Mori MVJR Build Log

More
14 Nov 2011 18:35 #14816 by BigJohnT
Replied by BigJohnT on topic Re:Mori MVJR Build Log
schmidtmotorworks wrote:

I wonder what the symptom of too much or too little acceleration would be.


Too much and your following error at rapid will be a lot not that makes any difference as your not cutting at rapid speeds. Also it might cause and over current trip on your drive. On the few (very few) I have tuned I just kept increasing acceleration until it was fast enough for me and quit right there.

John

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

More
15 Nov 2011 06:34 #14821 by schmidtmotorworks
Thanks to everyone's help I made a screen capture of my HALScope,
I never thought I would get this far.

Here is the program I was running
G20 G40 G90 G17 G91.1
O1 REPEAT[2]
G0 X4
G0 X0
O1 ENDREPEAT
M02

I tried MAX ACCELERATION from 5 up to 50, 50 seemed a lot more like how the machine ran with the old control but the pid error was much worse. I wasn't able to tune it out (but didn't try very long)

Some things stick out.

1. There seems to be sort of a "buzzing" to the motion, not severe though.
2. pid error shows some kind of activity all the time, even sitting still.
3. motor-pos-cmd, I was wondering if the middle of the curve should dip down to the same level as the ends.
4. f-error the scale is 5m/div, but I don't know what "m" is maybe it means 0.005"?
Attachments:

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

More
15 Nov 2011 10:40 #14825 by andypugh
Replied by andypugh on topic Re:Mori MVJR Build Log
schmidtmotorworks wrote:

Thanks to everyone's help I made a screen capture of my HALScope,

The accelleration is rather low, and emc2 is blending-out the moves, which explains why the machine is not travelling all the way. In fact, I am curious what the G61/G64 setting is.

Try putting a pause at the end of each move:
G20 G40 G90 G17 G91.1
O1 REPEAT[2]
G0 X1
G4 P0.5
G0 X0
G5 P0.5
O1 ENDREPEAT
M02

3. motor-pos-cmd, I was wondering if the middle of the curve should dip down to the same level as the ends.

Normally, yes. I admit I have rarely seen EMC2 blending quite so much.
It would be instructive to plot motor-pos-fb on top of motor-pos-cmd.

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

More
15 Nov 2011 10:51 - 15 Nov 2011 10:53 #14827 by PCW
Replied by PCW on topic Re:Mori MVJR Build Log
The large ferror (~7.5 Mills peak) looks to be mainly from not enough (or too much not sure) FF1 (notice that the error is directly proportional to the velocity (slope) of the position plot.)

I would try some small tweaks of FF1 (it should be close to one)

After you are done tuning, a small amount of deadzone should reduce the buzzing

Can you list your PID values, this might help as well
Last edit: 15 Nov 2011 10:53 by PCW.

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

More
16 Nov 2011 06:08 - 16 Nov 2011 06:20 #14854 by schmidtmotorworks
I did all the suggested tests and will post the results over a few posts:
I think i was able to keep all the scales the same today but the are different than yesterday.
The PID error seems to have smoothed out after running it at 100 accel for a while, I don't think I changed anything else..

I have to say it it really great to be able to do this, with my old control, I don't think this was possible.

1. I put a scale on the machine and saw that at MAX ACCELERATION = 20
a program going from 0.0 to 4.0 repeating 4 times it would come about 0.5 short of 4.0 but stop 0.750 short of 0.0.

2. I increased acceleration in steps finally up to 100, now it goes all the way to 4.0 and close to 0.0 but still short maybe 0.3.

This is the INI
[AXIS_0]

TYPE =                  LINEAR
MAX_VELOCITY =          9.842
MAX_ACCELERATION =     100
BACKLASH =              0.000
#FERROR =               .5
#MIN_FERROR =           .5

FERROR =               9999999999
MIN_FERROR =           9999999999


INPUT_SCALE =           -25400

OUTPUT_SCALE =          11.81
OUTPUT_OFFSET =         0.0
MAX_OUTPUT =            11.81

#MIN_LIMIT =             -22
#MAX_LIMIT =               0
MIN_LIMIT =             -22000
MAX_LIMIT =              22000
HOME =                  0
HOME_OFFSET =           0.04
HOME_SEARCH_VEL =       0.50
HOME_LATCH_VEL =        -0.10
HOME_USE_INDEX =        no
HOME_IGNORE_LIMITS =    no

# PID tuning params
#DEADBAND =              0.000015
DEADBAND =              0.000015

P =                     140
I =                     0
D =                     1.8
FF0 =                   0.0005
FF1 =                   1
FF2 =		      	0.00012
BIAS =                  0

Here is the graph with
MAX ACCELERATION =100
FF1=0.9 (set in calibration)
Attachments:
Last edit: 16 Nov 2011 06:20 by schmidtmotorworks.

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

More
16 Nov 2011 06:11 #14855 by schmidtmotorworks
Here is the graph with 0.1 second pauses
MAX ACCELERATION =100
FF1 = 1
Attachments:

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

More
16 Nov 2011 06:12 #14856 by schmidtmotorworks
Here is Here is the graph with
MAX ACCELERATION =100
FF1 = 0.8
Attachments:

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

More
16 Nov 2011 06:13 #14857 by schmidtmotorworks
Here is Here is Here is the graph with
MAX ACCELERATION =100
FF1 = 1.1
Attachments:

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

More
16 Nov 2011 06:14 #14858 by schmidtmotorworks
Here is the graph with
MAX ACCELERATION =100
FF1 = 1.2
Attachments:

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

More
16 Nov 2011 13:05 #14860 by andypugh
Replied by andypugh on topic Re:Mori MVJR Build Log
That is actually looking very good. The reason that you are not hitting zero (except with the pauses) is that EMC2 is not _commanding_ the axes to zero. (G0 moves are heavily blended for maximum speed)
If you use G1 and a high feed rate then you might see better following, and if you put G61 at the beginning of the code I think that both G0 and G1 should touch the 0 and 4 positions.
There is limited discussion of the modes here:
linuxcnc.org/docs/html/common_User_Concepts.html#r1_1_2

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

Moderators: cncbasher
Time to create page: 1.020 seconds
Powered by Kunena Forum