I took these pictures this morning. Let me know if there’s anything in particular that you wanted to see and I missed.
All the tension bolts are loose since the machine already is square, there are no cracked plastic pieces. Maybe I should enlarge the tool mount - could it be that those two rings are not stiff enough and twist on themselves? The Z assembly does not seem to be able to flex. I’ll try to get that on video if (when) it appears again.
Are you saying that I should pull the spindle up? I left it that much lower because this way I could have more clearance between the bit and the bottom of the Z rails, and I’d be able to cut in places where the whole Z assembly won’t fit when dropped down (like in the corners of the tray I’m using).
I’ve had small holes like that turn into triangles instead of holes with helical drilling before. I could tell if it was from the gcode, or marlin interpolating the curves into lines. They were close enough, so I didn’t really worry about them.
Mines are off enough to be useless, the four angled slots in the lower part should be 3x4.5mm but they basically turn out as slightly widened 2mm holes as per the pictures linked above; M3 bolts just won’t fit, let alone slide inside the slot… Maybe I should turn arcs off in Estlcam, just to be sure?
What do you think of these attached settings? I hope I’ll find some good combination soon because ordering more CF to cut test onto is going to be quite expensive.
I tried the settings above, the end result did not change much; I’ve attached a picture with the expected and actual dimensions of the cuts. Before cutting I’ve also replaced the tool mount with a single block version with cooling.
@vicious1, here are two videos of the flex - check your speaker’s volume before playing! The helical drilling looks too fast to me, does it follow the main feed or can it be tuned?
I really cannot understand where this flex comes from, the Z assembly looks tight to me. Cutting CF is getting disheartening, I ultimately built the MPCNC for this but I must have made mistakes along the way, it doesn’t look plausible to me that some people here are milling aluminium and I can’t get precise 3mm holes in a sheet of composite…
I can not tell from that video where that is coming from.
Stop wasting expensive CF and start with wood. Or better yet with HD foam. Then wood, then plastic, then CF.
Obviously something is not right, but it could be many things. You are right there you should be able to see it happen. Make a test piece with 30 of those slots and watch it until you can tell us what exact piece is moving. A shaky video does not provide much info. To me it looks like the whole thing is moving, or just the tool mount is not tightened up to the Z rails, At the same time that is exactly what happens when you have a bad bit and it is not cutting?
Sorry for the delay, I’ve been waiting for some new 2mm bits that are suitable for trying the said test cuts in wood.
I’ve tried to cut a tiled pattern and it is indeed showing an erratic (but consistent) behavior: the dimensions aren’t exact even in wood, but the errors seem to be the very same in every tile.
Ignore the vertical slack, briefly after shooting the video I noticed that the lower Z bracket slid slightly downwards. I pushed it back up against the leadscrew’s coupler and the slack vanished. The X/Y slack got better after having tightened the belts some more, but my experience with 3D printers is telling me that they might now be too tight. We’ll see if that’s true in the next few test cuts though, I’m no belt-based CNC expert.
I’m skimming through the assembly instructions again to check that everything is tightened in its place, and I’m noticing that the bolts on the sides of the bottom XYZ part might have gotten loose - or I may have not tightened them enough in the first place!
This is quite strange, since the tests that I initially did with the machine before the summer weren’t showing these effects - holes and slots were precise… Something that I’m overseeing must have worn out / cracked / bent. I’ll also try another test cut with arcs disabled, the fact that the errors are so exactly repeated between the tiles is very suspicious to me.
Based on the last video you in the album, it looks like one of your stepper pulleys is loose. I can’t move them when they are energized without skipping teeth. I’d check the tightness of the grub screw to make sure the pulley is firmly attached to the stepper motor shaft.
I finally got a precise cut in wood of this pattern! So far tightening the belts way more than I though was necessary worked, and disabling arcs in Estlcam probably took part in the trick as well (thanks for the pointer @barry99705). The pulleys’ grub screws were still tight.
I’ve also tightened the external bolts on the lower XYZ part, but they don’t seem to have a substantial effect on slack. I’ll see what kind of results I’ll get with CF now.
Making solid progress! The arcs is a tricky thing and can be tuned in config advanced. Depending on your firmware version it is probably best to leave them off, it just makes for a bigger gcode, no big deal. I had to tune this for last years MRRF plot with super tiny features.
Today I tried another pair of cuts in CF but they still don’t come out well: the dimensions got slightly better, but they still are nowhere close to the drawings. I tried the same settings above at 2.5 mm/s too but there were no noticeable improvements. Also no differences between using the diamond and the single upcut flute bits.
I think I’m missing something obvious here, since the cuts are good in softer materials (wood, PVC) but CF just won’t get machined precisely. I could give trochoidal milling a shot just for fun…
0.2mm width, 2mm DOC, 5mm/s feed, 1mm/s plunge. A little bit faster than the standard cut, but I can’t see any evident deflection in the tool with these speeds.
How far out are your tolerances and are they consistently undersize/oversize depending on feature? Using Kevin’s tip from his aluminum guide to fudge your tool dia on finish pass to account for machine’s deflection has helped me dial in the last several thousandths. Haven’t tried CF yet, will hopefully be making my water table soon. I’ve had some CF sheet laying around for a long time, I really want to cut it.
The slots/holes are consistently undersized, unfortunately the dimensions have not varied that much from the drawing that I previously posted.
Changing the the theoretical tool diameter could fix the holes, but then the outer contour of the part wouldn’t be right… I’ll take a look at that guide anyway, thanks.
Cutting underwater is really handy during the job, but is definitely a PITA while preparing/cleaning up. Especially when to dispose of the polluted water you have to carry the filled-to-the-brim container down a ladder and out of the garage :lol:
When you’re tuning your finishing passes in I’m finding it useful sometimes to run the contour or adaptive rough first, check measurements & then run the finish pass to see how much stepover you’re really getting on a finish & try to work that into the cutter dia fudge as well.
More fudgey but if you’re having consistent dimensional issues in identifiable feature types you can set up templates with whatever cutting data you need to get your tolerance & surface finish. I’m just getting into the templates in Fusion 360, they are awesome.
Do a full depth finishing pass and vary the thickness of it. I use 0.2-0.5mm depending on the material.
Also I know you want quad parts but if you are trying to figure out dimensions you should make a test part, square or octagon with a hole or two in it. Real number will help greatly here, some guys are wasting our time chasing 0.000001" others are 0.8" off. So if you mill a square you can check all sorts of things and give us exact numbers. X and Y accuracy, diagonals tell us if you are square, a small and a large hole help check your CAM.
I finally got the holes to be within a 0.5-0.75mm tolerance!
I’ve snugged up all the tension bolts and now the Z axis doesn’t wobble anymore: the bearings are much tighter on the vertical tubings and don’t slip anymore when pushed by hand. I’m only having trouble with the external-back ones (circled in green in the attached picture) that are not applying enough pressure on the tubings, but I’ll figure it out. @vicious1, am I doing this right? I’m not seeing bad effects on squareness right now.
I still have to solve the problem of the XY slack: the carriages move slightly when I apply pressure on the bit, even though the belts are really tight already.
I’m sure F360 is awesome, but I’ll check it out when they’ll publish either a Linux build or a functional web version. OnShape is a great web-based CAD system, I wish they would go beyond that and expand to CAM as well.
You could have over tightened or your Z isn’t straight. I can only suggest taking it apart and starting again. Move the Z axis all teh way down and all the way up and see if it changes.
With no tension on any of the bolts it should be snug, it is an interference fit, Making everything tighter is usually a bad thing. It is meant for balance not overall tension or we would just start out cranking them all down as tight as possible.
At the same time I think we really need to know how far off you are before you bother. What do you get milling a 50x50mm square with a 10mm hole?
When you say the carriages move slightly, you are referring to the X/Y carriages on the outer rails, right? Do both of the X move? How about both of the Y?
I’ve loosened all the tension bolts and re-tightened them wisely, trying to get a good fit using only two sets of them as I read in other posts here. Now all the bearings are snug and don’t move when (reasonably) pushed by fingers. When the Z axis is all the way up the bearings that I marked in the last attachment feel looser then when it is all the way down, but I couldn’t figure out the reason for this.
I cut a sample and took measurements: pics, measurements, design and gcode are all attached. There’s overall ~0.5mm of slop and some noticeable out-of-squareness, but please note two things: I took these measurements with my (chinese) digital caliper, so I guess that ±0.03mm of tolerance should be accounted for; this was cut in MDF since I currently don’t have any other stock that I can fully cut out a sample from - most of the other wood that I have right now is >2cm thick - so the caliper could have dug a little in it in some spots. I took multiple measurements and averaged them.
When I drilled the holes for the legs everything was perfectly square, I’d say the by fumbling with the tension bolts I must have put the machine out of square. Before every cut I use these to square the axes anyway.
Yes, I’m referring to those. When pushing the bit by hand - with the tubings centered in the middle of the axes, clearly - they seem to move by the same amount on both sides. I’ve eyeballed a 0.5-0.75mm movement - this was obviously tested when the motors were energized.
