I have a support ticket open where we are working through what might be causing this, but I figured I would open a thread here and see if anyone has had similar issues.
My machine seems to be leaving small gouges in parts, usually occurring at but not limited to feed direction changes, even if i slow the feedrate during said direction changes. All of these examples occurred during very conservative feedrates and low doc.
I have a Pro 4896, Nema 34, 12" Z travel.
These are not only a pain to sand out in 3D carvings, but is impacting how planar my CNC can make flat cuts. On the grand scheme of things they are small gouges, but still quite annoying and impacting my product quality and this error is affecting my production timeline.
So far I have:
Reinstalled mach 4, upgraded to the very latest version
Double checked PC specs, exceeds minimum
Checked all four linear bearings on Y axis, no major areas of resistance but I will be replacing them regardless
Replaced both Y axis drive spindles
Checked z-axis with avid support, looks fine to them
Simulated all related g-code in fusion 360, everything looks good there.
I dont understand when all the gouges are happening but I think the key phrase is ‘at direction change’. This would tell me that something is loose and is able to move when linear momentum changes. Causing a tic toc like a metronome bar at the spindle shaft. That is where I would start looking.
First, I tightened all the bolts to 20ft/lb on my frame and gantry, still got the same issue when I surfaced a test piece of mdf.
Next, I noticed that there was a rhythmic vibration when I put a hand on the spindle as it was cutting along the y, and took another look at the Y axis drives. Initially I over tightened the belt to the point where the gear on the motor was hitting the metal and made an awful sound, but I backed it off and replaced the belt after this. Not sure if this damage occurred before or after this but the intermediate gouges are gone, but the ones on the side persist after the 200ipm right angle direction changes.
Tram appears to be perfect in the middle of my 12" test piece, but gets further out of plane as it gets closer to the edge, seems related to the quick direction changes. The gouges are 0.12mm or so based on what I can measure with calipers, need to check them with a dial indicator to really know.
I see the same kind of machining marks on my 3D wood products. It is coming from direction change. No change of direction is instantaneous and wood is far from homogenous. With that in mind it is undercutting happening because of that change in direction call it “dwell” time and the material never being the exact same density or hardness . Think about a spindle running 18K rpm. In even one second it has turned 300 times, 1/4 second it has turned 75 times. How can that not cause an undercut unless on a very very rigid machine.
The solution may be to create different tool paths that take those direction changes off the part and not on top of the part.
Yeah minimizing direction changes while cutting is definitely a consideration. I may be able to switch to a steel frame when I get my new shop, hopefully that increases rigidity as well and cuts down on these even further.
.12mm may be meeting avid’s accuracy claims, but I have to remeasure with a dial indicator.
The image at the top is a 4 axis rotary parallel, tiny stepover. The one below that is a 3 axis scallop, the only one I have found that steps nicely down the crazy surfaces I feed it. Both of them have transitions on the part, doesn’t seem to matter if I slow down the transition feedrate or force it to slow down extra on angle changes. I have also been meaning to try the new geodesic toolpath option and see if that helps.
I read through the post a couple times to see if this is a new problem, or it has always been like this. If this worked fine before, and this is just new, then keep looking for a mechanical problem.
But if this is something that it always has done, then I would try something that has worked for me. The AVID machines are big, and somewhat flexible compared to the weight of everthing and how hard you can push them with the NEMA 34 motors. I found that running that fast, and with the default 50i/s^2 acceleration values tended to leave little chatter marks on the surface when I had direction changes. At first I just simply ran it slower and that took care of most of it (you are running 4x faster than I do for stuff like this :-)). When I added lasers, I finally realized it was more of the acceleration settings when you did direction changes, so I reduced to around 15i/s^2. I still don’t try to ever run these kinds of files over 100ipm, but I now have very little marks on my carvings, or wiggles on my laser lines.
Hard to tell if this has gotten worse over time. It looks like marks were present back in 2022, when I first wrote this toolpath within a year or so of buying my machine. However they look different and my clamping solution was much less rigid back then (plastic kreg side clamps → steel hold downs)
@Loren spiral could work nicely, the only issue i have with it is the uneven stepdowns on steep areas, causing ridges. Might be able to angle limit the toolpath and switch to contours for those regions.
I don’t have a solution to add, but anecdotally, I had a customer last week call me complaining on a custom broaching tool not cutting to size. When I went in to take a look, we found strange movements in the program. When we went through the program line by line, we found that Fusion had made the error and the machine and tool were simply doing what the machine was instructed by the code. This caused a lightbulb to go off and he immediately checked a finish boring operation that they have been having problems with and, what do you know, they found the same exact issue. They manually corrected the data and the problem(s) were gone. He is working with Fusion to diagnose the problem and they acknowledged the issue was on their end. Point being, have you checked your code at those places for numerical errors? Just curious.
@pketcham Thanks for your input, good point. Yes I have checked these 3d carves in fusion’s simulation, and it also looks good in nc viewer – they both seem to trace the surface perfectly. At this point I think the main contributing factor is vibration, and the best route forward may be the acceleration reduction and toolpath optimization mentioned earlier in this thread to reduce sharp turns. I will also be replacing all wear items on the CNC (bearings, belts, pinion gears) just in case that is a factor.
Did some more testing on scallop vs spiral, and transition feedrates for scallop. No modifications to acceleration yet. The vibration issue seems to persist regardless of transition feedrates. Spiral seems to cut perfectly.
Here is a video of the machine vibration during transitions, it doesn’t seem to happen on every pass, and the “hang” seems to vary in duration:
I’m not familiar with the f360 toolpaths (just a vectric user), but from watching it, I’m assuming it is basically running a circle, then jumping down to the next circle, so you are getting that jerk right at the transition where two or 3 axis are making a quick, short move? If so, I’m definitely putting my money on acceleration. I’d try 10i/s^2 as a starting point on X and Y motors just to compare to see if that is it. You can always fine tune later. I saw this kind of stuff on Vcarves all the time until I got it to take it easy on the gas and brakes.
@jjneeb Yes, on scallop 1 in the above image/video it is jerking as it transitions between passes. It appears to randomly jerk elsewhere also on this toolpath, not entirely on the transitions, shown here:
On scallop 2 it jerks on sharp angle changes on the surface, similar to during the transitions in scallop 1.
I am trying a bunch of things, including replacing the laptop that controls my cnc, the acceleration changes you recommended, rewriting my toolpaths to be as smooth as possible, and changing out the pinion gears. Thank you for your advice!
