Induced Voltage on Multi-Conductor Cable 1

[controlnovice]

We have a 38 pair (I think) multi-conductor cable about 600ft long carrying 120V signals for on/off valves (powering solenoids and position feedback). The signals are going to a DCS system.

During commissioning, it was found that while testing one valve, other position switches would indicate in the DCS. Our solution was to tie resisters on one side to ground to 'drain' the induced voltage. I don't remember what the 'turn on' voltage/amps is for the DCS.

I have several questions on this:
1. Is there any way to find this during design? We do wire sizing based on amps, voltage drop on distance, etc. But is there a way to calculate voltage induced on adjacent cables?

2. Is the resister installation a common method to fix the problem? I've seen/done it before, but have others? Is there a cleaner method? Are there any terminal manufacturers that have this built in?

Is the only clean solution to reinstall the cables in smaller multi-conductor cables?

ControlNovice

 

[felixc]

Six hundred feet is quite long to power solenoids and read data along the same cable.

The induction is directly proportional to the current.

I don't know if your driving circuit is capable of it, but if it can turn on and off when the AC crosses zero volts, it will reduce one source of noise.

Perhaps you could use one cable to carry the driving signals, and one other cable to carry the read signals.

Using filters at your inputs will also help. It depends what are the signals to be read. The load resistors are one way to filter out the induced noise.

 

[dspDad]

You probably have both capacitive and inductive coupling.
You might also have the ir drop showing up.
Twisted pair for each load will eliminate virtually all of these problems.

If you had to pick one problem, the capacitive coupling will probably be the biggest problem. This will be proportional to voltage rather than to current.

You might have 0.1 to 0.01 uF between lines. This represents 30k to 300k ohms of coupling, so a 1k resistor will reduce it to between 3 and 0.3 percent.

The worst case is if you already have twisted pair but aren't using matched pairs,and the best case would used two cables, each with twisted pairs running drive signals in one and sense signals in the other.

Dspdad

 

[controlnovice]

Well, okay. I know WHY the affect is there....

But my questions are still not answered.

1. Is there a way to design this affect out of the system in the upfront engineering.

2. Is there a cleaner solution than soldering resisters in the circuit?

 

[lrob]

I do not know anything about DCS systems, however coming from PLC background, does DCS system have ability to do remote stations? For upfront engineering setup remote stations then network back to main controls. Once again I have never worked with DCS, so this is part question part suggestive idea. Also for upfront engineering there was several suggestions to separate out going signals from incoming, placing them in separate cables.

 

[busbar]

Is each limit-switch contact on a dedicated pair? What are the input-impedance characteristics of the [120VAC?] DCS channels? Has a meg-o-meter been used to weed out cable-insulation problems? Are unused pairs in the multi-pair cable floating or grounded?

 

[controlnovice]

Irob:Yes, there is remote I/O for the DCS that we could have used. We didn't think the 600ft would be too long, or cause induction problems. We had existing signals coming from that distance, but to relays. When we installed the DCS, it became a problem due to the low triggering current (or voltage - Im not sure) in the DCS I/O card.

busbar: Each limit switch contact is on a dedicated pair. The input-impedance is 60k ohm and the detection is >1.4mA on, < 0.56mA off. I don't think the unused pairs are grounded - they are floating.

But I digress again....Can someone answer my original questions?

1. Is this something a design engineering firm should have known or calculated...so a better design could have been done AND
2. What is the solution? Is there a cleaner solution than to insert resisters into the circuit? Is the most clean to have remote I/O? Pull new cable - one for signals, one for power to solenoids?

 

[2dye4]

Several things can be done to up front make the
problem unlikely.

As DspDad said using twisted pairs on the returning signal
lines is the number one choice.
Also allways consider using a seperate cable for control
outputs and signal returns.
The ability of the outputs to affect the inputs is
proportional to the time rate of change of voltage on
the ouput wires. Using solid state relays with zero
crossing turn on will help much. If mech relays are
used there are RC packs that can be wired across the
relay terminals to reduce the rate of voltage rise in the
outgoing circuit.

 

[busbar]

There have been several threads worth searching here regarding long control loops. It may be wise to check voltage and current on falsely triggered AC-input channels to separate sensitive inputs from dspDad’s mention of likely “capacitive and inductive coupling”.

 

[BrianR]

controlnovice,
You can design the system to minimise induction between conductors.

Best design: Two cables seperated by several inches, one for power/control, another for feedback or whatever.

Otherwise, if you are locked into a single cable: ALWAYS use twisted-pairs for this sort of application, the more twists/m the better. The best suppression comes from having a floating load and grounding one of the wires at the source if possible. The suppression is typically 10-20dB better when the pair is also shielded, shield connected at each end, to the conductor which is grounded. If the load is not floating then suppression is much poorer.

The induced voltage is mostly due to the magnetic field surrounding the power conductor coupling transformer-wise into adjacent conductors. Twisted pairs greatly reduce the coupling.

Next: In all cases, suppress any inductive switching transients directly at the load. This reduces the high frequency component along the line and hence the induced voltage.

 

[dpc]

A star for BrianR's practical advice. 120V ac solenoid valves are well-known as potent sources of electrical noise, especially on de-energizing. I would not recommend powering this directly from a DCS output unless you had a big hunky output relay on the dcs output.

I agree with Brian that the solenoids should have local surge suppression.

120V ac power outputs should be run in separate cables, and preferably separate conduits from low-level signal inputs. (are you sure the inputs to the DCS are 120V ac? - this would be unusual)

Analog signal inputs should always be run in twisted pairs with invidual shields on each pair. Discrete inputs are also generally run in twisted pair cable, although often with just an overall shield.

Just another point on the solenoid valve power circuit - each valve circuit should be individually fused - these solenoids are notorious for failing to ground.

So, I would say the answer to your question is that an experienced design firm would not have designed a wiring scheme as you described and should have employed more noise prevention measures up front in the design.

 

[controlnovice]

Thank you all!!

I now know what to look for and design around in the future. For this instance, the project is complete. We installed the resistors and it worked, but this doesn't look 'pretty' according to the plant manager.

Do any of you know of a terminal manufacturer that makes terminals with 'grounding' resistors or another device (caps)?

 

[Skogsgurra]

controlnovice,

Adding (1 kohm, as suggested) resistors results in a lot of watts produced and heat in the cabinet. At 120 V you will get about 15 W/input - and that's a lot of heat if you have many inputs. Even at 10 kohms, you will get 1.5 W/input and 10 kohms is not low enough to eliminate the parasitic signal. It is much better to use capacitors. The size depends on your cable caracteristics, but a safe choice is 100 nanofarad between input and common.

That is a standard procedure and it will load the line enough to reduce the parasitic (mostly capacitive) voltage to harmless levels. Capacitors do not generate heat.

 

[resqcapt19]

One method that I have seen used for this problem is to install pilot lights on the input circuit, either in the field or at the DCS to control this problem.
Don