Servo Tuning FF

More
17 Jan 2012 02:52 #16832 by centaur
Hi Andy - just read this and it is a good starting point. I am totally new to EMC2 and feeling my way around. Please tell me just where these 'FF' values are set and how to get there. I am using Mesa 7i43 (small) with 7i33.
Thanx

Please Log in or Create an account to join the conversation.

More
17 Jan 2012 10:24 #16840 by andypugh
centaur wrote:

Please tell me just where these 'FF' values are set and how to get there. I am using Mesa 7i43 (small) with 7i33.

They are set in the HAL file, in exactly the same was as the PGain and IGain.
The actual values are probably read into the HAL file from the INI file, but putting them in the INI file will achieve nothing without a reference and a setp statement in the INI file.

You probably need to understand a bit about HAL pins and parameters.
linuxcnc.org/docview/2.5/html/hal/basic_hal.html

The pins and parameters of the PID component are described here:
linuxcnc.org/docview/2.5/html/man/man9/pid.9.html

Please Log in or Create an account to join the conversation.

More
29 Feb 2012 11:32 - 29 Feb 2012 12:21 #18195 by Fremder
Replied by Fremder on topic Re:Servo Tuning PID and FFn
Hi all,

verry good thread ...

to make SIMulated and optimicing my PID i have on my PC (office) the servo_sim.ini ...

what can i do to work simulated with the PIDs ? actuall i becomes ONLY the f.error joint 0 following error .
(he goes LINEAR from to to max-ferror ..)

i am stupid ? and the SIM its NOT for simulated a servo ? *g*
thanks for help.

PS: what its the POINT - i found NO samples for optimicing the PID and Co ?
PStart = 0
Pn = 0.5
Pmax = 5
.... then combine with the <I> <D> <F1> <F2> ? automaticly with ala pyhton ?

PPs: i goes to update the "office-pc" with the:
deb buildbot.linuxcnc.org/ lucid v2.5_branch-sim
deb-src buildbot.linuxcnc.org/ lucid v2.5_branch-sim
i hope this its the right way :-)
Last edit: 29 Feb 2012 12:21 by Fremder.

Please Log in or Create an account to join the conversation.

More
29 Feb 2012 17:57 #18211 by jmelson
Replied by jmelson on topic Re:Servo Tuning PID and FFn
Fremder wrote:

and the SIM its NOT for simulated a servo ? *g*

No, sim is simulated real time, not simulated servo.
MANY years ago, the old EMC had a simulated motor/machine,
but it was not terribly realistic. Unless you make detailed
measurements to the motors electrical and mechanical
characteristics and all the springiness, friction and masses
of the machine, simulation will be totally pointless, anyway.

Trying to develop tuning parameters on a simulation is
only of benefit for NASA, where you can't operate the
device in its natural environment.

Jon

Please Log in or Create an account to join the conversation.

More
12 Jul 2012 12:53 - 12 Jul 2012 12:56 #21822 by LAIR82
Replied by LAIR82 on topic Re:Servo Tuning PID and FFn
After reading these posts about servo tuning, I feel pretty good about tuning both servos on my turning center, but before I try that I want to make sure that I am at least close on my values in my "ini" file for my axis, before i throw power to the servos and have a major crash.

Following is a copy of my INI



any help would be greatly apprecieated.

Thanks

Rick

File Attachment:

File Name: X__INI.txt
File Size:1 KB
Attachments:
Last edit: 12 Jul 2012 12:56 by LAIR82.

Please Log in or Create an account to join the conversation.

More
12 Jul 2012 13:47 - 12 Jul 2012 13:49 #21824 by BigJohnT
I usually disconnect the connection if possible to do so without loosing the control feedback to the axis ball screw just to make sure my wiring etc is ok and no run away condition exists. If you can jog the servo a bit then connect the servo back up and carry on with your tuning.

We now have a Servo Tuning area... I'd move this post but seems like you can't move a post that is older than the category.

John
Last edit: 12 Jul 2012 13:49 by BigJohnT.

Please Log in or Create an account to join the conversation.

More
12 Jul 2012 14:05 - 12 Jul 2012 14:06 #21826 by PCW
Replied by PCW on topic Re:Servo Tuning PID and FFn
I agree that for first time servo power-up you should disconnect the motor from the mechanics, This is because the first time you connect things you have a better than 50% chance of a runaway.

So _expect_ a runaway and verify that you can power things down or disable the servo quickly before starting. As John says , once you have feedback working and can jog things, then reconnect the motors to the mechanics for tuning.

I would also add that you should verify the drive enable circuitry before starting.
you want a ferror to instantly disable the drives. You will need to have larger values of FERROR and MIN_FERROR as you start to tune
Last edit: 12 Jul 2012 14:06 by PCW.

Please Log in or Create an account to join the conversation.

More
12 Jul 2012 14:13 #21827 by LAIR82
Replied by LAIR82 on topic Re:Servo Tuning PID and FFn
Sounds like it could get real interesting then. I'm working on a cincinatti millacron 10cc from the 80's. The resolvers are mounted to the ball screws, which are belt driven from the motors, with no coupling in the ballscrew shaft to unhook to allow this.

Thanks guys


Rick

Please Log in or Create an account to join the conversation.

More
12 Jul 2012 14:22 #21828 by BigJohnT
In your case your more or less stuck with a live test so make sure your E-Stop works and is handy and manually move the axis to the center of travel before powering up. As Peter said double check your drive enable.

When I tested my CHNC I made a simple box with a couple of switches, one for enable, and one for jog and the jog input was from a variable voltage power supply so I set it at like 1/10 of a volt just to make sure jog plus and jog minus went the correct way. I did find out that with the drive enabled and no velocity input there was some drift... YIKES. So the moral of that story is don't jumper out the enable from the rear of the machine unless you can see the axis.

John

Please Log in or Create an account to join the conversation.

More
12 Jul 2012 14:35 #21829 by PCW
Replied by PCW on topic Re:Servo Tuning PID and FFn
It is possible to verify the feedback direction first. (and assuming analog servos here)

The very first thing is to verify that the position feedback works, is scaled properly and has the proper direction (prehaps by hand moving the motor shafts with the drives disabled and watching the Axis DRO)

to check feedback direction, you first determine which voltage input polarity drives the axis which direction This can be done with a custom hal file and manually setp-ing the analog output voltage or simply with a 1.5v battery connected to the drives analog input pins

Once you know which input polarity drives the axis in which direction write this down

Now you disable the servo drives(remove power), set a large FERROR and MIN_FERROR and start LinuxCNC with PID setup with a small P and all other terms = 0

Now when you hand crank the motor shaft the analog output voltage should change such that it would oppose the motion. That is if a positive voltage makes the axis go right, moving the axis by hand to the right should generate a negative voltage on the analog out pins, and moving to the left of the starting point should generate a positive voltage. If these are reversed you will need to reverse the PID output polarity

The other thing to check here is that the analog voltage varies smoothly from -10V to +10V as you rotare the motor shaft

Please Log in or Create an account to join the conversation.

Time to create page: 0.123 seconds
Powered by Kunena Forum