I looked at things a little more closely... DC (+) with respect to ground had 13 VRMS ~60-80 Hz. DC (-) with respect to ground showed the same. DC (+) with respect to DC (-) showed 5-7VRMS ~60-80 Hz. Interesting I installed a jumper from DC (-) to PE. The noise has significantly decreased. I am looking for some contributing factors, and what to look at next... There are 7 AC Drives in this panel... whoever installed the drives did run 3 phase line power and motor leads in the panel wireway with low voltage control wires... Is it possibly just the drives? The 120 VAC line power is clean (verified w/ O-Scope) Am I dealing with a failing DC power supply? What can you recommend to clean up the noise in the DC system? Is grounding the DC (-) common practice? I always assumed that the line side ground on the power supply also referenced the DC common to ground. JmK

[paulengr 44]

As a recent discussion about a year ago revealed...no, it's not. You can't assume that. In addition, one problem that seems to be notorious with AB cards is common mode problems. I usually have to tie the "common" or "-V" side of the cards to a "signal ground" as well. The biggest problem seems to be that very few of them are truly isolated and those that are, usually aren't very well isolated. This is generally a huge problem with analog signals especially but can also arise from some unusual places. I've seen noise conducted across a RIO communication cable before. After that incident in view of the nature of the signal (some sort of modified RS-485) but in violation of AB's rules, I ground the shields at one end only on each cable run (which is the SOP for shielded differential transmission lines). Don't forget the usual rules. That is, be aware that ground or "common" signals do travel across wires that have real impedances. Do whatever you can to minimize those impedances, and then make sure to break everything up with home runs back to a single common reference point (bus bar) to avoid having some of your commons "floating" on you.

[ianbuckley 0]

Paul's reply is right on. Let me answer some of your questions directly too. Yes it is possible the power supply is failing (assuming a switcher). To check this, disconnect the DC leads from the supply and measure right on the power supply terminals. Assuming the supply is good, you are picking up the noise from somewhere else. As for the 480, if the noise is truly in the realm of 60Hz, it would be coming from the 480VAC feeds into the drives, not from the outputs of the drives. The drives take the 480 feed, rectify it into a DC bus and then chop it up at a carrier frequency (usually 1.5 - 4 kHz). The noise you would see from the output of the drives would be centered around the carrier frequency (with harmonics). It is not uncommon to ground the return side of a power supply, dependent on the application, but they are almost never grounded internally. You don't say how many points are involved in the system, but you might look for connections, on the DC supply lines (+ and -) and ground wires in the field, that are loose or corroded.

[paulengr 44]

Just a thought...are you running your DC & AC lines together in close proximity for long distances (like in the same conduit)? Is this a matter of simple inductive pickup? Standard practice is to keep DC & AC lines as far apart as possible for this reason alone.

[JMK 0]

Thanks for the quick reply. The control panel is 72 X 72. The PLC is configured for 64 DC inputs, 64 DC outputs, a few analog IO, the HSC card, and a Device Net scanner. The DC control wires run in separate conduit from AC power and motor leads. However, the power and motor leads do run in the same panel wireway. The DC control power from the power supply terminate on a (+) and (-) bus. This system does live in a wash down environment, so I will have to verify corrosion on field device terminations. Since grounding the DC common, operations haven't seen any further issues. JmK

[paulengr 44]

Based on what you described, I can try to explain what's going on. Many people think that EMF is a black art. It is actually very easy to understand. It's just that it is not easy to pin down the culprit every time. There are certain things which will make it worse and certain things that make it better. Space has an impedance of 377 ohms. Any time you have a wire carrying a signal where you set up an impedance close to that magic value, you will start conducting energy into free space. This is the definition of an antenna. There are two basic kinds of antennas, dipoles and loops. A dipole is simply a single piece of wire that is not connected at one end (that end connects to "space"). It radiates energy radially away from it. Most antennas that you see are whip antennas. The second kind is a loop antenna. This is just what it sounds like...having a complete circuit. So every one of your circuits is actually a loop antenna! And all your dangling, unused wires are whip antennas. But those are usually much less of a problem in a typical industrial scenario. The physical geometry of the antenna determines it's impedance. And the impedance is frequency-dependent. So as the loop gets larger or the whip gets longer, or smaller, then the frequencies that it is most efficient at radiating will change. Antennas go both ways. So any time you have a complete circuit, it is both a transmitting and a receiving antenna. Every one of your DC circuits (positive power supply->device->negative power supply) is effectively a "loop" antenna. So are all of your AC circuits. Typically though AC circuits transmit and DC ones receive. These are absolute truths at all times. No matter how you wire things up, the fact is that you are going to be creating antennas, both transmitters and receivers. How you deal with this fact is what determines whether or not you will have problems with EMF. We can get into details about polarization, E-fields vs. M-fields, etc., but in reality these are all a lot of extra details in an INTENTIONAL antenna system. You're not dealing with intentional systems and you would spend months tearing your hair out if you tried to figure it out. That's what microwave engineers spend their entire life doing. If you're not trying to achieve maximum efficiency, there are some basic rules of thumb (guidelines) that work very well and keep you out of trouble. Like anything else, you may be able to get away with ignoring them some of the time but sooner or later, it will come back to bite you. For instance, if you run your signal wiring away from your power wiring, you will reduce the problem (radiated power decreases at the square of the distance). For the same reason if you follow the rule of crossing your power leads at "right angles", you will reduce the problem (cuts the coupling power in half). Always provide a low impedance path back to ground from all of your circuits. This rule isn't just for safety. The EMF reduction reason is because then you've formed a resistive divider. Assuming that you have somehow formed an efficient antenna (Murphy's law pretty much gaurantees that this is the case), then you will have EMF present. If all of your connections are high impedance, then the EMF will be a significant signal in your circuit. If however you have a low impedance path to ground, then simple ohm's law (R=R1/(R1+R2)) gaurantees that most of the signal is conducted harmlessly to ground and your circuits will see almost no EMF. If you are using metal conduit or duct or other raceways and you ground them, you have created a Faraday cage. This prevents signals from travelling in or out of the wiring inside. If you run grounded, shielded cabling, then you are doing the same thing. This is the principle behind coaxial wire and why it works so good for RF signals. If you abandon or change circuits, don't leave those extra wires connected to anything. All they do is create whip antennas which conduct EMF into your live circuits. When you are grounding, always ground all your cables/conduits at ONE END ONLY with a low impedance path back to ground. If you think of your entire system, starting from the last transformer secondary upstream, you want your grounding system to resemble a "tree". The tree is rooted at the ground stake for the transformer. If you have multiple alternate paths in the grounding system, these alternate paths form loop antennas. Your ground connection can actually be creating the EMF that you are trying to avoid. If you run twisted pair wiring (Ethernet cabling), this almost totally avoids the loop antenna problem. Each half-twist is perfectly balanced with the next half-twist on the cable. Each half-twist will create a signal that is opposite in polarity on the cable, so the EMF fields tend to cancel each other out. Although it's good practice to run Ethernet cabling away from power cables anyways (square of the distance...), testing has shown that in practice it is nearly impossible to induce EMF into CAT 5 cabling in the first place. Since CAT 5 is so good at rejecting interference, I have used it in practice for a lot of other things than it's intended use. If you start doing the same, do yourself a favor. Pick a jacket color for your Ethernet cables and stick with it. Any place you use CAT 5 for something else, do not use that color. Right now when you open a panel in the plant, you will see a lot of blue. Some is DC wiring. Some is blue hose (AB DH+ or RIO). Some is Ethernet. Some is CAT 5 used for signal wiring. But you can't tell at a glance what is what. It's almost as bad as troubleshooting an airplane (FAA rules require ALL wires regardless of purpose are white). Edited 31 May 2008 by paulengr

[milldrone 0]

Thank you paulengr, Your post has earned a spot in my bookmarks folder

[JMK 0]

Paulengr - Thank you for the detailed defination! I will also bookmark this post for future reference! JmK