Motion Challenge/Problem

[davidg68124]

Just an update guys.


With all of your help, and Rockwell's Jim Svarovsky's help, we were able to use the phasing suggestion earlier and apply it in Contrologix Motion.

There were a few hurdles, but we managed to get it with an average error of 1/32" or less.

It required using 2 additional Virtuyal axi on top of the Virtual Encoder axi, but it works well and accurately.

During the move, I monitor the error and issue a incremental move on the servo while it is following a virtual master axis, to catch it up to the position of the virtual encoder axis. When I am in position I stop. The pahsing commands are so smooth that you do not even see it until it stops. When it stops, it usually overshoots a bit so it backs up. This is the only visual cue that anything is happening outside the norm.
Thanks for all the help.

David

[Peter Nachtwey]

Originally Posted by davidg68124

It required using 2 additional Virtuyal axi on top of the Virtual Encoder axi, but it works well and accurately.

Now I am confused as to why.
It should have simpler than that.

[davidg68124]

The reasons for the virtual axi.

I have 2 encoders. Both required a virtual axis to momic the count and allow me to gear to them. There was a knowledge base article on how to configure a virtual axis to follow an encoder. By doing this, I am able to take advantage of the Motion Commands in Contrologix to do things such as gearing and position re-defining easily.

I then had to create 2 other virtual axis. One I called the Master Command Axis. I geared the physical axis to this. This allowed me to issue a master MAM instruction on the virtual axis without superceding it with any phasing MAM's in the motion planner(I have learned alot about the motion in contrologix as of late). In other words, when I issue corrections to the physical servo, these correction MAM's will not supercede the master move command.

I then had to create a virtual axis, called Master Accumulation Axis that is geared to the encoder virtual axis which has the right conversion constant to convert the counts to inches and accumulates the position in inches.

I then compare the actual accumulated postion to the master command axis position and issue correction MAM's as necessary to "catch" the physical axis up to the virtual accumulated distance axis.

It seems like alot more than I might need to do, but it is all necessary as I just could not get any other combination to work using Contrologix Motion instructions.

I will attach a document to show the flow process. I ended up converting everything to inches for ease of conversion. The one to note is Diff_Corr = true

It works very well.

Hopefully this explains and clarifies.

David

[OkiePC]

So, one virtual axis to control (master) the physical axis, another virtual axis driven by the physical axis (feeback), and some math to sum the error back into the:

a> actual position of the master virtual axis

or

b> command position of an existing MAM (overwrite with a new MAM at present speed?


Either method is familiar to me, except using the feedback encoder as a virtual slave (geared to the physical).

That seems like a smart way to close the outermost loop, and get the Logix CPU to apply the right UDTs and servo update settings to the external encoders.

Is there a smooth transfer when the error is large?

Thanks for sharing your hard-earned understanding.

Paul

[davidg68124]

It is absolutely smooth. Even when it is correcting itself back into position after overshooting. You can notice that it is not exactly a smooth motion, but it is almost there. I surmise if I decrease the scan time on the correction task, it would almost smooth out completely.

I have added some offset and I get nearly a zero error on my moves.

David

[davidg68124]

I was already going down this path when I decided I wasn't going to get there using only one Master Virtual axis. It was Jim's suggestion to use this method.


I have to admit, this is my first really GOOD experience with timely and helpful Tech Support. I am very pleased with how they came through when I needed it most.

David

[davidg68124]

To answer your question further, I really do not let the error grow that large. I limit the correction to .1" maximum. I just do it enough that it is able to correct accurately.


David