8i20 and bldc, upgrading 2.5 to 2.7

More
23 Jul 2019 23:03 - 24 Jul 2019 00:11 #140364 by johns00056
My machine has been working great for four years.

I am putting together a ,new, complete replacement spare computer.

The computer and Mesa 5i25 are new. Everything else is unchanged.

I decided to take the opportunity to update from 2.5 ubuntu to 2.7 debian 9 stretch.

I had to change several things in the ini and hal files.

I got it so that linuxcnc will start without error but the servo motors have very little torque and I get a following error immediately when I try to move an axis. If i set the following error to a very large number, the axis will move but is extremely weak.




the 8i20.0.0 current will go up to about 4.3 when I turn the motor pulley by hand.(before it errors)
This is about 30 times higher amperage than what I observe when hooked up to the old computer.
the 8i20 bus-voltage looks good at 166 volts

File Attachment:

File Name: 5i25green_...7-23.ini
File Size:14 KB

File Attachment:

File Name: hm2-servog...7-23.hal
File Size:25 KB
Attachments:
Last edit: 24 Jul 2019 00:11 by johns00056.

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

More
23 Jul 2019 23:10 - 23 Jul 2019 23:11 #140365 by PCW
There was a change in the way the driver scales the 8I20 current,
I believe between LinuxCNC 2.5 and 2.6 (in any case a fairly long time ago)

The current used to be scaled 0 to +- 1 for full scale current
Now (since 2.6) its scaled in Amps
Last edit: 23 Jul 2019 23:11 by PCW.

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

More
24 Jul 2019 00:52 #140374 by johns00056
PCW, thanks for the help.
By "the driver" do you mean- hostmot2, bldc, pid , or something else?

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

More
24 Jul 2019 03:52 #140389 by PCW
LinuxCNCs HostMot2 driver

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

More
27 Jul 2019 17:55 #140757 by andypugh
Is this working now?

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

More
28 Jul 2019 12:28 #140796 by johns00056
Andy, thanks for the interest. I got it running.

bldc is working fine, I did not change anything in it.



old computer- MAX_OUTPUT = 4.3 -This number is wrong because it is out of range. It should be 1. (I probably won't change it on the old computer)

new computer- MAX_OUTPUT = 15 -(amps) I changed this to 15 and it helped a little bit, but, the motors were still too weak.

OUTPUT_SCALE = 4.7 ---- This number is probably wrong on both the old and new computers but I am not going to change it. I have spent many days on this problem and the machine works well enough as it is.


I re-tuned the servo motors. (torque mode) I ended up multiplying the previous settings for P by 13 and D by 9.5 and adding some FF2.


old computer 2.5 ubuntu

P = 2500
I = 0
D = 15
FF0 = 0
FF1 = 0
FF2 = 0


new computer 2.7 debian 9 stretch

P = 32500
I = 0
D = 143
FF0 = 0
FF1 = 0
FF2 = 2

Now the motors are really strong, stronger than they have ever been.


The remaining problem is - often (about 25% of the time) when a motor is stationary it will start to hum or buzz a little bit. If I put my hand on the motor pulley ,the humming will stop ,and stay stopped
when I take my hand off.

The machine has been like this for five years. I usually push the e-stop button when the machine is not running to stop the buzzing.

I am concerned that the encoders will wear out or that the encoder couplings will loosen on the motor shaft , eventually.

I tried increasing the dead band , but that just increases the apparent backlash in the axis. If I push the motor pulley to one side of this backlash ,sometimes it will just continue to hum at that location.

If I reduce the P by a very large amount ,the humming will stop but,then the motors just seem too weak.

Maybe I am supposed to measure the motor torque to set the P , or something.

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

More
28 Jul 2019 16:26 #140811 by andypugh
If the previous output was controlling a 0-1 scale and now it is controlling a 0-15A scale then you would need to multiply the gain by (about) 15 as you have found.

You could experiment with using more I and less P.

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

More
29 Jul 2019 22:55 #140929 by johns00056
Thanks for the good explanation of how the scale works. It makes more sense now.

I am still working on the tuning. I will try what you suggest.

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

Moderators: PCWjmelson
Time to create page: 0.114 seconds
Powered by Kunena Forum