Delayed stop response from rapid move

10 Feb 2019 07:08 #126127 by hatch789
Question for you guys. If I do a long rapid move on X or Y, I get a follow error after about 11 inches. Right now my rapid error is .5" and increasing that to 1" has allowed me about a 22" rapid move before the error. What this suggests is a cumulative increase in the F-ERROR as my machine moves rapidly on X or Y.

Now my encoder scale is 12000 and I'm wondering if I need to increase something so that LinuxCNC it can keep up with my encoders? I have a noticeable delay when I stop a rapid move on my keypad before the servo actually stops moving. The longer the rapid move, the longer the delay before it stops. -Also adding to my cumulative error theory.

So I figured I was missing something small here ...any help would be appreciated.
10 Feb 2019 12:47 #126140 by tommylight
Not properly tuned.
10 Feb 2019 16:27 #126154 by hatch789
It is properly tuned at 30 IPM and my f- error is near zero. But when that puppy rapid moves... That's when I see this. If I try to remove f-error at the rapid speed, won't it make the slower speed moves show f-error below the zero line?
10 Feb 2019 16:49 #126159 by tommylight
You might have it tuned near enough for short rapid moves or long slow moves, but it looks like it needs just a bit more fine tuning to avoid error-ing out on long fast moves.
Had the same issue some 2 weeks ago with a machine i retrofitted 2 years ago, they began using it for long profiles (6 meters) so that issue creped it's ugly head on moves longer than 5 meters. They never had that issue for anything as it never did moves longer than 3 meters.
On long moves that f-error accumulates till it goes past the 1mm limit.
All that was needed to get rid of that issue was add another decimal value to FF1.
11 Feb 2019 01:48 #126186 by hatch789
I dropped my rapid to 80ipm and it fixed the issue. When it was going 100ipm my encoder got behind and then the further it went the bigger the F-Error. So dropping my rapid to 80 instead of 100 was an easy fix. It is still plenty fast enough so I'm cool with that and now even when moving rapidly I have almost no F-Error at all.
13 Feb 2019 23:08 #126416 by andypugh
It sounds to me as though LinuxCNC was commanding moves faster than the axis was physically capable of following.
14 Feb 2019 01:03 #126423 by hatch789
Hi Andy, The servo motors can definitely move a lot faster. I've tested it with the manual voltage commands and my finger on the F2 key. During experiments with this and watching the scope I see the F-Error increase and then after about 1/2 a second second it sharply increases and goes out of sight. I think by having my scale at 12000 it's maybe going beyond what my 7i43 card can count per second or something odd? Doing the math that is 12000 x 5 = 60K pulses per inch. If I'm traveling at 100ipm that's 1.666ips or 100k pulses per second. My parker servo is set at scale 200,000. That one I had to lower even more down to 30 inches per minute and that one can DEFINITELY fly many times faster if I give it 10v. So it's something else I'd like to try to understand so I can correct things.

My quick fix of lowering my rapid jog down to a level that doesn't produce an F-Error seems to be fine but it limits me on how fast I can go.
14 Feb 2019 01:38 #126428 by andypugh
The 7i43 encoder counter can count at MHz. And if it was losing counts then you would lose position, and that seems not to be happening.

Perhaps the PID has a max output setting that is wrong (or a max error) so that it never gets to the full potential output.
14 Feb 2019 03:05 #126429 by hatch789
Hey Andy,

Here are my 2 config files. Everything seems to be working great so I'm confused where this behavior is coming from.

File Attachment:

File Name: Tree_with_...2-13.hal
File Size:11 KB

File Attachment:

File Name: Tree_with_...ders.ini
File Size:3 KB
14 Feb 2019 12:10 #126444 by andypugh
I can't see anything obviously wrong. I assume that when you say that you dropped the rapids to 80 ipm you are sayign that you set all the [AXIS}MAX_VELOCITY to 1.33? If those were already at 1.333 and you only changed the [TRAJ] MAX_LINEAR_VELOCITY then that might explain what you see.
Time to create page: 0.088 seconds
Powered by Kunena Forum