I tried to calibrate my cnc yesterday. I startad with the x axis, i attached a digital calipers, made sure that it would measure correctly and not moving. I then moved the x 5mm in 5 steps (1mm steps) This is what the program say before its calibrated. The cnc movement is about 2mm (I'll come back to this later). So I measure this five 1 software mm step. Reading the calipers and divides it by 5 so to get a average say was 2.30mm. Now to the above 2 mm, the first measurement I took was 2mm with calipers. So after getting the average I was confused about the measurement so I started to measure several software steps of 1mm. I was not pleased with my findings, the range was from 1.97-2.48mm. Sadly I cant find the precision of my caliper but say has +-0.05mm at most. Is it normally this fault measuring (sorry forgotten the right word for it) How can I improve my measurement to calibrate my cnc. Is a digital readout scale a good thing to use and can it be used as an encoder in bcnc? I did fasten down my caliper on the top of the x axis, made sure that the thing to measure hole depth hited the CNC x axis straight on. Moved the cnc to push it in.
Hmmm... Inaccurate? Shouldn't be. Is it screw or belt driven? Mag/glass scale is doable, but grbl/bCNC can't use it. You'd have to switch to LinuxCNC, which is pretty complex, and I don't think you'd gain much accuracy once you have the machine tightened up and properly calibrated. I can't quite follow that calibration method, but something about it doesn't seem right. We'd have to know more about the machine to offer too much more though.
If you get inconsistent measurements between moves, check that the pulley isn't slipping on the motor shaft. A drop of locktite in between helps immensely, and make sure the setscrew is on the flat of the motor shaft.
I'm much better to read English than to write, its quiet annoying. I read alot of technical manuals and specifications but the word I seek just skipp my mind at the time. Google translate say its margin of error but I'm sure I have read a word with the same meaning but nicer and shorter. Any why, its belt driven at the time with plans to converting it to ball screws or other more ridged solution. this is my solution. never mind the numbers showing it was just for the pic. as you see its flat against the plate in the lower part of the pic and therefor it is parallel to the ALU beam so the measuring has zero angel error and against the x axis plate. the i open up bcnc. set it to move 1mm, so it compresses the caliper. that means if i zero it before i get a actual moment reading on the caliper. click the software to move 1 mm five times. calculate the average on each move. like this 12,35=A*5 A=2,47mm then do it again but only one move and write down each to see the differences. one klick on software=1,98mm one klick on software= 2,19 one klick on software=2,10 one klick on software=2,35 one klick on software=2,50 do this a few times to get the whole range from lowest to most movement. on above example i get a differences to 0,52 mm(its more like 0.4mm but i cant remember) well i try to calibrate it as good as possible and by measuring as good i am able i will receive the most precision. but as you mention i probably do not gain super much by using mag scale. as for the grbl/bcnc not able to use the input from the mag scale. they are both open source, so its possible to add a few lines of code. but the big question is, is it worth the hassel? what do you recommendation on measuring to calibrate it?
Ha! That happens to me all the time, even with English as a first language, so don't worry about it. That actually looks pretty good as a measurement system, I just wasn't getting it from the description. The variation, unfortunately, looks fairly normal for belt-driven mills. Do you have the belt clamped tight and tensioned properly with spring tensioners? Any amount of slack in the system, with a heavy carriage, may result in variable moves. Double belts could be worth a shot. Give this recent thread a read: Workbee - precision ... It's a bit long, at five pages, but it'll give you a good idea of what you can expect on your own machine. The eventual solution after several months was simply to switch to screws, unfortunately.
To Add to what Rob said, make sure you don't have your wheels too tight, this will lead a more backlash than typical. And I recommend Dual belting (Double belt) for sure, I've been using it for about 2 years with no issues. Martin Barfoed is the father of Openbuilds DualBelt as far as I know. Do your calibration over a long travel such as 300mm using a good ruler. Cheers Gary
well the z axis was lose so that probably did solve it. well i knew i have to upgrade to screw before i bought i. my plan now is to get it working with descent precision for 1 year then start to upgrade it.
have i done the calculation correct, it seems that im getting too small number. i have following this guide:Stepoko: Powered by grbl Hookup Guide - learn.sparkfun.com the motor is a 200 step per rev pulley diameter :18.24mm microstep 1/4 pulley circumference 18.24 x π now i use the formula in the link above. (200/(18.24 x π))/0,25=13,96095..... default $100 and $101 is 250 and when i set the machine to move before calibration i did get software mm : actually movement mm 1mm : 2,332mm. in % its more then 133% to much. when on the subjekt how much micro stepping is usual? 1/8 or 1/16 or 1/32?
Pulley circumference is irrelevant with toothed pulleys. How many teeth? Pulley teeth * belt pitch = one-revolution travel distance. Divide that by steps, and you have your distance per step. Invert that for steps per mm, of course. For example, a 20-tooth pulley running GT2 on a 1/8-stepping 200-step motor. 20 * 2 = 40mm per rev. 40/200/8 = 0.025mm per step. 1/0.025 = 40 steps per mm. Remember, this is just to get you in the ballpark. You might have 39.783 as a final number after proper calibration, this just gets your machine not running 2mm for every 1mm you put in! With belts, I'm not sure I'd go below 1/4, if even that low. 1/8 is the last stage before torque starts dropping off a cliff, but since you likely need all the torque you can get I'd probably go 1/2-1/4 stepping. Understandable. At least this way you can cut the plates for the new motor orientation yourself!
Typical for a belted system is 1/8 micro, any less gets a bit noisy and not smooth. Are you using typical 20tooth pulley? If yes, then it’s 20*3mm = 60mm per turn then 1600/60mm = 26.66 steps per mm in the grbl setting. This is with 1/8 (1600) microstep. Cheers Gary
Gary made a point earlier that I think you missed. Don't use such short movements, use the longest you can. That will give you far better accuracy. Secondly, doing the math of steps and ratios is good but you still need to measure and tune your steps/mm. Belts stretch, screws are imperfect, things wear. There are procedures laid out but it basically entails for each axis: Measure the actual movement of a jog. Use the largest distance you can. Larger distances minimize measurement error. Compute a new $10x value via new_value = old_value * (jog_distance/measured_distance) Plug the new value in and repeat the process until jog distance equals the measured distance
If you're using a 6"/150mm digital caliper, make sure you're using a minimum of 100mm of it, preferably 125-130mm. This is pretty important. Also, once you've dialled it in in one part of the machine, try it over a different section of the axis to make sure it still agrees- you want consistency everywhere. This is what I meant by my "get you in the ballpark" comment above. Typically once you've put in your theoretical calculated value, you'll have to do at least 2-3 rounds of this commanded/measured ratio process to get your true precision value. Super important!
