Received the HDM last week. Been busy getting setup and making a pallet for my SMW fixture. Well today I finally got to cutting some tests. From a 12”x8” sheet of acrylic i cut 10 ~2”x3” nested rectangles. According to my Mitutoyo calipers all pieces were with 0.01mm of each other. Yes millimeters! The parts aren’t accurate (but that can be dialed in) so next I’ll start on calibrating the steps but I’ve got to say this is great news to start my week with.
I’m waiting for one of you HDM owners to start asking Carbide 3d to incorporate “ballscrew mapping” into Carbide Motion.
I think I’ve had repeatability that good and accuracy within 0.02mm. Make sure your endmill is on-size and that you have a good finishing strategy.
Yeah I’m thinking i can get this dialed in with a little step adjustment. Really excited to see how far i can push production… i think my pre and post work will end up being the bottle neck
I didn’t have to adjust steps any. With the ball screws it’s more simple math than belts. How far are you off?
I’m off by: +0.056mm on X, +0.1mm on y
So I was going to adjust the steps in the firmware using the formula:
Distance traveled x steps per mm / steps commanded and put that value as the new steps per mm. Is there another way?
That’s the only calibration method I know of.
But I also suspect that 0.056 mm could easily be the diameter of the endmill. Have you put a micometer to your endmill to verify it’s diameter? When we verified a new machine at Boeing we sent off the calibration circle/diamond/square test with the endmill used so the metrology people could take that into account before the machine was put into service. Even temperature was taken into account.
You may be chasing several different variables, none of which you’ve said you’ve eliminated. Also, most calipers (even the best ones) have a +/- 0.05mm variance themselves. So verifyably measuring below that in accuracy will require a micrometer or other more accurate metrology equipment. @Vince.Fab measured a test he did with calibrated metrology equipment and got accuracy to below 0.01mm.
Finally the metrology training and being in on multi million dollar mill verification, I experienced in aerospace is coming in handy in real life!
Please don’t put a measuring tool w/ metal anvils and so forth against delicate cutting edges — it’s better to cut a slot and measure what was cut instead, or use a specialty tool designed for measuring things w/o touching them (optical comparator?)
Will, true enough, I’ve seen pics from PreciseBits Application: Woodworking
that confirm damage| from measuring with calipers and micrometers. My workaround is to measure at the very top of the flutes, an area that is seldom if ever actually cutting anything.
With my luck, the day after I did that, I’d have a project where that tool would be needed (and probably used) for a full-depth finishing pass, resulting in the damage to the cutting flutes showing all around the perimeter of the part.
Good point, that’s why I keep my final cut/finish mills separate and un-measured since the tools precise diameter is not so critical for that type of cut.
If you are going to do the indirect, measure a slot method, measure run-out first. You’re adding in extra variables that need to be accounted for and we’ve seen many a well-intentioned hobby machinist start chasing phantom errors…
Machining slots is also subject to material deflection/rebound (ex in wood and fibrous materials), and machining with single flutes (vs a 2-flute for example) can cause more deflection since only one side is pushing against a wall and the cutter bounces around between walls on a microscopic level.
Measuring movement directly with something like a 2" dial indicator would be my preferred way to quantify accuracy if you’re going to use that data for calibration.
I’m very happy to report back that I’m hitting within 0.01mm precision accurately on nested parts over 12". I won’t be sending these in to a lab for exact measurements but for my needs they are prefect. Slip fit precision… check! accurate to better than a thousandth of an inch…check! repeatable in a nested production run… check! All that on a $5k machine… color me impressed!
What changes did you make?
Just adjusted the stepper values in the GRBL firmware. $100 ans $101 i think… Very minor adjustments figured using the formula above. Still need to calibrate Z but X and Y are
I wouldn’t think there’d be a variation in $100-$102 values with ball screws.
@nuance.merger How/what are you measuring?
I’m cutting 3mm cast acrylic and measuring the parts with Mitutoyo vernier calipers. I’ll test movement with Mitotoyo .0002” indicator today. The parts I’m cutting sit in a known socket with 4 locating pins. Before i calibrated the parts did not fit properly. I ruled out incorrect measurement of the bit because the error was not consistent on X and Y. I’m no engineer and have no formal training or precision industrial measurement equipment but if the parts fit the parts fit right?
But yeah i think regarding ballscrews unless you’re using very high grade precision rolled (which i doubt is In use here) they aren’t going to be exactly perfect. I’d also assume there also can be variance in the individual steppers causing very slight error. Now i would guess 95% of HDM users would be fine with it as is out of the box (within .1mm) but for my needs a bit of tweaking was needed. This post wasn’t to criticize the HDM but to praise it and share my findings.
Just for reference my previous CNC after endless calibrating was only able to get within about .1mm error over 12” cutting the exact part nest and needed tweaking in CAD/CAM to correct the issues.
There is assumably always some travel error, how much would be at least partially indicated by the class of the screw, P or T Class, 1,2,3 ect. Then you have all of the stacked error from the other components. Fixed vs Rotational ball nut, max rapids ect. Ballnut preload, adjustable or otherwise. Linear bearing quality and setup. I think most of your industrial style controllers allow one to map this kind of stuff out over the machines limits, as its probably not going to be linear. Linear scales are another option.
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.