Induction Problem

All good solid and useful infos, gained from real experience I suspect!

Furthermore, low level signals sent over relatively long distances, often need to be sent as twisted pairs with a differential transmitter and receiver, especially where cables with higher level signals are close, mains voltages for example...

Changing the mains voltages to DC (for low current applications) can also often help, but sometimes difficult to arrange/achieve after the event!......

Otherwise cables need routing "apart" as you so correctly mentioned!!

Couldn't have said it better. One solution I have heard is to put a very small ceramic capacitor in parallel with the tacho terminals. This may help to dissipate some of the transience induced by motor starting etc. However it's a bandaid solution. All of the above mentioned are definately preferable. Hope this helps.

As Tornado says, ground one end only of shield. A common resolution to this problem is to ground the source end only to power supply negative (not panel ground). This would apply to all signal cables.

If possible isolate zero volt side of tach output from all other zero potentials (grounds) If you have an oscope you can see spikes from the AC power you are near. If you are checking voltage with a meter you will see a small amount of voltage if you check for AC on your DC tach. Either device is helpful for checking your progress. I have lived in your exact shoes recently, and isolating tach is the best after you have switched to twisted pair with one side grounded shield.

Other areas to consider:

Check the grounding of the individual weigh feeder motors, and measure the frame voltage potential to true ground while starting each motor. We had a problem with a feeder retrofit where the net weight would jump 50 KG whenever a motor started. We could not find the issue until our scale tech was zapped by the loadcell shield wiring. The motor was attempting to ground itself through the scale wiring.

If your weigh feeder controller is PLC-based consider using a communication network (EtherNet/IP, DeviceNet, etc) to transmit the signals. Mount the tach and weight transducers directly on the weigh feeder frame to ensure they float at the same ground potential as the motors. Having a local network also allows for easy control expansion in the future.

First is to look at this on an oscilloscope, is it 50 Hz on top of your DC output? Is it some high frequency from VSD or other stuff? If it's hf you may be able to filter with ferrite ring or capacitor filter across the tach output, you can see the effect on the 'scope.

These problems are often cause by loops, grounding should be in a tree formation, no loops. Also by power & instrument interaction, separate the power gounding to a single point, and similarly the control and instrumentation.

If you can isolate the tacho output signal from ground and receive it via two wires at a differential amplifier, then any pick up is cancelled as the amplifier only sees the difference between the two wires. The differential amplifier output will feed whatever circuit input you have at the moment.

I have some 4-20mA isolation modules at the moment, power in, 4-20 mA in, and 4-20mA out all electrically isolated. I think they exist in volts modules also, you might try to isolate the tacho volts from the tacho line volts because your problem may be an electrical circuit from the tacho power supply, is the -ve 12 volts grounded? Where?

It is not just the signal wires which need attended to. A.C. variable drives should have a filter/choke/transformer mounted close after them to capture the worst transients. This reduces the interference caused by the cabling to the motor.

GM1964,

Agreed, mine are Mitsubishi and they are mounted on a EMI filter which is supposed to block all those transients & spikes. All the inputs are electrically isolated from the mains.

I did have a problem similar to this with an encoder, the pulses had little glitches when I looked on the 'scope. The screen wasn't earthed at all, cured it with an earth at one end, I suspect this guy has a similar connection problem.

<...reduce this Induction problem...>

Change the transmitted signal to a current-based one, such as 4-20mA.

If the tacho DC output is showing readings on a DC moving coil (insensitive to most noise) which are "not-reasonable, then it is most likely the AC input signal to the tacho is being corrupted by pick-up. You need to look at the screening and grounding of the connection to the tacho probe.

Routing signal wires next to variable-frequency drive cables is a recipe for disaster. One time I encountered a problem where a 4-20 mA signal became corrupted by noise simply by being lashed onto a drive conductor with a plastic wire-tie, and this was even on a properly shielded cable! Never, never route signal wires in the same conduit as power conductors, especially power conductors feeding or coming from VFDs with high-voltage spikes and high-frequency harmoncs.

One possible solution would be to replace the voltage-based tachometer with a pulse-type encoder, where the on/off pulse signal would be more immune to noise, but this would require signal conversion back to analog at the weighfeeder controller, and even then would not be a guaranteed solution given the amount of noise you're probably experiencing.

When you look at all the time, effort, and money you might spend on trying to solve the noise problem, installing a separate run of conduit for the signal wires starts looking pretty cheap.

I agree with Tonykuphalt's remarks on putting signal & power cables together being a recipe for trouble.

Good specifications I have seen for plant installations require about the following distances between signal cables and other cables in runs.....

15 cm to DC control cables.

30 cm to AC cables (115/230V single phase).

100cm to high current or voltage cables.

-- and of course, separation of DC & single phase AC & power also applies.

At the design stage, just putting signal cables in a tray/hangers at top left of a duct, with other cables bottom/right can avoid trouble. But when the cables have been installed together and twisted round one another (always seems to be the case for the one cable you want to separate) fixing it is a major problem!

Shielding is no substitute for separation, at power frequencies effective shieldings are very expensive and less effective than separation.

67model