forum.cncdrive.com

Hi, it's driving me nuts cutting 3D wing mould shapes, the way the movement slows down dramatically in the more curved area. When using Mach3, I didn't have this problem. I've asked about this sort of thing previously when I had jerky movement in 2D curves, and I did a lot of experimentation, and got the tolerances sorted out to give a good compromise between speed and accuracy in tight curves.

The settings I'm using are:
MCE 0.1
MLE 0.05
MLAL 10
MLUL 25

But it's driving me nuts now with these slowed-down parallel finishing cuts which are repeated thousands of times so it really slows the process down. I'll try to upload a video of the problem.

It has been described several times, but I describe it again:

MAch3 has no path deviation control, it cuts off path details, round corners as it feels it will be fine.
With the UCCNC you have control over the path tolerances, the settings are in the General settings, the Constant velocity parameters.
For information about how the parameters work please read the users manual.

If the tolerance is set low then ofcourse the controller must slow down to follow the path.
MAch3 can do the job fast because you cannot set the path tolerance, so it can cut off from the path whatever size it wants in order to maintain fast feedrates.

Also MAch3 has a trajectory bug when it overrides the acceleration at least 2 times the value you set for the axes.
This means that you can set the acceleration at least 2 times in the UCCNC than what you set in MAch3 and it will still work reliably with your machine if helf the values worked reliably with MAch3.
This will make the overal machining speed faster than in Mach3 even with tight tolerances.

Thanks, yes, I understand that. And I do appreciate the higher accuracy of UCCNC, which is important to me for cutting highly developed, high performance model aircraft airfoils. But ... to clarify my question ... is there anything I can do to increase the speed in the arc sections? I'd be OK with it slowing down in tight curves like the actual leading edge radius, but it's dramatically slowing down in sections where the curve is not tight at all and such a slow speed is not necessary.

After posting this yesterday, I created a subsection of the wing model to the side and ran a whole bunch of air cuts, changing parameters each time to try to get improvement. I used a stopwatch to time 10 passes (5 back and forth). It's only a small tailplane (horizontal stabilizer) around 11cm across. I tried loosening up the MCE and MLE settings. I loosened up the tolerances in the CAM program. I increased all the feedrate parameters (approach, engage, etc). I tried, one at a time, everything I could think of that might help, but no matter what, it still took the same 33 seconds for the 10 passes. Nothing helped. I'm an amateur so perhaps there's something else I'm missing.

For most things I can live with it but these wing surface finishing cuts are at a fine stepover so even for a small tailplane like this there are more than 2000 passes. So the slow moves really extend the cutting time dramatically, plus the machine does a lot more speed up/ slow down on each pass as well.

If you can suggest anything at all that I can do to increase the pass speed, please let me know.

maybe we could have an acceleration and accuracy doesnt matter option?
i would like to see how the chinese dsp controllers do it- as they dont seem to ever slow down. (not checked their accuracy though)

HI AvB I have the opposite experience I had the issues described in Mach 3 with ESS and moved to UCCNC and UC300ETH and have not had any of the slowdowns you describe.

I'd be interested to test your code or part of it out on my DIY machine to see if it does the same thing?

No worries, I can send you a file. But it might be hard to compare apples with apples unless we know the settings are comparable. What tolerances are you setting in your CAM? (Mine is usually 0.01 or less). And also what CV settings are you using (MCE, MLE, MLAL, MLUL)? And have you run tests to check for accuracy and sharpness?

Two years ago I ran a lot of tests to get the CV settings as loose as possible without losing accuracy/ sharpness, as reported in this thread: viewtopic.php?f=2&t=1750

Hi AvB,

Could you post a section of this code, and a settings dump from your machine? Are these G18/G19 G2/G3 sequences? How are you generating the gcode?
I read and appreciate your prior post on 2D curve speed optimization, and have been looking into some similar issues myself.

I know you had previously mentioned you were using fairly low accel values and that Z was lowest. Have to wonder if that's related to the present issue.

Have you been able to catch F_act during these operations?

I've now switched to Mach4 and don't have the slow movement issues any more. It's working well for me.

Mach4 does not have path deviation control, so it erases path details as it feels.
It might be good though for reliefs which does not require much precision.