Tempeture Controller

Merry Christmas
Status
Not open for further replies.

JdoubleU

Senior Member
We just had a new feed water tank put into our boiler room. On the controller for the motors we have a tempeture controller that is powered with 120 volts. It recieves a 4-20ma inputed from a level sensor and sends a 4-20 ma signal to a to a valve in order to allow water to fill the tank. The Temp contoller has a set point value on it and when it goes below a certain level it opens the valve. The problem is that it everyonce in awhile the current value shoot up high sticks there which in turn tell the valve not to open and then our tank eventually runs out of water. I called tech. support on the controller and he said they are really sensitive to noise. The cable being used is 2 pairs of 18 gauge wire with a bare ground all wrapped in a shielded cable. The bare cable is grounded. Could it be that noise is traveling on that ground.
 

StephenSDH

Senior Member
Location
Allentown, PA
If it is 4-20ma it is not noise that is your issue. Current loops are fairly noise immune.

Put your multimeter in the circuit when it happens and take a reading.
 

JdoubleU

Senior Member
Thank you for your response. Why would this not be an issue with the 4-20ma control. I did remove the line from being in the same raceway as the 208 3 phase feed and it seems to have helped. This issue befor was happening every 20 minutes to 2 hours and now it only has happened once in the last 2 days.
 

StephenSDH

Senior Member
Location
Allentown, PA
Noise is a voltage distortion. The voltage distortion is canceled because it is induced on both the supply and return wire. Plus most analog inputs are very slow response to remove noise.

Again, make sure you don't have an integral wind up. I don't know your process, but if you can set both your integral and derivitave value to zero and try to tune the process first with only proportional only. Only if needed add a small amount of integral.
 

KentAT

Senior Member
Location
Northeastern PA
The cable being used is 2 pairs of 18 gauge wire with a bare ground all wrapped in a shielded cable. The bare cable is grounded. Could it be that noise is traveling on that ground.

Are you saying that your input cable has four 18AWG conductors? How are you using those 4 conductors?

Are your level sensors loop-powered?

Or are you providing them with 120VAC within the same cable as the 4-20mA signal (not good)?

Remember, the shield drain wire must only be grounded at one end. The other end should be cut back and taped or heat shrink applied.

What is the mfg. and model nos. of the controller and sensor?

Kent
 

JdoubleU

Senior Member
Noise is a voltage distortion. The voltage distortion is canceled because it is induced on both the supply and return wire. Plus most analog inputs are very slow response to remove noise.

Again, make sure you don't have an integral wind up. I don't know your process, but if you can set both your integral and derivitave value to zero and try to tune the process first with only proportional only. Only if needed add a small amount of integral.

How are they cancelled? What is meant by integral wind up? Sorry to ask such questions.
 

tkb

Senior Member
Location
MA
Are you saying that your input cable has four 18AWG conductors? How are you using those 4 conductors?

Are your level sensors loop-powered?

Or are you providing them with 120VAC within the same cable as the 4-20mA signal (not good)?

Remember, the shield drain wire must only be grounded at one end. The other end should be cut back and taped or heat shrink applied.

What is the mfg. and model nos. of the controller and sensor?

Kent

I would add, that the shield be connected at the source of the signal.
 

StephenSDH

Senior Member
Location
Allentown, PA
How are they cancelled?

The noise is induced on both the supply and return wires, so since both wires have the same noise present they rise and fall together and it doesn't change the ma input value.

What is meant by integral wind up? Sorry to ask such questions.

You have a PID controller so it needs tuned gains, P, I, & D.

The Proportional Gain (P) is easy to tune. Set your Integral (I) and Derivative (D) to zero. Start with a small P value for gain and increase until you get the response you want. Too much and the process will oscillate.

If you are happy then stop here. Leave your I and D values at zero. If you need to close the error between your setpoint and actual over time, you can add some integral. The amount of integral you add controls how long it will take to close the error between SP and PV. This integral value winds up over time. So it takes time for it to wind up, and then when there is a state change you need to realize it takes time for it to wind down. So your output can be stuck on or stuck off for the duration it takes to wind down.

I can't help you with setting the derivative value. It serves as a shock absorber to limit the response, but isn't needed all that often.
 

ELA

Senior Member
You said the controller tech stated that they are really sensitive to noise.
If that is the case I would not run the signal level lines in the same raceway as the power leads. The general rule is never run signal along with power.
Of course there are exceptions but the safe way is to keep them separate. "Electrical Noise" is a broad term and can refer to many things. It does not only mean 60 hz induced currents. It also can mean high frequency transients that can be coupled to signal wires in the common mode ( not effectively cancelling in each conductor). These transients can cause temporary upsets to control systems.

That being said there are many other variables. I have worked with a lot of liquid level systems in the past and the level transducers themselves can be more or less reliable.
I would talk with the tech. some more on exactly which parts of the system are most sensitive.
Are the level transducers loop powered or do they have a separate power supply?
 
Status
Not open for further replies.
Top