Hi guys, I have an OX CNC build (GRBL, 4xDRV8825, 4xNema 23, Dewalt D611 router) with some issues. When cutting circles, the motion seems very jaggy when looking from closeby, the final result has a very rough finish to it. The jags in the final result are not random, you can actually see the stepper motor steps towards max/min on the X and Y axis. Things I have done so far: - verified that my circles are made with arc commands and not a bunch of small vectors - checked belt tension on all axis - checked set screw on the stepper motors - checked wheel tension on the v-rail on all axis - added fans to the drivers to prevent overheating - changed the stepper drivers DRV8825 to cheap TB6560 Changing the drivers made the biggest improvement (but the finish is far from smooth), I’ve added some pictures of my issue and comparison of the drivers, the circles were cut with a diameter of 50mm (or +/- 2 inch). Is this to expect from a belt driven setup? Are there any ways to improve this result? Thanks! Tom
OH, Tomster, i feel your pain, I am struggling with the same issue, and i cant find the answer WHY and nobody is also telling me or helping me to solve this lets hope we will find the issue and how to fix that
What does the finish look like when you cut square and say, hexagonal shapes out? Hard to diagnose things from a single data point. Also are you climb cutting or conventional cutting? CNC should be climb cutting for surface finish, but it's highly likely that your spindle is overpowering your belt's ability to hold it. Try a conventional cutting move so that you maintain constant tension on the belt from in front of the spindle rather than trying to hold onto it from behind. Seems a little less likely to "bounce" in my head, but it may make no difference. More definitive tests would be 1) try a higher spindle speed and a lower feedrate so you get a lower chipload per tooth, this will reduce the cutting force and spindle power at the expense of cycle time, and 2) Try maxed-out spindle speed and high feedrate but at very shallow depths of cut, say 0.5mm or something, and use plunge-pass or (probably better) a helical path. This is basically HSM, where the idea is to improve cycle time on smaller spindles by minimising cutting force. Similar idea to the first option, but a more modern approach. The smoothing between drivers is probably a combination of higher current capacity and smoother signal handling. StepSticks just aren't great, even the better ones. It does seem to congregate around the areas where one axis comes to a stop and reverses- this is a very "low-bandwidth" move and the cutter's tendency to dig in is going to be more obvious than in areas where the step rate tends to even out the "chunks". I'm just now noticing that it looks like you've cut this in shallow passes and the "bumps" are in exactly the same spots, so I'm going back to the idea that it is, in fact, the belts- specifically the low rigidity and low resolution inherent to them (microstepping the motors only helps to a certain extent, since after about 8x microstepping your torque curve drops off a cliff). In all likelihood, you just need to switch to screws. Belts are really for laser/plasma cutters, 3D printers, and maybe lightweight spindles like PCB mills in some cases. Trying to hold onto hundreds of watts of power with a little rubber belt isn't going to produce the effects you're looking for.
