Hi there, I have a Stronghold pro from Ratrig (Rat Rig StrongHold PRO CNC) with rather solid (to my understanding) rails and ball screw transmission. I noticed a consistent imprecision when doing long cuts (90 minutes for example), mostly on the Y axis, that has two motors. The frequency of the imprecision is consistent, but the error is not the same every time, which seems to clear the software part of the problem, and hint towards a more mechanical issue. I see few possibilities of improvements and hints of possible issues: - the setup itself: I made a grid on my spoilboard, in order to have physical stoppers that would help when turning the piece and keep a square setup, in relationship to the axis of cutting. But I did notice that the error doesn't come necessarily from flipping the pieces, since it comes before sometimes, on the first face itself. I did notice I had to be very careful when making this grid (follow the X axis, then the Y axis) because I had crazy big errors when I just had the programmed toolpath doing its thing (making the most efficient cut, but not the most precise, mixing X and Y axis sequences). Making the spoilboard showed the limits of the machine in terms of precision, but eventually improve the systematic character of the process. Nonetheless, it could be improved (stronger pins, more stable and thick spoilboard). - the Y axis is by far the most inconsistent, and shows some jamming when jogging to fast (over 3950mm/min), especially close to the motors. It feels like they are not perfectly synchronised. - the way I setup the zero each time: the system to zero the spindle feels slightly imprecise. It's the Openbuilds probe, and depending on the way the bit touches the probe, it could be slightly off by a few decimals of a milimeters, as the bits are not perfect cylinders, and if they touch in a different place of the flute, it might give a different result. How can I improve this? - kind of a beginner question, but DO I HAVE TO home the machine between each piece I do (every time I restart the same toolpath)? I'm doing 6 pieces at a time, zeroing each piece every time. - finally, a friend who works with CNCs for a long time pointed at the limits of the type of motors I have (Nema 23 Stepper Motor - High Torque - 1.8degree/step, 345oz-in), insisting that these are less precise and slower than servo motors. What motors should I look at to increase the precision? What driver would they then use? I guess I should then change the whole setup Motors/controller? I have an Openbuilds Blackbox Controller. Looking forward to reading you all!
Probing - don't use a bit to probe XY - use an accurately measured rod - actually probe XYZ, switch to the bit then probe Z only. Motors stalling - small inaccuracies building up - slow down. You might achieve 4000 mm/min but probably need to let parts of the machine "bed in" first. Post a back up of your grbl settings here so we can recommend changes to get you going. You have a pretty good hobby level machine - it's not going to compete with an industrial vertical mill but, carefully built, tuned and sensibly driven, should be capable of +/- 0.1 mm or better. Talking of tuning - have you calibrated your steps/mm? Openbuilds Control has a wizard for that, but calibrate over the longest distance you can accurately measure, not the default 100 mm. You do NOT need to rehome the machine every time you run a job as long as you have solved the other problems - you have an "open loop" control system. The only way the controller knows where the machine is, is by keeping track of the steps it has ordered the stepper motors to do. If some of those steps don't actually happen (eg because you are trying to get move too fast) then the controller will "think" the machine is somewhere it isn't. Alex.
