Jerk motion, what am I doing wrong?

[rmat]

Hello everybody,
Im Matej and Im new here. I have the controller for two weeks and now that I figured it out and set up my new machine up, so everything runs smooth, it was time to cut some metal.
In short, Im using Fusion 360 to generate Gcode and I have servos on my machine. Machine is router, but build as more as a gantry mill, so very stiff costruction, point of making was aimed especially at high speed machining.
So up to now everything as planned, but since my first cut, nothing but dissapointment. Machine runs very "jerky" to say the least, and I have tried everything with messing with parameters on CV.
Also, Ive read that uccnc has problems with arcs, Ive posted files with poits only, and maybe just a little improvement, but still no luck getting smooth motion.
Since I have had some expirience with motion controllers and also MACH, I cannot understand what is the deal here. Im probably doing something wrong, but cannot figure it out.
So, please, please help. I would already need this machine cutting, so Im willing to try just about anything.

Im pasting part of the, code, if anybody is willing to look at it I would highly appriciate it.

(1001)
(T6 D=6. CR=0. - ZMIN=-10.5 - flat end mill)
(T10 D=20. CR=0.5 - ZMIN=-0.5 - bullnose end mill)
(T11 D=6. CR=0. TAPER=45deg - ZMIN=-1.7 - chamfer mill)
N10 G90
N11 G0 G53 Z0.

(Adaptive1)
N12 T6 M6 (flat end mill D=6.)
N13 S15916 M3
N14 G64
N15 G55
N16 M7
N17 G43 H6
N18 G0 X-4.34 Y45.382
N19 Z15.
N20 Z-9.099
N21 G1 Z-9.699 F3000
N22 X-4.326 Y45.357 Z-9.884
N23 X-4.283 Y45.283 Z-10.052
N24 X-4.216 Y45.169 Z-10.184
N25 X-4.132 Y45.024 Z-10.27
N26 X-4.039 Y44.864 Z-10.299
N27 X-3.451 Y43.932
N28 X-2.797 Y43.045
N29 X-2.082 Y42.207
N30 X-2.053 Y42.178 F2700
N31 X-0.763 Y42.313 F3000
N32 X0.532 Y42.373
N33 X3.126 Y42.413
N34 X5.721 Y42.416
N35 X8.315 Y42.42
N36 X45.284
N37 X45.932 Y42.402
N38 X46.557 Y42.227
N39 X46.803 Y42.109
N40 X47.061 Y42.017
N41 X47.446 Y41.849
N42 X47.808 Y41.635
N43 X48.141 Y41.378
N44 X48.43 Y41.078
N45 X48.68 Y40.745
N46 X48.862 Y40.416
N47 X48.983 Y40.152
N48 X49.072 Y39.876
N49 X49.157 Y39.516
N50 X49.174 Y39.336
N51 X49.27 Y38.695
N52 X49.367 Y37.076
N53 X49.399 Y35.455
N54 X49.418 Y28.97
N55 X49.419 Y1.73
N56 X49.404 Y1.081
N57 X49.233 Y0.456
N58 X49.076 Y0.116
N59 X48.954 Y-0.237
N60 X48.862 Y-0.416
N61 X48.68 Y-0.744
N62 X48.425 Y-1.085
N63 X48.127 Y-1.388
N64 X47.79 Y-1.648
N65 X47.421 Y-1.858
N66 X47.157 Y-1.981
N67 X46.881 Y-2.073
N68 X46.521 Y-2.153
N69 X45.603 Y-2.268
N70 X44.681 Y-2.341
N71 X43.757 Y-2.374
N72 X41.163 Y-2.414
N73 X38.569 Y-2.416
N74 X35.974 Y-2.42

...and so on....

Best regards, Matej

[cncdrive]

UCCNC has path tolerance control just like industrial controllers, mahc3 does not. Tune the constant velocity parameters properly for your job.
Read the UCCNC manual to understand how the parameters work.

[rmat]

Thank you for reply.
I understand about mach3, but Eding has it also, and runs smooth and both accurate.
Ive read trough manual a couple hundred times just in case if I missed something and I think I understand what id does and how, but even if I put on corner error max some crazy number like 100 it doesent blend my parameters in one smooth motion. I understand that it cant keep full speed all the time, so im okay if it brakes, but this is running so if you put your hand on machine, you can feel every point. It feels as the machine wants to run in semi CV mode??Unfortunately this shows on workpiece also.

Now, Ive seen some youtube videos of guys cutting with this controller, so this is why im asking what am I doing wrong here. Please have a little paitience with me. I like the controller, so I wanna figure it out.

Best regards, Matej

[cncdrive]

It runs smooth if tuned properly. You can rune it to very tight tolerances and then if your g-code movements have jerks then the controller will follow the jerks closely, so there will be jerk.
If you tune it loose then the tolerance will be high making high tolerance path errors, but the motion is then more smoothed out.
You should find the optimum with the settings for your machine and your jobs. The limitation is the earth physics, unfortunately you cannot override that with any software or hardware.

Also make sure that you are in G64 constant velocity mode and not in G61 exact stop.

Yes, there are fortunately many thousands of people using these controllers, you are correct about that.

[ger21]

Try increasing the linear unify and linear max settings.
What are your acceleration settings? The higher the accel, the smoother it will run.

[rmat]

It runs smooth if tuned properly. You can rune it to very tight tolerances and then if your g-code movements have jerks then the controller will follow the jerks closely, so there will be jerk.
If you tune it loose then the tolerance will be high making high tolerance path errors, but the motion is then more smoothed out.
You should find the optimum with the settings for your machine and your jobs. The limitation is the earth physics, unfortunately you cannot override that with any software or hardware.

Thanks for reply, I was hoping there was some more straightforward solution for my situation. Im a little surprized, since this is working with high speeds toolpath very simmilar to an old ISEL controller we have on our old machine. It doesent want to run smoothly no matter what if I use these modern toolpaths.

Also make sure that you are in G64 constant velocity mode and not in G61 exact stop.

Im aware of that and postprocessor outputs correctly G64 for CV and G61 for drilling cycles ect, so this is taken care of.

Try increasing the linear unify and linear max settings.
What are your acceleration settings? The higher the accel, the smoother it will run.

I have played with this also, but it gives mixed results, alhough I increased linear addition also, then it just skips lines, or it acts so.

My accel is 600 at the moment, linear max 0.03, Linear addition 1, Linear unify 2, Corner max 100!!, Arc radius 0.1

I personally would stick with acceleration on 600, becaose if I increase further I can get into a "servo effect" troubles, so I need to find some sweet point as cncdrive suggested.....
Any ideas with what numbers would I start with these parameters?

Best regards, Matej

[Robertspark]

At 3000 unit/min feedrate and 600 units/s/s try a corner error max of 4.2 and it's as good as it will get.

r = v^2/acc

The acceleration profile is trapezoidal, not S curve at present

How many steps per unit is each of your axis set at?

[rmat]

At 3000 unit/min feedrate and 600 units/s/s try a corner error max of 4.2 and it's as good as it will get.

r = v^2/acc

I figured it out, it doesent do anything from 5 up, just desperate I guess, so willing to try anything.

The acceleration profile is trapezoidal, not S curve at present

How many steps per unit is each of your axis set at?

Im aware of trapezodial, in fact most of planners are, at least those in reach.

Im at 400steps/min, so theoretical 0.0025mm/step.

[Robertspark]

Im at 400steps/mm, so 0.0025mm/step.

Ask yourself "do you need to have a step resolution of 0.0025mm?"

Dont shoot messanger.... just something to consider.



Jerk / Jolt is defined as units / s / s / s (units/s^3)
With a stepper if you know the step size and the acceleration you can quantify the jolt that the machine will be creating and its components experiencing.
Given the acceleration profile is trapezoidal the acceleration goes (near) instantly from 0 mm/s/s to 300 mm/s/s and then acceleration is constant until ~ the feedrate is achieved and then the axis acceleration decelerates to 0 m/s/s and maintains the feedrate.

Jolt / jerk is worked out as follows

Jerk = accel / [ SQRT ( 2 x step distance / acceleration) ] m/s/s/s



So in your case at 300 mm/s/s and 0.0025mm step resolution

Your axis has to accelerate from rest to 300mm/s/s in 0.0025mm, and the jolt / jerk experienced is ~73.5 m/s ^3

to lower the jerk / jolt you can increase the step size and or lower the acceleration

I don't believe I've ever seen reference to an acceptable level of jerk / jolt
The closest I got was regarding elevators / lifts : https://en.wikipedia.org/wiki/Jerk_(physics)

It is reported[5] that most passengers rate a vertical jerk of 2.0 m/s3 in a lift ride as acceptable, 6.0 m/s3 as intolerable and for a hospital environment 0.7 m/s3 is suggested.

I have a relatively lightweight plasma gantry with large width belts (big teeth + wide) direct driven by largish steppers and have much higher jolt / jerk (~520m/s/s/s) but it is all to do with what is acceptable to you.... plasma can use course steps and requires high acceleration and my finish is less important to me than the shape being close to what I require as once its mig welded & angle ground it makes little difference. The effect of jolt in one plane (Y) is worse than in the other (X) because the X only has the carriage and the Y has the gantry weight.

If you doubled your step size to 0.005mm (still less than half the average human hair thickness), you have a 30% reduction in jolt / jerk

If you half the acceleration to 150mm/s/s you'll get a ~65% reduction in jolt / jerk.

[rmat]

Ask yourself "do you need to have a step resolution of 0.0025mm?"

Dont shoot messanger.... just something to consider.

Not really, it was just nice number, and since we can have onough kernel to spare, I used it.

Jerk / Jolt is defined as units / s / s / s (units/s^3)
With a stepper if you know the step size and the acceleration you can quantify the jolt that the machine will be creating and its components experiencing.
Given the acceleration profile is trapezoidal the acceleration goes (near) instantly from 0 mm/s/s to 300 mm/s/s and then acceleration is constant until ~ the feedrate is achieved and then the axis acceleration decelerates to 0 m/s/s and maintains the feedrate.

Jolt / jerk is worked out as follows

Jerk = accel / [ SQRT ( 2 x step distance / acceleration) ] m/s/s/s



So in your case at 300 mm/s/s and 0.0025mm step resolution

Your axis has to accelerate from rest to 300mm/s/s in 0.0025mm, and the jolt / jerk experienced is ~73.5 m/s ^3

to lower the jerk / jolt you can increase the step size and or lower the acceleration

I don't believe I've ever seen reference to an acceptable level of jerk / jolt
The closest I got was regarding elevators / lifts : https://en.wikipedia.org/wiki/Jerk_(physics)

It is reported[5] that most passengers rate a vertical jerk of 2.0 m/s3 in a lift ride as acceptable, 6.0 m/s3 as intolerable and for a hospital environment 0.7 m/s3 is suggested.

I have a relatively lightweight plasma gantry with large width belts (big teeth + wide) direct driven by largish steppers and have much higher jolt / jerk (~520m/s/s/s) but it is all to do with what is acceptable to you.... plasma can use course steps and requires high acceleration and my finish is less important to me than the shape being close to what I require as once its mig welded & angle ground it makes little difference. The effect of jolt in one plane (Y) is worse than in the other (X) because the X only has the carriage and the Y has the gantry weight.

If you doubled your step size to 0.005mm (still less than half the average human hair thickness), you have a 30% reduction in jolt / jerk

If you half the acceleration to 150mm/s/s you'll get a ~65% reduction in jolt / jerk.

Will try today and report results. Many many thanks for taking the time to explain this.

Best regards, Matej