Optimizing speed controller of drive

On some documents I found two methods to tune the speed controller of a drive (PI-regulation):
1. Set integral time to a very large value(f.i. 1000ms)
Input a constant input signal (1..5% of maximum value) to
optimize at a friction-optimal operation point.
Increase Kp until controller oscillates (evaluate
the manipulated value). Reduce Kp to half of that value.
Now reduce Tn until the controller oscillates again. Set
Tn to double of that value.
2. Transient function/ Step respons :
Set integral time to a very large value(f.i. 1000ms)
Input a step reference. The manipulated value must not go
into limit. Observe the actuals speed and adjust the Kp
the response is as fast as possible but there is no overshoot. No decrease Tn until there is just one
overshoot (20-50% of maximum value).
Which of the two methods should be prefered? How do you tune the speed controller of a drive (supposed you don't use a wizard in the drive software). Did you ever manually tune the current controller of drive (supposed you don't use a wizard in the drive software).
Thanks

No suggestions?

PLCUser, first, I can safely say that using the autotune feature in the drive will get you a nice stable system but it will not be optimized. Autotune is always quite conservative.

Second, there are lots of seat-of-the-pants methods to do PI tuning. None of them will get you the best possible tuning. To get that, the purists on this BBS will insist you use mathematical methods and, if you need the very best, I would agree with them.

Third, in my experience, the vast majority of speed loop applications can be adequately tuned by turning off the I time or gain and running the P gain up until you start to see instability. You will see instability in the torque before it shows up in the speed. I then back off from that point around 10% (not 50% as you mention above). Then I bring the I time down or the I gain up until I get instability. I back off from that point about 10% and check to see that the performance meets the application. It is rare that it isn't good enough.

DickDV,

Thanks for your suggestions. Can I conclude that you evaluate the 'instability' only during a constant movement? I know it will be application dependent but what speed setpoint (Hz or % from max) do you implement? Do you evaluate step responses also, and if so what speed setpoint do you use then?

I routinely look for instability at the lowest and then 80% of the highest speeds the application will run. I don't bother with step change response unless the application is dependent on that kind of performance. If it is, I would do it at the same two speeds.

DickDV,

Did you ever manually tune the current controller? I never did so far, I always keep to the drive autotune for that.

DickDV,

I saw in your profile you are a drive engineer. What brand do you support AB, Siemens, Control Techniques?

Plc_User said:

DickDV,

Did you ever manually tune the current controller? I never did so far, I always keep to the drive autotune for that.

Click to expand...

I have, it is quite different from tuning in velocity mode. The control output is proportional to the acceleration instead of the velocity. The acceleration feed forward is much more important when in torque or acceleration mode.

I've always used the autotune for the internal or current loop.

I have a long background with Reliance and now ABB drives.

When the motor is used for a gearing application (f.i. a flying saw), during gearing the motor is constantly accelerated deccelerated to follow the master signal. If this is a noisy master with high accelerations, accelerations of the slave will also be high.
I have seen it in an application where the speed controller was set very aggresive that the drive frequently tripped in overcurrent.
Considering that the load on the moving carriage of the saw (the part that has to move synchronously with the master, and that contains the saw) is always the same, would it be a good idea to do step responses with lower speeds but also with the maximum possible speed and then assure that the speed controller is not so aggressive that the curren limit is never reached? I would even take it a little further, wouldn't it be better to do step moves constantly alternating between forward and backward. In that case the motor is accelerated with maximum acceleration from backward to forward and the other way around. The alternating movement can be performed by the commisioning software, so that won' be a problem.
Second question is : does someone has experience with P- and I- parameters that are speed dependent. In that case you have two possible setvalues to create a straight curb. If you do have experience, how do you set it?
Thanks.

I have used a threshold detector to use higher gains at low speed, for positioning response, but make the drive a little lazy at higher speed, where the exact tracking is not as important (on certain applications). This sure settles the drive down and reduces heat losses!

I wish people would just say what they are doing up front.

Plc_User said:

When the motor is used for a gearing application (f.i. a flying saw), during gearing the motor is constantly accelerated deccelerated to follow the master signal. If this is a noisy master with high accelerations, accelerations of the slave will also be high.
I have seen it in an application where the speed controller was set very aggresive that the drive frequently tripped in overcurrent.

Click to expand...

This is a common problem. A trick is to gear to a virtual master. The virtual master never has noise and can be used to generate feed forwards for the shear axis. The physical master updates the virtual master. In some cases the physical master is doesn't update the virtual master during a cut. In some cases the cut causes a pull or disturbance in the material being cut that affects the master encoder which will cause feed back to the shear. By decoupling the physical master from the virtual master during the short cut time this 'feed back' problem is avoided.

Considering that the load on the moving carriage of the saw (the part that has to move synchronously with the master, and that contains the saw) is always the same, would it be a good idea to do step responses with lower speeds but also with the maximum possible speed and then assure that the speed controller is not so aggressive that the curren limit is never reached?

Click to expand...

No, you want the motion as smooth as possible at all speeds. Minimize accelerations as much as possible. Of course as you go faster the accelerations will increase dramatically. You motion profile has a lot to do with this but you have said nothing about it except you have mentioned making step moves which are bad.

Provide a cut length, cut time and line speed. I can modify a program I have and show you what a good motion profile should look like. I have canned flying shear programs where I can input the cut length, cut time and lines speed and the maximum velocities and accelerations are calculated.
A trend or graph would be good.

I would even take it a little further, wouldn't it be better to do step moves constantly alternating between forward and backward. In that case the motor is accelerated with maximum acceleration from backward to forward and the other way around. The alternating movement can be performed by the commisioning software, so that won' be a problem.

Click to expand...

What? I am confused. Don't you have software that will generate the motion profile for you? Don't do the step moves. A good motion controller should be able to get back to the next cut and sync up smoothly. This should be simple.

Second question is : does someone has experience with P- and I- parameters that are speed dependent. In that case you have two possible setvalues to create a straight curb. If you do have experience, how do you set it?
Thanks.

Click to expand...

Yes but you shouldn't need to change the gains on a flying shear. The load doesn't change.

You should probably be using a PID since you are not simply controlling velocity but also phase or position.

I wonder if you aren't running your drive into saturation and exceeding the torque limits.