Hello Open Builds, I began building a large Open Builds CNC machine a while ago and I almost have it all working, but I'm hung up on this one issue. The z-axis seems to randomly shift upward, increasing the distance from the build at times when it should be maintaining its height. The spindle is quite heavy so it seems especially odd that it shifts up. It shifts a mm or two at a time per every ~250mm which is enough to ruin any of my builds. None of the other axes seem to have this problem. At first I thought it might be electromagnetic interference (from the 2.2kw spindle) but the wires are shielded. The motor makes a bit of a crunching noise when it happens. I'm running a tinyg2 controller (Arduino due with 2 GRBL shields). Perhaps its some firmware issue. The current gcode involves, row by row, shaving off layers. (Each pass in the x-axis is a row) I've tried testing the CNC machine using Coolterm, but whether or not the spindle is on it does NOT shift in the z-axis. This does not fix the problem as I can not use Coolterm to run the build. I'm uploading the gcode using Chilipeppr. Maybe that is causing the problem. I'll do some more testing, but if anyone has any other ideas please let the thread know. I ended up doing a work around by changing the gcode to use less passes and take less time overall. Instead of half a mm per pass in wood it now cuts at the final depth (.5 inches ~12mm). Thank you for your help!
slow down the Z, ie change the max rate and the acceleration settings to lower values, say 1/2 or 3/4 of what they are now, and run a file that did it before. if it still does it, then there is probably a loose grub screw (allowing a pulley to rotate) or a sticky spot on the rails/leadscrew that is causing extra resistance 'sometimes'. an alternate is to reduce the microstepping setting, ie if it is now 1/16, set it to 1/8 (and change the steps/mm setting to match)! why? stepper motors get weaker the faster they run, spindle is heavy therefore run slower. microstepping loses power, so lower settings are more powerful, but you don't want single stepping either as that suffers from vibration issues.
This is a good solution for a z-axis that is slipping and falling down due to the weight of the spindle. I should have clarified that in this case the z-axis is moving up, increasing the distance from the build. It also does this at times when it should be maintaining its height. Thank you for the response though.
AND for when the Z axis is missing steps when pushing down into the work (-: does your gcode plunge straight down into the work? if so then you should try a ramped entry. which points to interference... but prove it by reducing the acceleration because that is much easier than separating all the wires (-:
This is interesting, just can back to the forums with a Z-axis issue, not sure if related. My Z-axis (on large-ish double-belted Ox with high torque NEMA23 for all motors) gets slightly deeper as Y increases, and shallower again as Y decreases. The slope is perhaps slightly-less than 1%. It is inconsistent, sometimes hardly appearing at all, at other times approaching 1%. I made a jig and very carefully checked the X-crossbar-to-perf-board distance across the entire perf board. Could this also be a Z speed issue? I am using very conservative speeds. It doesn't seem to matter how many passes I am make, by the way. And all I'm cutting is 5mm plywood with 6mm end mill, Dewalt tool.
firstly, nothing is flat until you make it flat. (-: I would attach some 6mm MDF to the base and then mill it flat. if you draw lines on it with a sharpy and then mill say 0.25mm deep you will quickly see where the highs and lows are. now you can compare that 'flat' to your yardstick and adjust the frame to get the best results. the perfboard will follow the frame, but maybe not instantly, depends on the thickness. note that if you move the frame, even half an inch, it will no longer be flat since it will conform to the table it is standing on, and tables are never flat, and if made of wood, will change shape.... secondly. perfboard = some form of wood. wood changes shape depending on the weather. put some MDF on there and mill it flat, for those jobs that matter. or make a probe, probe the surface and let the software correct for the warp. bCNC can do this with a GRBL controller.
Of course this all makes sense, but my question would be that given the perf board is 3/4" MDF, why not simply mill the surface of the whole perf board? I had always thought this might be necessary - your comments suggest it is. Thanks.
I ended up doing a work around by changing the gcode to use less passes and take less time overall. Instead of half a mm per pass in wood it now cuts at the final depth (.5 inches ~12mm). This solution won't work for materials much stronger than wood, or for anything that has to be precision, even with a $5000 CNC machine.
