Pretty sure they will, it dont take much cutting force in my experience. I have several 17’s on hand but not the time…yet!
The re-ignited interest in R/C have forced forward a redesign of my CNC v2, I want larger machine area in order to cut molds for composite models. I am now limited to 350x410mm but for a 1,5m DLG I need close to 1000mm to do full half span carves.
The ShapeOko was launched with NEMA17 motors and a hot-wire will have less cutting forces so there should not be a problem.
Yes it is, this model was the main reason to get a ShapeOko2. 762 grams in this stage for 1m50 span and detachable is something that is not possible with hand tools, the machine allows me to build with less wood. Then the machine became the hobby, I did some strength calculations and optimised my machine for balsa and plywood.
The best upgrade was to make stronger V wheels for a SO3 type Z axis with an 8mm ballscrew (balsa is max 12mm thick) and full length support for the Y axis (6x100mm alu) and of course sandwiched the X axis makerslide with an 4x40mm strip. A very silent 400W Quiet Cut (runout 0,01mm) controlled by the Arduino. Final upgrade was to replace the 6mm belt of the X axis to 9mm, this really helps to reduce backlash on X, I kept the 6mm on Y.
In order to see how good the machine is now I designed a square where each leg has a part with backlash and a part without backlash. Next to the square a circle. If you have your ballscrew up and running I would love to see the results to compare.
BacklashTest.dxf (11.2 KB)
I can run that later for sure if you want, what kind of feed rates/bit info do you have on your calibration routine?
My design is weak/flexible in the X-direction due to the MDF side walls, but it is doing quite decent 
The Makita mount was completed today in 25mm Alu, 900mm/min @ 1mm DoC with a 6mm end mill. Still need to drill out mounting holes and tap them before I can assemble/square up the new Z-slider.
I wanted to see the backlash created by the friction of the wheels on the gantry and stretching of the belts so went slow on cutting the plywood.
MDF is a very flexible material, with a gauge mounted on the spindle mount and pushing it down gently shows no sign of flex on Z but if you push on the 22mm MDF you will see flex appear at even low pressure. So for accurate wing moulds you need to make sure the bed is rigid too.
The weak link now is flex in X direction because the MDF do flex. I can get it to flex when cutting deep in aluminium (the shaft start to rub when I am deeper than the flute length) which give a low enough frequency / vibration to agitate chatter. The next point of flex is in the Z/Y plane.
I have also tuned down the acceleration on X because rapid change in X direction would cause the gantry to flex since most of the weight is high above the Y-rails.
I have the Z torn down at the moment because I am assembling the new Z-slider
Update - I finished assembling the new Z and performed a quick test cut. The sound of the router is very different, more “connected” to the machine and less rattle/vibration noise. Needed to recalibrate steps for Z and is is pretty close to spec.
The test carve was two circles with 12mm diameter / 2mm depth and 20mm dia with 1mm depth + a rectangle of 50x30mm at 0.5mm depth in aluminium. RPM 18k with 6mm 1F running at 900mm/min and 0,5mm DoC.
I have a manual calliper and the circles measure about 11,8 and 19,8mm, and the rectangle about 0.2mm undersized. I have not measured the bit actual diameter but my machine has some flex in X/Y.
As far as the Z goes - it is FIRM! I manually reset Z to the top surface of the alu, placed a 8kg load on top of the Makita and jogged Z up 0.03mm before it cleared again. I used a voltmeter set to continuity mode.
Very happy with the new assembly Now I need to improve the X & Y 
Operational, this is the first piece of machinery my CNC made to improve itself 
Still need to tie in the small things like homing switch, mister, drag chain etc.
The design is made for a new larger CNC with higher gantry to I cant utilize the lift range as-is. Might make a shorter, temporary Z-plate to use in the meantime
A quick tramming adjustment was made and over a 8" span I have about 1mm of Z-discrepancy. Not perfect yet.
That is probably the spoilboard bending on the load, not the Z axis. Z axis looks good.
Sounds like you are going to build a real big machine, for inspiration, this is the RBC kits machine that produces the MIG29 rbc kits machine
My spoil board is 11cm thick, torsion box design with 76mm ribs and 21mm top/bottom board. But the SBR rails are mounted vertically on the side of the torsion box so the SBR16 blocks want to rotate under compression loads. The MDF is the link.
The back of the gantry have a epoxy/sand (EG) casting firming it up but I lack the same support for the side walls. I am considering doing a second casting which will include the side walls. This will however add another 7kg or so to the gantry.
I was originally considering doing a full EG machine with similar work space as I have today but desire longer work span. In that size EG become highly impractical for my back 
So my plans have changed and one axis will be extended to >1m work space, the width of the current gantry is sufficient for now and will be carried over. Yes, I have seen the RBC Kits work horse, dual spindle and all I also have plenty of sheet balsa on hand so all I need to do is design something… need more hours haha
We all do, welcome to the club
BTW - I haven’t tested to accuracy yet, but I expect do take a hit but 2mm DoC with 6mm WoC went without a hitch 900mm/min (chip thickness 0.54mm and MRR=10cm^3)
Here is a old Z / new Z comparison:
Prior shot was with a 4mm bull nose bit with large stepover (roughing) and the pencil point toward a vertical running area on the curved surface that visually is quite different (more flat) than the rest. This was due to a slight misalignment of the Z-stepper and ACME screw + a motor bracket that would bend ever so slightly when the ACME turned. The resistance during rotation wasn’t even and this is how it showed:
This next picture is with the new Z in place, and with much finer bit/stepover and there is no artefact present any longer
This test was done with a material that isn’t suitable for mould making, next step is to source some Corian or other suitable material and make a test mould for either the rudder or horizontal stabiliser which both are small enough for my current CNC work space.
Corian seems a good idea for making a mould. Make sure that you have it at room temperature before you start cutting. Corian expands 15 times more under temperature changes than aluminium this can have an influence to your cutting quality.
Good info Erik, thank you
Trying to source Corian… I knew it wasnt cheap but 4 pieces of 400x250mm would run me about €150 in total + shipping… Steep… 1m tabletop (63cm wide) start at €550…
I can get aluminium cheaper, might go that route despite it require more work to get smooth.
Test piece required 
Kitchen tops are sometimes made of Corian, if you can find a local installer they may have scrap from kitchen sink cuttings.
Also MDF really works well if you finish the surface with clear floor varnish.
Yes, I havent found one local yet though.
MDF on the other hand is easy enoug to get hold of, I have been considering carving the molds 0,5mm oversize and add fiberglass/epoxy to it for a recarve to actual size.
When you brush an acrylic floor covering paint to MDF it will absorb the paint and reinforce the MDF, the “hairs” on top of the wood can then be sanded off very easily as the deeper you get into the material it will be harder preventing you from sanding away material. Depending on how much is absorbed you need several coats.
Start with grid 220 and continue polishing untill you are satisfied. I use EzyKote on balsa which is a specialist product but have read several posts on acrylic floorpaint working as good as that. The only downside is that acrylic is slow compared to epoxy and may need a week or more to harden.
