PID Tuning 2

[DNort]

I am trying to tune a PID controller to control a motor via a variable speed drive. The vsd used is a siemens micromaster 410 and ramp up and down times are low (1.5). I cant seem to get the speed output to remain steady. When in manual mode it produces a perfect steady output but in automatic (PID) the speed output is very unsteady.

Can anyone help with this?

Regards

Derek

 

[DRWeig]

Need lots more information.

Perhaps if you gave us your existing proportional gain, integral time, and derivative time, plus a graph of the input and output, we might have a suggestion. It would be good to include the process variables -- for example, is your input pressure? Flow? What are the values currently stored in the drive for all configuration parameters? What is the characteristic of the load? Dead time?

Tuning questions are tough to answer from afar.

David St. Clair's methods have never let me down so far:

http://www.straightlinecontrol.com/

Good on ya,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[DNort]

Hey my existing propportional gain the automatically set value is 2.0 integral 20ms but I have tried a thousand variations of these values and still cant seem to get a steady output.

 

[DRWeig]

The system may not be controllable, that happens sometimes (very oversized, or prone to large upsets in short times, or other). It doesn't happen often, though.

Your VFD may have dozens of configuration parameters that could be limiting your ability to control in a stable manner.

Your comment that you have tried a thousand variations of the tuning constants indicates to me that you don't have a good understanding of tuning control loops. There are systematic ways to tune a PID loop that yield optimal results (or close to it) quickly. It might be best if you got hire someone local to come have a look. There are too many variables involved to troubleshoot in a forum.

Let us know how it works out!

Good on ya,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[PNachtwey]

It would be good to collect some data in the form of time, control output and speed. From this you can develop a system model. Then you can tune your system.

You don't just tune a PID. You tune a system. This means you need to know something about the system you are tuning.

Some controllers have a auto tune function that will automatically tune a system for you but these will only work if the system is relatively perfect and then a auto tuner shouldn't be necessary any way.

So do you know the gain and time constant of your system? If not you should find out.

Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

 

[iop995]

Try these :
* Lower the derivative value to 0, we will not change this value from zero after this first step.
• Lower the integral value to 0, easy second step.
• Raise the proportional value 100.0
• Increase the integral value to 100
• Slowly lower the integral value and observe the system’s response.
• Since the system will be maintained around setpoint, change setpoint and verify if system corrects in
an acceptable amount of time. If not acceptable or you would like a quick response, continue lowering
the integral value.
• If the system begins to oscillate again, record the integral value and raise value to 100. Just like me,
you got a little greedy trying to get the quickest response.
• After raising the integral value to 100, return to the proportional value and raise this value until
oscillation ceases.
• Lower the proportional value back to 100.0 and then lower the integral value slowly to a value that is 10%
to 20% higher than the recorded value when oscillation started. (recorded value times 1.1 or 1.2)
Change the setpoint and watch as the system tracks quickly and efficiently. If you experience an overshoot that
is not desirable, consider using the setpoint ramp parameter. It is most useful at system start-up or when a large
setpoint change is introduced during system operation.

Details may be find on internet.

 

[PNachtwey]

@iop995, are your gains specifically for the micro master? A gain of 100 means what?

I prefer a simpler method like this.
I made a rough video that is unedited but short.

http://deltamotion.com/peter/Videos/AutoTuneTest2.mp4

My system requires a derivative gain because it is in torque mode. Notice how I had to specify that I was tuning a system in torque mode and not a velocity or voltage mode. The point is that I wouldn't give advise until I know more about the system.











Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

 

[sreid]

What are you using for a tachometer and how is it connected to the motor?