is the Dynomotion controll as confusing as it seems? - Page 6

# Thread: is the Dynomotion controll as confusing as it seems?

1. Originally Posted by TomKerekes
Its not clear you have a good working, tuned, tested configuration. Post Plots showing how the Servos are performing.
working on tuning, I have all the servos where I can command them to rotate fwd and back, and they count.

2. ## Re: is the Dynomotion controll as confusing as it seems?

Here is a couple screen shots of what im getting when attempting to balance, I already tuned this axis.

3. ## Re: is the Dynomotion controll as confusing as it seems?

Hi Gunmachinist,

Well at least you show movement

But it isn't following what it should be doing at all and in fact moving the wrong way. (Blue line travels one way Red the other) (output Green just slams between +100 and -100)

Change the sign of the Output Gain so the motor drives in the right direction.

The max output of only 100 limits the drive to 5% - maybe ok for now to test very slow moves first.

Remove the Feed Forward. That is used later to tweak things. We want to learn to steer down the road slowly first, before we try to anticipate up coming curves.

Zero the Axis before doing the plot so the Command and Position are not miles apart.

The Velocity is set to a crazy high value. 5000000 counts per second. But because the acceleration is set low (13000) the trajectory speed is greatly limited. It would take something like 300 seconds to accelerate to your set speed. Set a more reasonable velocity of 20000 counts/sec

Set the Deadband Range to 0. It just adds 1 count of confusion at this point.

Set I and D to zero

Try a P gain of 1

Regards

4. ## Re: is the Dynomotion controll as confusing as it seems?

Ok, we may be cooking with gas. How do I determine which perimeter to change when looking at the graph?

5. ## Re: is the Dynomotion controll as confusing as it seems?

Ok, we may be cooking with gas.
Yes in the second plot the Red is following the Blue. Poorly as it is it is trying, stable (not oscillating), and nothing is saturating (being clipped).

How do I determine which perimeter to change when looking at the graph?
There isn't any simple answer. You must study the graph and realize what is going on. You must read up on each parameter to know what it does.

The Blue plot is what you are asking the motor to do. It is determined by the motion profile parameters and the size of move.

The Green plot is what is being commanded to the motor to do.

The Red plot is what the motor actually did.

Error is the difference between the Red and Blue Plots.

Error creates the Output based on the P I D parameters (and Filters). P Gain drives the Output proportional to how big the error is. I Gain accumulates output based on how big and how long the error has existed. D Gain drives the output based on how fast the error is changing.

Imagine trying to steer a car down the center of a winding road. You look at how far off center you are and turn the steering wheel based on that. If you make huge corrections when only off a small amount you will likely weave, overshoot, and crash. If you make tiny corrections when off a large amount you won't follow the the center well. The idea is to find the best balance.

The next step would be to increase P to try to reduce the errors without going unstable.

Please determine what your resolution is. Post 98 we described how to do this but you didn't respond. Without that information we can't relate the counts to real distances and velocities.

Its difficult to teach/help someone when they don't provide feedback on what parts they understand. Please add more detail into your responses.

HTH

6. ## Re: is the Dynomotion controll as confusing as it seems?

I was able to get a measurement on the counts, when I measure 1 rev. I get about 800 +/- counts, I also tried 5 to 10 revolutions and get anywhere from 770-840 counts. I also played with the P gain and if I get over .065 I get a fault and the machine shuts off power, I have to disconnect power and reconnect to get everything back. I also adjusted the V and A perimeters and get no real change that I can tell.

7. ## Re: is the Dynomotion controll as confusing as it seems?

I was able to get a measurement on the counts, when I measure 1 rev. I get about 800 +/- counts, I also tried 5 to 10 revolutions and get anywhere from 770-840 counts.
Sounds like 800 counts per rev, but that doesn't tell us what the resolution is without knowing other particulars of your machine. Please do what we requested in post 98.

I also played with the P gain and if I get over .065 I get a fault and the machine shuts off power
That would indicate we are trying to accelerate too fast.

V and A perimeters and get no real change that I can tell.
What values did you try? How much did you reduce A? At some point the acceleration should be so gradual it should not fault.

8. ## Re: is the Dynomotion controll as confusing as it seems?

Originally Posted by TomKerekes
Sounds like 800 counts per rev, but that doesn't tell us what the resolution is without knowing other particulars of your machine. Please do what we requested in post 98.
I can't complete the second part of that, the servos are not connected to the machine and I haven't been able to get the limit switch to work yet.

9. ## Re: is the Dynomotion controll as confusing as it seems?

I can't complete the second part of that, the servos are not connected to the machine
Can't you at least estimate how far the machine will move if the motor were to do 1 revolution? Are there lead screws? What pitch? Directly connected to the motor?

and I haven't been able to get the limit switch to work yet.
You shouldn't need limits to do a move.

10. ## Re: is the Dynomotion controll as confusing as it seems?

I believe may.need them, when the servo does a move it will continue to rotate at.the end. For example it will rotate one way the back, but when it get to the end of the cycle it will continue to rotate until I disable it.

11. ## Re: is the Dynomotion controll as confusing as it seems?

Is this the correct wiring for the limit switch or should I have connected the Pos. through Opto 0 +/_ and neg through Opto 1 +/- ?

12. ## Re: is the Dynomotion controll as confusing as it seems?

You didn't label the black wire.

From post 77 it seems you have a dual power supply with common grounds. I think this because there seems to be just 2 terminals both called "COM". This would be easy to verify with an Ohm meter. With power off and everthing disconnected it will either measure 1 ohm or less or 1 million ohms or more.

If they are common grounds then the optical isolation will be defeated and any noise and spikes from the long limit switch wiring going all over the machine will be coupled directly into KFLOP/Kanalog potentially causing problems or damage.

Please verify the supplies are isolated. If not then obtain an isolated supply.

13. ## Re: is the Dynomotion controll as confusing as it seems?

Hi I'm a little late to the party, but do you still intend to use the tachs?
In this day and age they are never used in servo applications, and are a vestige of a long gone era, fortunately, I have (re) used them in the past but removed the tach brushes and hooked just the motors up to the AMC drives and used the torque mode rather than the old velocity (Tach) mode.
In the torque mode, there is virtually no tuning of the AMC needed.
Just my 2c.
Al.

14. ## Re: is the Dynomotion controll as confusing as it seems?

Originally Posted by Al_The_Man
Hi I'm a little late to the party, but do you still intend to use the tachs?
In this day and age they are never used in servo applications, and are a vestige of a long gone era, fortunately, I have (re) used them in the past but removed the tach brushes and hooked just the motors up to the AMC drives and used the torque mode rather than the old velocity (Tach) mode.
In the torque mode, there is virtually no tuning of the AMC needed.
Just my 2c.
Al.
I'm assuming the small brushes are the Tach brushes? I may give it a try and see if it is better with my system.

15. ## Re: is the Dynomotion controll as confusing as it seems?

Originally Posted by TomKerekes
You didn't label the black wire.

From post 77 it seems you have a dual power supply with common grounds. I think this because there seems to be just 2 terminals both called "COM". This would be easy to verify with an Ohm meter. With power off and everthing disconnected it will either measure 1 ohm or less or 1 million ohms or more.

If they are common grounds then the optical isolation will be defeated and any noise and spikes from the long limit switch wiring going all over the machine will be coupled directly into KFLOP/Kanalog potentially causing problems or damage.
I don't know if they are common, I believed that 2 comm ports meant they were separate, 1 for 24v and 1 for 5v. (fu*" if I know)

Please verify the supplies are isolated. If not then obtain an isolated supply.
I believe they are but ill check again.

16. ## Re: is the Dynomotion controll as confusing as it seems?

Originally Posted by Gunmachinist
I'm assuming the small brushes are the Tach brushes? I may give it a try and see if it is better with my system.
Yes, the tach will be the smaller of the two pair.
Al.

17. ## Re: is the Dynomotion controll as confusing as it seems?

P.S. Don't forget to change the DIP switches from tach to torque mode on the AMC drive.
Al.

18. ## Re: is the Dynomotion controll as confusing as it seems?

Originally Posted by Al_The_Man
P.S. Don't forget to change the DIP switches from tach to torque mode on the AMC drive.
Al.
I took your suggestion and modified my X axis servo and I believe it made a difference. Thanks for the advice..

Here is a screenshot of my current X axis balance attempt.

19. ## Re: is the Dynomotion controll as confusing as it seems?

I was able to get the X axis hooked up today and did some initial balancing. The table moved about .25 each time I ran a balance cycle. Where do I need to go from here.

20. ## Re: is the Dynomotion controll as confusing as it seems?

The table moved about .25 each time I ran a balance cycle.
With a Move size of 2000 counts that would indicate the resolution of your system is about 8000 counts/inch.

Where do I need to go from here.
Its hard to help when you don't tell us what you did, what happened, and what you understand.

One of the first steps in tuning is to maximize the D Gain. In the previous Post D was 5. Now you have 2.5. You should probably go back to 5 or higher. See steps 3 and 4 here.

Note D Gain introduces a type of drag which requires some error to overcome. V Feedforward can be used to compensate for this. Set V FF to D * 90e-6. For a D gain of 5 it would be: 0.00045

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•