Getting joint 0 following error?

More
29 Mar 2020 23:39 - 29 Mar 2020 23:42 #162042 by Catch22
Thanks PCW I changed the jumpers and set the resolution and it's working now, setup an indicator on all axis and looks good by hand movements. Now do you think I should setup in volocity mode by hooking up the tachs, or should I keep going in torque mode till I get everything working. I put the timing belt back on the x axis and tried jogging it .100 but it moves and shuts down. And doesn't move the proper amount either.
I feel like I'm getting closer all the time, thanks for everyone's help.
Last edit: 29 Mar 2020 23:42 by Catch22.

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

More
29 Mar 2020 23:55 - 29 Mar 2020 23:58 #162044 by PCW
Note that you have about a 50% chance of getting the feedback backwards so it immediately
runs away rather than moving under control. The fix for a runaway axis is to reverse the analog
output polarity in the ini file by changing the sign of the OUTPUT_SCALE parameter
( assuming the scale readout is already in the correct direction )

Also note that the tuning procedure is very different for torque and velocity modes
and decent initial values for velocity mode will not work at all for torque mode
Last edit: 29 Mar 2020 23:58 by PCW.
The following user(s) said Thank You: Catch22

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

More
30 Mar 2020 02:49 #162051 by OT-CNC
Maybe I'm missing something, but I don't think these drives work in torque mode?
I don't think you can run the motors without velocity feedback from the tach to the drive. How are you able to tune the drives without the tach connected?

My workflow was the following before hooking up to Linuxcnc:
1) Set dip switch settings on drives based on manual to match your system
2) Set pots to initial settings in manual and set current limiting to match your system
3) Decouple motor from machine (remove belt)
4) Wire estop/enable for drive testing
5) Make connector for tach input and +/-10v signal (you need a special crimp tool for supplied connector)
6) Wire motor and input power
7) Apply power and expect immediate runaway. If so, reverse tach or motor leads. Dip switch reversal didn't work for me. Maybe works on BL.
8) Adjust pot for drift. So motor is still when no input voltage is present. See manual for temporary grounding of control input .
9) Adjust loop gain pot CW until motor squeals, then back off a turn or so.
10)Test analog input with 1.5v battery CW and CCW motor rotation. If drive faults, over current condition etc is sensed or drive is damaged.
11) If you get an imbalance in response CW vrs CCW as in one direction running rough, decrease reference gain pot.
In my other post, I could not tune these drives with linux as this pot was set full CW and machine ran very rough in one direction. I'm on hold with this step at the moment as my external rpm gage is not working. You want to adjust this pot to get the desired motor rpm at 10v. Drive has 10v out controllable via pot to test but I didn't follow that as I already made the connector. You can later do it with the 7i77. You want to avoid going from 10v to 0 without tapering it down as the motor will stop very hard.
Idea is to set and record your max motor rpm so you have something to go by for configuring the ini.
You may need to redo some of the pot adjustments once the belts are back on.

Now that you know that the drives are working, you can hook them up to linuxcnc and as PCW pointed out, you run a chance of the feedback from linuxcnc causing a runaway. Make sure E-stop works.
The following user(s) said Thank You: Catch22

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

More
30 Mar 2020 21:26 #162145 by Catch22
Ok so I'm running in volocity mode now, everything is working but when I try to jog the x axis .05 it moves and then keeps moving back and forth. Could this be because the motor isn't tuned properly?

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

More
30 Mar 2020 21:29 #162147 by tommylight
Yes.
If you already have the motors/drives tuned, not you need to tune Linuxcnc.
The following user(s) said Thank You: Catch22

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

More
30 Mar 2020 23:15 #162183 by OT-CNC
How much is it moving back and forth? It may just be dithering in between encoder counts. If you have backlash, it will be more noticeable. When you don't command it to move, the position display will reflect that in the last digit bouncing around +-.0004". At least that's how it was on my machine with the old drives. If you have hand wheels, it will look amplified. You didn't say what machine this is.

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

More
31 Mar 2020 00:58 #162191 by Catch22
It is a Delta 3600 and had the Anilam Crusader M control on it.

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

More
31 Mar 2020 22:54 #162314 by Catch22
Ok I've been looking and can't seem to find out how to tune linuxcnc. Do I just change the pid and the ff1 in the ini file, or is there a way to use the config.

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

More
31 Mar 2020 23:10 - 31 Mar 2020 23:12 #162316 by PCW
The easiest way to do this is to use the "Calibrate" menu in Axis in
conjunction with HALScope. This allows you to tune the PID terms
"live" and write the changed parameters back to your ini file.

Here is a basic LinuxCNC velocity mode tuning tutorial:
gnipsel.com/linuxcnc/tuning/servo.html
Last edit: 31 Mar 2020 23:12 by PCW.
The following user(s) said Thank You: Catch22

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

More
01 Apr 2020 13:39 #162348 by OT-CNC
PCW, I'm following this as I will be back to tuning again on my machine in a few days. Do you have any additional pointers for the PID tuning process for those of us using linear encoders with relatively coarse resolution?
Also, how's the deadband figured into this (v2.8)? I also have .0004" table travel between encoder counts. Is it best to set this to 0 during tuning?
The following user(s) said Thank You: Catch22

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

Time to create page: 0.245 seconds
Powered by Kunena Forum