Shapeoko 2 - cuts ~0.5mm short, tried almost everything!

jacquard
Posts: 13
Joined: Fri Dec 09, 2016 10:48 am

Shapeoko 2 - cuts ~0.5mm short, tried almost everything!

Post by jacquard » Thu Dec 15, 2016 12:59 am

Hi there, a few months ago I recently completed probably the world's slowest Shapeoko 2 build over the last 18 months (which included building an entire workshop/hobby building for woodworking, where it was to live!)

I've built it all up and it is working smoothly, however I am having an accuracy issue I simply cannot get rid of. When I carve something, like a 50x50mm square, the dimensions are about half a millimetre too short - like 49.5x49.5mm. It doesn't scale with size, e.g. if I did a 500x200mm square, it will come out as 499.5x199.5..

Everything else mechanical about the machine is fine. It doesn't miss steps, no unusual noises (other than bit chatter), doing steps/mm calibration checks by plunging holes comes out perfectly accurate, and using a pencil onto paper instead of carving is also accurate (though I haven't checked this to the nth degree). It's just cutting that's the issue.

I've been reading the past forum posts like a sponge for the last month or two trying to see if anyone else has had similar problems (and found solutions that might help me), but despite looking at several possible issues and investigating them, the inaccuracy persists!! :x

I plan on rebuilding the machine and doing it as precisely as possible; since I know how it goes together and how it's all supposed to work, I'll be able to do this much more quickly and competently. Maybe 1.8 days instead of 18 months :D

Sorry for the long post of info and data, but from looking at the other threads I've noticed a good correlation between supplying lots of objective, measurable data and getting a great result. So I'm going to join that particular cargo cult as I have nothing to lose :lol:

First of all, some background about my build. It is a build kit from 3dtek.xyz that included a few big upgrades out-of-the-box:

* Shapeoko 2 mechanicals with 1000mm makerslides (effective carving footprint 785x785mm)
* Dual Makerslide X Axis, 8mm thread Z axis
* NEMA23 motors for X, Y1 and Y2 (dual motor Y axis), NEMA17 Z axis
* upgraded to GRBL 0.9j, Arduino + CNCShield electronics (separate driver for each Y motor)
* Machine base is 19mm MDF sheet, not clamped to table
* 400W 48V quiet cut spindle. 12000rpm (always been used at max RPM)
* Pieces being cut are held in place with a large wooden vise integrated to work area


I also get quite a bit of bit chatter; I am cutting into pine and oak mainly, with feed rates 800-900mm/min, 1.0mm plunge with a 3.175mm single flute endmill, and 12000RPM. Notably - I get almost zero bit chatter with a 2mm double flute endmill, all else being equal, and the inaccuracy is still there...

---------------

Here are the various checks and changes I've done based on my forum research for possible issues:

Belt tension – tensioned X and both Y belts to the same values (approx 2kg force per 25mm deflection). I was measuring this with the "fish scale" technique.

V-wheels – all eccentric nuts have been set such that the V-wheels don’t slip at all when trying to move a wheel separately from the assembly it’s attached to, and no tighter.

Threadlocking – All nuts that are accessible post-assembly have been replaced with nyloc nuts. Threadlocker applied to threaded insert of Z stepper spacer, and the two bolts holding Z stepper mount plates to Z axis Makerslide. (This made a BIG difference to audible vibration, as those spacers were rattling loose) The eccentric nuts haven’t got any locking treatment (Nyloc nor threadlocker) but I haven’t had any issues of these coming loose.

Calibration of steps/mm values after tensioning – completed by drilling holes 600mm apart and measuring actual edge-edge distance, and making steps/mm changes based on the results. Performaed immediately after belt tensioning, and again after about a week to see if there is any further change from belt stretch. Calculated values are X is 39.752, Y is 39.753.

Runout Measurement – endmill in use for testing (3.175mm nominal) reads as 3.17mm on caliper, and makes holes of 3.16mm (Inaccuracy in cheap digital calipers?) Used a torch to project a sharp shadow of the endmill on a white background, and then spun the spindle at various RPMs. No visual indication of significant runout.

X axis rail stiffness – I've put some M4 bolts through the gap between the two X-axis rails, supported with a small washer above and a 25mm wide washer below, held in place with Nyloc nuts. 8 washers in total (approx. 11cm between each)

Spindle/Endmill Squaring – used “wire with two right angles” to ensure collet is square in X and Y directions. Done almost perfectly, approx. 0.5mm inaccuracy over 500mm. To get Y axis squaring dialled in, I shimmed the lower part of the spindle mount by 0.33mm using a feeler gauge finger piece.

Squareness/Parallel Y Axis Rails – measured the diagonals between identical points of the rails using a tape measure. Measurements are within 1mm.

Heights of Y Axes, Both Ends – left rail, 99.73/99.75mm; right rail, 100.19/99.55mm

Height of X Axis Rails, Both Ends – left side 118.61/118.81mm, right side 119.14/119.41mm

Backlash in X/Y axis – I haven’t done a specific backlash test yet, but given the tightness of the belts, and the magnitude of the error, I doubt that backlash is the main cause of what’s going on.

Stepper Current – stepper motor current has been increased to ensure no binding (was originally a problem on Z axis due to runout of nuts securing it to the bearing). There is no noticeable (from cuts) or audible skipping of steps. Drivers are cooled with a 120mm fan and have never shown any thermal cutouts.

You can probably tell by now that I'm really out of ideas on what to try next, short of pulling the whole thing apart and rebuilding it!

---------------

When I finally do do the rebuild, I'm going to be doing a few further tweaks and upgrades that I'm hoping will either (a) improve accuracy and stiffness, and (b) make it a more competent machine. I'd appreciate any advice anyone can give about how much (if at all) any of these would contribute to the issues I'm facing at the moment.

Tramming Wasteboard – I’m not concerned with Z axis accuracy right now, I’m focusing on X/Y at the moment first. I just recently bought the bit and collet to do this, will do it once I change the size of the router. Of course when I rebuild I'll use a fresh piece.

Y axis dust shield – to go along the Y axis from wasteboard to makerslide, to protect the slide rails and Y axis wheels from dust. like in this Youtube video. Actually that guy on the video has heaps of great ideas and I plan on doing a lot of the upgrades and tweaks he has on his videos.

Improving Y axis stiffness – put two brackets to secure Y axis along its length

Revisit X axis stiffness fix - I'm considering moving from doing the bolts tweak, to an aluminium bar tweak similar to Marius Hornberger's video I linked above.

Dust shoe – will make a simple one for the quiet cut spindle, and another if/when I upgrade to Makita

Reduce router size – I won’t be doing any work over 600x600mm, so will be reducing the size of the machine approx. 150mm in each dimension. (625x625 effective area vs 785x785 currently)

Improved base/torsion box – first off I'll firmly bolt the machine down to the worktable it's assembled on, and at some stage I'll make a torsion box for the smaller sized base. Will include a removable wasteboard (instead of over-plunging directly into the machine base as it is now) and also some holes/threaded inserts for clamps.

Z axis fix for larger routers – will be modifying the Z stepper mount plates as per the "Shapeoko 2 Z Axis Fix" Inventables project to make vertical clearance for a future Makita router upgrade

Makita router – is on the cards once I have the money to buy the unit, mounts and new collets. Not really a priority as I doubt I’m really at the limit of the current spindle, but the idea of running the machine much faster once it’s all dialled in is very appealing.

ACME Z axis screw – will probably do this on the rebuild “because I can”.
Last edited by jacquard on Thu Dec 15, 2016 2:05 am, edited 3 times in total.

jacquard
Posts: 13
Joined: Fri Dec 09, 2016 10:48 am

Re: Shapeoko 2 - cuts ~0.5mm short, tried almost everything!

Post by jacquard » Thu Dec 15, 2016 1:31 am

In case it's of use, here are my GRBL configurations..
  • Grbl 0.9j ['$' for help]
    >>> $$
    $0=10 (step pulse, usec)
    $1=255 (step idle delay, msec)
    $2=0 (step port invert mask:00000000)
    $3=5 (dir port invert mask:00000101)
    $4=0 (step enable invert, bool)
    $5=0 (limit pins invert, bool)
    $6=0 (probe pin invert, bool)
    $10=3 (status report mask:00000011)
    $11=0.050 (junction deviation, mm)
    $12=0.002 (arc tolerance, mm)
    $13=0 (report inches, bool)
    $20=0 (soft limits, bool)
    $21=1 (hard limits, bool)
    $22=0 (homing cycle, bool)
    $23=0 (homing dir invert mask:00000000)
    $24=25.000 (homing feed, mm/min)
    $25=500.000 (homing seek, mm/min)
    $26=250 (homing debounce, msec)
    $27=5.000 (homing pull-off, mm)
    $100=39.752 (x, step/mm)
    $101=39.753 (y, step/mm)
    $102=315.000 (z, step/mm)
    $110=1500.000 (x max rate, mm/min)
    $111=1500.000 (y max rate, mm/min)
    $112=400.000 (z max rate, mm/min)
    $120=20.000 (x accel, mm/sec^2)
    $121=20.000 (y accel, mm/sec^2)
    $122=12.000 (z accel, mm/sec^2)
    $130=760.000 (x max travel, mm)
    $131=760.000 (y max travel, mm)
    $132=50.000 (z max travel, mm)

jacquard
Posts: 13
Joined: Fri Dec 09, 2016 10:48 am

Re: Shapeoko 2 - cuts ~0.5mm short, tried almost everything!

Post by jacquard » Thu Dec 15, 2016 1:52 am

Some photos of the assemblies, in case they help. Note that in these photos I've removed the X axis stiffening screws, and the vice for holding things onto the work area.

Machine and Work Area
Image

X axis
Image

Y axis
Image

Z axis
Image

Gratuitous photo of the workshop built to house the machine (among other things) :D
Image

WillAdams
Posts: 8628
Joined: Mon Apr 09, 2012 6:11 pm
Location: Pennsylvania --- south of the Turnpike, East of US-15
Contact:

Re: Shapeoko 2 - cuts ~0.5mm short, tried almost everything!

Post by WillAdams » Thu Dec 15, 2016 2:38 am

Looks like a nice setup!

Seems as if you've done a very good job of trouble-shooting.

My suspicion would be that the spindle / endmill is getting pulled into the cut --- please try cutting a diamond-circle-square test, leaving a >0.5mm roughing clearance, then make a finishing pass which cuts to the final dimension --- may need to try both conventional and climb milling for the finishing pass.
Shapeoko 3XL #0006 w/ Carbide Compact Router w/0.125″ and ¼″ Carbide 3D precision collets

jacquard
Posts: 13
Joined: Fri Dec 09, 2016 10:48 am

Re: Shapeoko 2 - cuts ~0.5mm short, tried almost everything!

Post by jacquard » Thu Dec 15, 2016 2:49 am

Thanks for the quick suggestion - I've been spending so much time tilting at windmills trying to resolve this in hardware/calibration, that I haven't had chance to learn a proper CAD/CAM package yet - I'm still running Easel, yet learning Fusion 360 will be on my new years' resolutions list!

I did do a circle-diamond-square test semi recently - I put in the wrong stock thickness so cutting it out was a bit scrappy but it is attached. Stupid me forgot to mark the orientation of the test so it's of limited use, but I'm "fairly sure" as per the pic, that the bottom left is towards 0,0 on the machine.

Image

In case the numbers aren't too clear;

Diamond - should be 30mm square - 30.41/30.48/30.87/30.82 for each edge
Circle - should be 48mm round - 47.82 in the horizontal, 48.10 in the vertical, 48.45 on one diagonal
Square - should be 55mm square - 54.62 on one side, 54.54 on the other

BTW - this was just done in Easel, so didn't have an opportunity to do roughing and finishing passes etc.

WillAdams
Posts: 8628
Joined: Mon Apr 09, 2012 6:11 pm
Location: Pennsylvania --- south of the Turnpike, East of US-15
Contact:

Re: Shapeoko 2 - cuts ~0.5mm short, tried almost everything!

Post by WillAdams » Thu Dec 15, 2016 1:11 pm

That helps a bit. Given that the numbers are asymmetric, either the machine isn't square, or cutting forces are pulling it out of square --- given the work you've done, and the size of the machine, the latter seems the case.

Suggest processing a circle-diamond-square in a tool which allows you to control path orientation, determining which combination of roughing and finishing passes in what direction yield the most consistent dimensional result.
Shapeoko 3XL #0006 w/ Carbide Compact Router w/0.125″ and ¼″ Carbide 3D precision collets

HPGreyling
Posts: 123
Joined: Sat Nov 15, 2014 11:37 am
Location: South Africa

Re: Shapeoko 2 - cuts ~0.5mm short, tried almost everything!

Post by HPGreyling » Fri Dec 16, 2016 4:49 pm

I am thinking runout.

I also have a SO2 and if I run at those feed and plunge rates I do get these wildly frustrating odd numbers. Even with those bolts through the middle you still get that bit of flex if you're going to fast.

Try to reduce the feed speed to around 500-700 and plunge to around 250.

jacquard
Posts: 13
Joined: Fri Dec 09, 2016 10:48 am

Re: Shapeoko 2 - cuts ~0.5mm short, tried almost everything!

Post by jacquard » Sun Dec 18, 2016 11:59 pm

Thanks all for your feedback. My plan of action from here is:

a/ improve chassis stiffness and intrinsic accuracy by rebuilding it at 850x850mm, with some further stiffening mods, loctite, and an eagle eye on the square and calipers

b/ try conservative feeds and speeds and work from there

c/ learn a proper CAD-CAM toolchain like Fusion 360 (highly recommended by Ben at 3dtek)

d/ if and when needed, move to a Makita spindle when time and spare money allows

WillAdams
Posts: 8628
Joined: Mon Apr 09, 2012 6:11 pm
Location: Pennsylvania --- south of the Turnpike, East of US-15
Contact:

Re: Shapeoko 2 - cuts ~0.5mm short, tried almost everything!

Post by WillAdams » Mon Dec 19, 2016 1:59 am

Sounds like a plan --- I really suspect that there must have been some underlying engineering / FEA which resulted in Inventables creating the 750mm size, so reducing the total span, esp. along the X-axis is sound --- for the Y, my inclination would be to let that length be, and just add side supports.
Shapeoko 3XL #0006 w/ Carbide Compact Router w/0.125″ and ¼″ Carbide 3D precision collets

jacquard
Posts: 13
Joined: Fri Dec 09, 2016 10:48 am

Re: Shapeoko 2 - cuts ~0.5mm short, tried almost everything!

Post by jacquard » Mon Dec 19, 2016 8:44 am

The main reason I have to shorten the Y that it fits the table more cleanly.. at the moment it juts out some 170mm from the edge, and in general gets in the way. My wood stock is usually 450/600 (cut from 900/1200/1800/2400 sheets) so the difference between 600 and 780 working space doesn't yield much extra utility.

All this CNC tinkering can get a bit obsessive... I'm looking at my mechanically-but-not-yet-electrically-built Ordbot Hadron, and thinking about how to re-purpose all its pieces into a second, smaller CNC machine.. hmm the pair of 470mm makerslides would make a decent X axis gantry, the NEMA17s aren't as grunty but are better than a kick in the head, ... and so it goes ... :)

Not helping me from falling down the rabbit-hole is that the workshop gets too dirty with wood dust to keep a 3D printer anyway, so if I need one in the future I may as well buy a second hand "indoor-friendly" one and leave it in the study..

Post Reply