Multi Axis Slow Feedrate

Hi everyone. I know this topic has been covered before, but I would like a bit of advice specific to my machine as I'm not that well versed in LinuxCNC.

I've built a 5 axis rotating nutating (head-head) type CNC and it's been running well for the past 1.5-2 years. However, I'm getting quite slow feedrates on certain multi axis operations and it's becoming increasingly frustrating.

As an example, this image shows a test file where the X and Y axes barely move, the Z and A axes don't move, and it basically just rotates the C axis. At a specified feedrate of 4000mm/min, it's only cutting at about 600mm/min. My max velocity for linear axes is 25,000mm/min, and angular is 4500deg/min. Acceleration is relatively high, and 4000mm/min should be achievable.

 

I'm using Fusion for the CAM, and the above test file is using swarf machining. This can only be output as a bunch of small lines, so I've gone as coarse as possible with the tolerance and that's set at 0.1mm, and I'm using P0.1 Q0.1 for smoothing. And I'm using TCP.

Correct me if I'm wrong, but is the slowdown in the feedrate due to the 1 line lookahead in multi axis?

I have a separate LinuxCNC config setup to run my machine as a simulation, and it uses a different ini file to my machines so I can test files prior to running them. With this ini file I've bumped up the max velocity a bit, and really bumped up the acceleration, and I've been able to get the required 4000mm/min. However, these velocity and acceleration speeds are very unrealistic for my physical machine.

So my question is, is this just how LinuxCNC is? Or am I doing something wrong? Or is there a workaround for this issue? I've also tried using inverse time, but it gives me the same outcome as programmed feedrate. Let me know if there are any files I should post (ini, etc). Thanks.

Here's a couple of photos of my machine.

 

 

Thanks Tommy. I had tried G93 Inverse Time, but didn't see any difference to G94 Programmed Feed.

I went back and did a bit more testing (with G93 and G94). The only thing I've found to work (for both G93 and G94) is to up the velocity and acceleration speeds to unachievable levels, or lengthen the small line segments that make up the toolpath. However, this will result in a more inaccurate part. And I still can't get close to the required feedrate while keeping the short lines to a decent length.

Might be a limitation in advanced planning of non-XYZ moves? My understanding is that for non-XYZ coordinated moves, there's no lookahead, so presumably it has to come to a complete stop after each segment.

forum.linuxcnc.org/32-documents/53351-pl...nding-vectors#306151

I'm sure I saw this in the official docs somewhere, but I can't find it at the moment.

ETA: Missed that you already saw that... Assuming it works as expected with single-axis moves, it sure seems like the above is what you're running into.

Thanks Andy. The single-line lookahead is the conclusion I've come to also.

Any advice on where to go from here?

Is there an alternative trajectory planner that could be better suited for multi axis? Is PathPilot an option, and can it be modified for more that 4 axes? What would it take to merge the Tormach TP into LinuxCNC? Or is there something other than LinuxCNC that would be better for simultaneous multi axis machining?

I did have a look through this a couple of days ago.

To me it seems like a very worthwhile use of time to get this up and running. I think it would be very valuable to anyone wanting to do multi axis work.

I'm unfortunately not that experienced with coding. However, this would be a huge benefit for me to get working, so I'm going to spend some more time going through it.