Allen Bradley ControlLogix AC inputs high due to induced voltage

[kmoser]

Hey all,
I'm having this issue with an Allen Bradley AC input card(1756-IA16I). I was looking into another issue and found that the inputs on the card were acting strangely. The inputs are on with ~36V AC induced because they run about 400' or so in a 2" conduit with a bunch of other wires, which is all existing so this has probably been an issue since whenever it was all installed. The obvious fix would be to replace the wire with shielded cable but I wanted to see if there were any other solutions to either pull down the voltage at the card below 30V so that it doesn't trigger the inputs, or if there is a way to do so within the plc.

 

[Corsair]

Sometimes you can get lucky with a 10K ohm 10 watt resistor in parallel with the input. All you need to do is drop the voltage comfortably below the threshold for the module.

And yes - I realize that a 10 watt resistor is oversized.

Less resistance may help more as long as you do not exceed the allowable current of the input sensor.

 

[Corsair]

Rant about module density

The problem may be worse these days than it used to be because of input module density. Twice the points in half the space means 4 times the power density unless the input currents go down - which they have.

If my memory serves me correctly we used to have module currents in the 15 milliamp range. Now they may be half of that. Add a practice called capacitive coupling to the mix and it's no wonder that these noise issues emerge.

There's other issues with switch contact whetting, etc.

Smaller is not always better. Nobody has figured out how to shrink a watt or an electricians fingers.

 

[ck8177]

You could try putting a resistor in series with each affected input to drop the voltage. 2-4 kΩ may do the trick, but I've only had one cup of coffee so I would double-check my math. Long term, I would definitely look at re-routing the control wiring so it doesn't share conduit with any higher-voltage circuits, or at least using shielded cable.

 

[Corsair]

Clarification

It's been bothering me that one answer to this post is a resistor in parallel with the input and the other is a resistor in series. Which is right? The answer depends on the situation - the best solution may be both.

An AC input module has a 'Must turn on' voltage. If the input voltage is above that level the PLC will interpret the input as a 'one'. Let's say for this discussion that voltage is 90 volts. There is a 'Must turn off' voltage. If the voltage is below that level the PLC will interpret the input as a 'zero'. Lets say for this discussion that voltage is 30 volts.

Between 30 and 90 volts the PLC is figuring the input status as a zero or a one but the manufacturer is not saying which value. It's not guaranteed. There could be several identical input modules with the same voltage and some may be at zero and some at one. The control system integrator must see that the input is above 90 when it should be on and below 30 when it should be off and avoid the undefined band between 30 and 90.

The next thing to determine is the 'input resistance'. It's actually an AC impedance and maybe not linear with voltage due to diode turn-on issues but it can be approximated as a DC resistance. Find out how much voltage is on the input from the noise. Also find out how much voltage is on the input when the sensor is turned on - typically 120 vac. A series resistor will drop any input voltage that comes in. If a resistance value can be found that drops the noise voltage to the module below 30 and yet still keeps the sensor voltage above 90 the series resistor is the answer. If the series resistor drops the sensor voltage below 90 then it may not be the best solution.

A parallel resistor may work if the noise voltage is from a 'high impedance' source. It may pull down the noise voltage without reducing the full 120 volt signal when the sensor is on. If it pulls too much current it can damage the sensor.

This is an inexact science. It may not be possible to find a combination of resistors that works. A combination that works today may not work tomorrow with other VFD speeds or equipment changes. Resistors can be good for dealing with predictable leaky two-wire sensors but they are not ideal for noise suppression.

The ONLY real answer is to avoid the situation completely. Isolation is the first choice (separate conduit), shielding is the second. Then and only then can you go home and confidently move on to your next problem.

 

[Corsair]

Rare answer - for general information

If there are a small number of input signals - and the noise is really nasty - and you have enough conductors through the pipe - and you're not concerned about electrical code issues for the moment - and you've got a power source at the sensor end - and you've tried other methods - and you really want to go home and get some sleep -

I had a pump call signal at a water plant going to a starter at a well. The control pair was in the same trench as the power feed to the well with the cables laying next to each other . I could pull in an ice cube relay from the induced noise. I tried to damp it out with a 60-watt incandescent bulb - which just glowed at me.

The TEMPORARY solution was to use a spare control power transformer to generate isolated balanced-pair voltage from one end. The other end fed an isolated relay coil. Now the input leads were being used in a differential mode where the induced noise voltage was the same on each wire. The noise (mostly) canceled.

With microphones you have high-impedance 2-wire connections to phone plugs. For better noise rejection you go to a 3-wire balanced line with a 3-prong connector. Same principle.

It worked well for a couple of weeks until we got a data radio between the two ends. Isolation beats suppression every time.

 

[Saffa]

Isolation beats suppression every time.

This is why I use fibre optic and remote I/O as much as possible these days. Those big ol' multi-cores we used to run for hundreds of metres back to a central location just make me cringe these days.

 

[kmoser]

Between 30 and 90 volts the PLC is figuring the input status as a zero or a one but the manufacturer is not saying which value. It's not guaranteed. There could be several identical input modules with the same voltage and some may be at zero and some at one. The control system integrator must see that the input is above 90 when it should be on and below 30 when it should be off and avoid the undefined band between 30 and 90.

The next thing to determine is the 'input resistance'. It's actually an AC impedance and maybe not linear with voltage due to diode turn-on issues but it can be approximated as a DC resistance. Find out how much voltage is on the input from the noise. Also find out how much voltage is on the input when the sensor is turned on - typically 120 vac. A series resistor will drop any input voltage that comes in. If a resistance value can be found that drops the noise voltage to the module below 30 and yet still keeps the sensor voltage above 90 the series resistor is the answer. If the series resistor drops the sensor voltage below 90 then it may not be the best solution.

It looks like the input resistance for the card is around 8kΩ, so a 3.3k resistor in series would drop the voltage enough to prevent the inputs from staying high. It would drop the voltage coming back from the contact in the actuator to the plc to about 85V, so I may drop the resistance down a little bit more to get under our 30V still but keep the closed contact voltage higher.



All of the conductors in the conduit are 120V out to contacts on actuators and the corresponding conductors that come back to the plc for the inputs. There is just a large number of them in a conduit that runs down a long drive, the power ckts for the actuators are totally separate as are the analog signals to and from the actuators. We will be looking into replacing the conductors with shielded cable once we get back into warmer weather, so I just wanted to make sure I was on the right track for a temporary solution.

 

[Corsair]

I hope you don't run into this - but with the 3.3K and the 8k in series you increase the load resistance seen by the noise source. Sometimes it seems that this higher impedance allows the noise voltage to go up and you don't get as much voltage reduction to the input module as you hoped for. This may lead to an additional resistor in parallel with the input module.

Don't get discouraged - there should be a combination that will work. I didn't look at the specs on your input module - maybe they are closer than the 90 and 30 volts that I mentioned for discussion.

Solving problems like this is one mark of a true integrator - someone who isn't just another pretty face with a laptop.

 

[Corsair]

Putting a resistor from switch input wire to neutral out at the actuator may keep the input wires from acting like antennas.