Noise on an analog input

[brycebsp 0]

Hello all. First time poster. I am hoping someone here can offer some advice or a solution to a problem I am having. I am working on a system that consists of a control panel (With PLC and HMI) and located adjacent to the control panel is two VFD's in their own enclosures. All the enclosures are steel. On the PLC there is a mixed analog I/O module with 4 (4-20mA) inputs and 4 (4-20mA) outputs. I am using one input connected to hydrostatic pressure transmitter for measuring water level. I am using two outputs from the module to send speed reference to each VFD. The problem that I am having is that when the VFD's are running with a load my analog input bounces around. If I go into the register for the analog input and watch the raw value it will randomly jump up or down as much as 100 points. The resolution of the analog input is 12bit (0-4095 raw value). I have tried disconnecting the transmitter and feeding a steady mA signal from my Fluke process meter and I still have the same problem. Example... I fed a steady 12mA in and start the VFD's signal begins to bounce around. Shut off VFD's and bounce stops. Things I have tried so far.... 1. Checked grounds for good connections and potential ground loops (SEVERAL times) 2. Replaced Analog Module. 3. Replace entire PLC 4. Added DVDT filters to the outputs of the VFD's. 5. Added rungs of logic that do a 3 sample average of the analog input in hopes of quelling the size of the bounce. 6. Tried an Allen-Bradley PLC instead of the one I am currently using (EZ-Automation) and had the same problem. 7. Tried connecting .01 microfarad caps inline with shields on 4-20mA circuits. 8. Replaced 24VDC power supply. Other things to note: all the analog signals that go in or out of this analog module go through Dataforth DSCL22 isolators. Also if I disconnect the motors from the VFD's (no load) and ramp the drives up and down I have no problems. The entire system is mounted on a steel platform mounted 15' in the air. It is located adjacent to a river so the idea is to have the control equipment above flood levels. I feel like this big steel platform is acting like a giant antenna and causing my problems. I realized this is difficult for anyone to solve via e-mail. I am just hoping some of you can make suggestions of additional things to try.

[Clay B. 1]

Had a problem similar to this years ago. I found that my problem was the ground loop of the motor and drive. Instead of taking the motor ground directly to my ground lug I took it back to the drive and then took a lead from the drive to ground. Somebody else may be better able to explain why that works ( It was explained to me but I have forgotten the reason).

[Chris Elston 78]

Clay B. is right.. If you have VFD rated CABLE, and that comes back to the panel. What everyone wants to do is land the ground to the ground bar on the panel or a GREEN/YELLOW terminal block, (the kind that bonds the terminal direct to the DIN RAIL, then into the panel ground. What you want to do INSTEAD is make sure the ground from the VFD motors, go DIRECTLY back to the VFD it came from. So if your using terminal blocks, change to the soild green terminal block, (the kind not bonded to DIN RAIL, and pass the ground through the terminal block and back to the VFD it came from. There are two ground lugs on a VFD, one is for the motor return ground, the other ground lug is to the panel ground. Is that how your motor and VFD are grounded?

[panic mode 246]

i hope your motor cable is shielded (ok, ok, just checking), see that shield is not removed more than needed (some 2"). you mention motors (plural)... how many and how exactly are they wired?

[Clay B. 1]

Do you know the reason behind this Chako. I have been trying to remeber why it is today and I think there are just too many beers between now and then.

[PdL 71]

Also do not ground the shield anywhere but the inverter. The closer the shielding runs up to the inverter, the less chance EMI emmission the leads radiate can be picked up by other conductors in your panel. We always have a small earth rail as close as possible to the motor connection terminals on the VFD, and use a clamp that encapsulates the shielding all way round. I'll see if we have a panel in the workshop and take a picture. Also at motor side best way is to use an EMI gland. Here's a picture of what I mean. We always have the motorcables run up to the inverter outside of the ducting.

[brycebsp 0]

Thanks all. I will investigate where the motor ground is connected when I go to site next. From memory I believe the motor ground goes to copper ground lug on the VFD enclosure back plate and then is bonded by a wire from there back to the VFD ground. The motor does not have shielded motor cables. The motor is part of a submersible pump therefore Submersible motor cable is used from the pump back to the VFD. The motor is 45HP and as I mentioned earlier the VFD's are each in their own cabinets seperate from the control equipment. Of course there are some control connections between the VFD and the control system. Control wiring is in a seperate conduit from the motor leads and all conduits are galvanized and grounded. I am thinking about changing out all the control wires to twisted shielded to see if that makes a difference. Any additional thoughts are appreciated.

[TConnolly 13]

Are you saying that your analog signals are not using shielded twisted pair cables? Call me crazy, but yeah, that could be the problem right there. I spec Belden 8760 for most of my applications. Edit to add: Make sure you ground all the instrumentation shields at one end only (not to be confused with inverter cable sheild) preferably in the control panel. Edited 5 Oct 2007 by Alaric

[Clay B. 1]

Even though your VFD's are in seperate cabinets they could still be sharing a ground with your controls. Keeping those seperate is always a good thing. Also what Alaric said is pretty much a standard as far as I know. Analogs are always a twisted pair with a sheild and the sheild is grounded only on one end. I always ground at my control cabinet and will apply heat shrink to other end to reduce the risk of grounding that end and making my sheild an antenna. If the VFD does not have 2 ground lugs, put your motor ground wire into the ground lug on the VFD and come out from there to your ground buss. The object is to create a ground loop between your drive and the motor so the RF noise created by the carrier frequency of the VFD does not get introduced into your ground. Do have one other question. Do you ramp down your pumps are do you just use the VFD's to get up to speed and stay there and coast stop when you shut down? If you do ramp down, have you had a high failure rate out of your drives?

[paulengr 44]

Not that this is necessarily your problem but I've had problems with poor common mode isolation with Allen-Bradley 4-20mA inputs before. The way that this manifests is that with a single 4-20mA loop connected, all is well. Once you add the second one (or more...the pattern is not always so easy), all of a sudden ALL the loops go haywire. The solution is to give the 4-20mA receivers a common reference on one side or the other. It depends on your particular wiring but quite often this means simply shorting together the returns of all the 4-20mA receivers. I've seen this happen even without VFD's.

[brycebsp 0]

Twisted shielded cables are being used for all analogs (4-20mA's) there are some Relay outputs from the VFD's that are 120V and run in seperate conduit. These were the wires I was talking about changing to shielded as well. I have several spare Analog in's and outs that are being used. They are essentially sitting there open. Could terminating these make a difference? If so what is the proper way to terminate them? To CLAYB yes the motors ramp up and down as required to flow match. Have been using this Brand of VFD for over 5 years with very few failures. I really had hoped the DVDT filters would have resolved my problem since the noise seemed to only be present when the drives have a load on them. Thanks for the input so far. Bryce

[Clay B. 1]

Are the unused Analog I/O's cabled? What I mean does a cable come out of them. If it does you might want to disconnect them from the Analog cards. Also Paul mentioned something that is also true. Making the commons truly common to each other on the analog cards helps. One thing that was stressed to me when I was learning about controls is that noise is additive. A little nosie from several different sources is just as bad as a lot of noise from one source. If you can get your hands on a O-scope You might try and bring things on-line one system at a time and see how the signal degrades. One way I have gotten an O-scope is to talk to my local motor rebuilder. These guys always have an O-scope and are usually happy to help you with problems like this.

[TConnolly 13]

Its generally good practice to short the terminals of unused analog Inputs. INPUTS ONLY, NEVER OUTPUTS, and some PLC manufacturers recommend this in their instructions. That probably won't help much for your noise.

[PdL 71]

Note that having unshielded motor cables is still a major issue when having EMI problems. You are trying to find other solutions but the your starting point should be shielded cables... With DVDT filters you mean a 3 phase AC choke reactor in line with the motor ? You can ask the VFD manufacturer if they also have DC choke reactors which some drives have as an option. The combination of the AC and DC reactor will be the most effective. Have you tried setting the switching frequency of the drive to it's lowest setting ? Might introduce ambient noise in your motor but still better than electromagnetic noise in your cables

[paulengr 44]

Disconnect everything from analog outputs (leave them open and terminals only). Short the inputs of the extra analog inputs at the card. This might be a silly question, but did you ground any shields at ONE point only, with a separate signal ground vs. general ground? With AC, the goal is to achieve proper bonding and grounding. For the grounding side of things, the transformer secondary should be grounded at a single point. In other words, the neutral and ground bars are tied together at the first point of distribution and NOWHERE ELSE. From that point forward, everything else should be bonded. This means that all metallic cases, motor housings, etc., should be shorted together and all should connect to your grounding point. The purpose of the bonding (different from grounding) is to prevent dangerous voltages from appearing on any components in the event of a fault (shock hazard) as well as to conduct enough current to trip the ground fault protection (circuit breaker/GFI/fuse). It is possible to run distributed grounds (grounding at multiple points) and get away with it but my experience is to stay away if at all possible. Although this has been practiced by the utilities for years, just ask anyone in the business if they've fought a "stray voltage" issue yet. If they have, then they will probably agree that distributed grounds are a sin. If you have distributed grounds, chances are highly likely that you will have ground loops and conduct current across your grounds continuously (and "ground" will not mean "ground" anymore since if there's a current flow, this means there's resistance and voltage present). In an ungrounded delta system, the only real significant difference is that you don't have a neutral. Frankly, I'd recommend even considering using a resistively coupled neutral rather than a capacitively ("ungrounded") coupled one. At least then you have more control over your system voltages relative to the local Earth ground. Especially with 3-phase equipment, the neutrals and grounds tend to inductively pick up voltage if nothing else. So the best way to handle 4-20mA circuits is to make sure to run a completely separate grounding system, tied together at a single point (where you connect your DC negative to ground). All shields should be grounded to the same system, preferably on the end closest to the DC signal ground, at a single point. If you tie your shields at BOTH ends, this creates a loop antenna which easily picks up any stray magnetic fields in the area. Just as with an AC system, treat your shields and "signal grounds" (negative terminal from the DC power supply) as separate systems so that you don't induce noise from one to the other. They are capactively coupled anyways if the wiring runs are sufficiently long. Check the current loop readings yourself. There are three methods I use. First, measure the resistance across the receiver since almost all receivers consist of a precision resistor and an analog/digital converter reading the voltage across that resistor. Then connect it back up and use Ohm's law (V=IR) to figure out what the current loop signal is. Second method if you have one is to use an actual ammeter if you have one with the right range to make it work. Third method is to get your own high quality (metallized film, NOT cheapo penny carbon) resistor and use that as a current probe. The goal here is simply to verify whether YOU can see the noise or not. Another possibility is that you may have a crappy DC power supply in the excitation. If possible, change the current loop from internal to external excitation and provide your own excitation for the loop. Use a high quality DC power supply for this. I have had good luck with switching power supplies from Omron and Sola for this. I don't receommend linears. At one time linears were less noisy because industrial switching power supply designs were very poor. But these days, linears are mostly useful as multipurpose power supplies and cabinet heaters. If you want to check "how good" a power supply is, simply put your meter (true RMS) across it (with a constant load! and make sure it has something for a load!) and switch from DC to AC. Divide the AC voltage it reads by the DC voltage. Finally, check the actual resistance of your current loop at the excitation source (voltage generator). Current loop receivers will typically use between 100 and 1000 ohms of resistance at the receiver, with a typical drive maximum between 1.5 kohms and 10kohms depending on the transmitter capability. If you are seeing more than this, check your connections.

[dinesh patwal 0]

how to configure the temperature in mitsubishi plc...

also the AD CARD CONFIGURATION