Request for feedback on a 4/5th axis for a CNC drill

Hi,

I was wondering if someone could have a look at attached 4/5th axis design?

The design is made in FreeCAD and does not contain most of the detail yet. The design file can be found on my Github at https://github.com/inatic/4-5th-axis-cnc.

It is intended for drilling holes through 10x10cm and smaller softwood timber. Holes can have many different orientations, though the angle with the surface will at least be 45°.

As holes need to be drilled 'through' the wood they can always be drilled from the top or sides, and the router will at maximum need to be horizontal.

Components are:

* 15mm thick aluminum plates (EN AW-6060 T66)
* cross roller bearings as joints (Hiwin CRBE03515A)
* a 90x90cm aluminum extrusion connects to the machine (link 1)
* custom GT3 timing pulleys (130 teeth, radius 62mm) attach to links 2 and 3
* nema 23 steppers drive the axis (20 teeth pulley, radius 9mm)

Belts are tensioned by adjusting the position of the motors.

Forces are:

* Leadshine offers Nema 23 stepper motors with a holding torque up to 3.1Nm
* At the 20 teeth pulley radius of 9mm this translates to 344N applied to the belt
* A 10mm wide timing belt is rated at a maximum of about 200N (GT3) or 280N (GT5)
* At maximum belt strength, a holding torque of 12,5Nm (GT3) or 17,5Nm (GT5) is available at each axis
* The tip of a drill, extending 200mm out of the router, would be 350mm off axis center
* At the tip of the drill a force of 35N (GT3) or 50N (GT5) would be available to push sideways against the material (see attached image).

Would this force be sufficient for the drill to be able to enter the material? And more generally, is this a sound design?

I'm still at the initial phase, and would appreciate any input, suggestions and comments.

Best regards,
Jim

References:

Bearings: https://www.hiwin.hu/en/Produktfinde...ngs/CRBE/20591
Timing belts: https://www.electric-skateboard.buil...nd-belts/71058

Interesting idea and rather ambitious.You will need to take great care to ensure that the 4th and 5th axis are exactly aligned and that the spindle axis is also coincident.My other observation is that the stepper that operates the 5th axis is projecting and you are likely at some point to wish it wasn't there as it will limit access to tight spaces.You might not have a need to get into limited spaces at the moment but once you have a machine, there is no telling what kind of work might come along.If the 5th axis stepper could be flipped through 180 degrees or moved above the 4th axis mounting plate it might open things up a bit.

On the operating front, will you have the means to calculate a toolpath based on the huge tool length your drawing shows?If you were simply running a 3 axis machine it would be a lot simpler,but as I'm sure you know, tilting the spindle will require calculation of the Z axis location in X and Y to compensate for the angles at which the 4th and 5th axes are operating and these distances will vary with the tool length.This tool length will need to be a part of the calculation.As a fellow Freecad user I know there isn't a module of Freecad that will allow this level of calculation and no post processor either.Which means some other system will have to be implemented. I look forward to watching the machine evolve and appear in the solid.

Hi Vum - a 10mm wide timing belt although strong enough is not stiff enough for the job. I'd use 16mm minimum and 25mm preferred. 3.5kgf and 5kgf is not much of a drilling force. It will probably drill but not very quick. Prefer 20kgf min. A drill press can push 100kgf easily for instance. By hand probably 20kgf is reasonable. The holding force does not translate to the actual force as this is the electrically "stalled" force. Once the motor is moving the forced can drop off considerably. Try to get the motors speed/torque curve for more accurate force data from the supplier.

That length drill will not be very stiff and will "slide" off the job a little when it tries to start (as the edge not the point will contact first) consequently the machine has to be very stiff to enable a good start . ie belts and machine have to be very stiff. You may need a stepped drill or centre type drill or a flat end mill to create a start on the work piece before you can long drill. Peter

Have you ever tried drilling a hole that way? As Pete points out above, it's not really going to work without getting in there with another tool and making a starting place for the drill tip. Without that, the start of the hole will be off as much as an inch or so, and where the end of it comes out will be anyone's guess. If you really need accurate holes going all the way through a beam at that angle, a better way to get them would be to rip the beam lengthwise, cut diagonally across each side with a ball-end cutter, and reassemble it so that the channels line up to form the holes.

Hi routalot, peteeng, awerby,

Thank you very much for your replies!

I've started off by looking at the possible error due to belt elongation under maximum load.

Following calculations use stiffness of a HTD5 belt instead of the GT3, for the latter I don't seem to find any values online. Source for the numbers and the calculation is Mectrol's timing belt guide (link below).

Specs for the HTD5 belt (Mectrol guide p. 12-13):

T1all (max allowable belt tension for 25mm belt width): 1290N
Csp (specific belt stiffness): 23400N/mm

L (full belt length) = 500mm
L1 (tight side length) = L2 (slack side length) = 250mm
b (belt width) = 10, 16, 25mm

Calculating belt elongation (Mectrol guide p. 29):

k (resultant stiffness coefficient of tight and slack sides) = Csp*b*L/(L1*L2)
Fst (static force, I'm using the max. belt tension for a belt width of 10mm) = 516N
deltaX (elongation) = Fst/k = 0,27mm (b=10mm), 0,16mm (b=16mm), 0,11mm (b=25mm)

Elongation causes an angular error depending on the circumference of the driven pulley:

radius pulley = 62,5mm
c (circumference pulley) = 393mm
ae (angular error) = deltaX/c*360 = 0,25° (b=10mm), 0,15° (b=16mm), 0,1° (b=25mm)

At a load of 516N on the belt, this angular error causes a positional error at the tool tip:

d (tool tip distance from axis center) = 350mm
positional error of tool tip = 350mm*2*PI * ae/360 = 1,5mm (b=10mm), 0,95mm (b=16mm), 0,6mm (b=25mm)

For counteracting this error:

1. The hole could be partially drilled by the router, and I manually drill through the rest of the beam. A drill extending 10cm from the router head would bring tool tip distance at 250mm. Positional error would be reduced to 1,1mm (b=10mm), 0,7mm (b=16mm) and 0,44mm (b=25).

2. Significantly reduce the force applied when starting the hole, i.e. move very slowly while cutting away the first material. When the static force on the belt is divided by 5, the positional error is reduced with the same factor. At a belt tension of 516N, about 90N would be available at the tool tip (350mm off axis center). At 100N of tension at the belt that would only be 18N, so way less than the 18kgf minimum recommended above by Peteeng.

3. Wood drill bits typically have this pointy front. It might be possible to start the hole perpendicular to the material, to avoid the drill from slipping away, and then change to the correct orientation.

4. Use a tool changer and start off by locally milling the surface perpendicular to where the hole should come. The cost of this would probably be prohibitive though.

A combination of increased belt width, moving slowly when starting the hole, starting perpendicular to the timber surface, and a shorter tool will probably be best. It is also clear I should keep the router head as close to the axis center as possible.

Stepper torque/speed:

I've also had a look at the torque/speed curve for a possible Nema 23 motor (Leadshine D57CM31, see graph in attachment). The drive pulley (20 teeth) needs to rotate 6,5 times for one rotation of the driven pulley (130 teeth). If a full rotation of the big pulley was required in 1 second, the stepper would reach about 400rpm and torque would drop from 3,1Nm to 2,5Nm. This might be important when changing orientation while drilling.

Best regards,
Jim

References:

Mectrol guide: http://www.mectrol.es/media/catalogo...33565941b0.pdf

Hi Jim - Look up the ATL series of belts, much stiffer then the HTD (except in 10mm) and designed for positional applications. They have a better tooth fit then belts designed for power transfer.....They are in the Mectrol guide....Peter

Thx Peter,

I've put the calculation for axis accuracy and torque in a spreadsheet and am updating the design to get enough force and accuracy at the tooltip.

https://github.com/inatic/4-5th-axis...iming-belt.ods

Best regards,
Jim

Hi Jim - Just remember that accuracy at full force is not that important as this condition will be for roughing not finishing cuts. But it does need to be stiff so it does not vibrate. Finishing cuts need to be stiff and accurate. Finishing cuts will be light on force but if the system is not stiff enough it won't cut but skip along the surface. cheers Peter

Hi,

Could you have another look at the changes made to the design and say what you think?

I've tried to get adequate force while keeping belt elongation to a reasonable minimum.

Construction:
* aluminum plates are 20mm thick now
* pulleys and belts are 20mm wide as well
* driven pulley diameter has been increased to 200mm

Drive:
* motors have been moved out of the way
* nema 34 motors are used because of the force applied by the belt on the motor shaft

Both drives (plates, motors, belts and pulleys) are identical to limit the number of unique parts. At the moment all belts are 5mm GT2, this will change to a type better fit for positioning.

Best regards,
Jim

FreeCAD design: https://github.com/inatic/4-5th-axis...xis-v0.2.FCStd
Torque and error calculation: https://github.com/inatic/4-5th-axis...iming-belt.ods

Hi Vum - You may not have enough teeth engaged to transmit the full torque. The mechtrol design doc has guidelines for teeth engagement. You also may need a tensioner. I've found you can't achieve enough tension just by hand. or you need a notch to put a screwdriver in to use it as a lever to tension belt. I use a bracket and a bolt to push the motor with. Peter

Hi routalot,

The 5th axis motor has been moved out of the way. It now integrates in the plate and is indeed one less thing to keep track of. The new version can be found on Github if you like to have a look.

I was planning to generate the G-code for operating the machine right in FreeCAD, but will have to get back to you on exactly how.

For now I have a FreeCAD script in which each beam is represented as a Python object. A direction vector represents the main beam axis, and each hole is represented by a position and an orientation vector. Beams will be fed lengthwise into the machine, so the transformation of the beam's main axis to the machine axis will be applied to the hole vectors as well. As such I suppose the angles for C- and B-axis can be obtained by projecting the hole's orientation vector onto the XY- and XZ-plane of the machine.

What kind of G-code is necessary for 5-axis movement or how to generate it will need to be figured out shortly :)

Best regards,
Jim

FreeCAD design: https://github.com/inatic/4-5th-axis...xis-v0.2.FCStd

Hi Peter,

Thanks again for your advice. The driver pulley had to be increased from 15 to 18 teeth to transfer the full motor torque to the belt, and I've also added a tensioner to each of the motors.

Having the pulley directly mounted on the motor shaft, the maximum radial load for the shaft seems to be a limiting factor as well. For a Nema 34 motor, Leadshine specifies this maximum to be 300N. For anything higher than the current motor torque (which transfers 314N to the belt) the driver pulley will need to be on its own shaft.

Greetings,
Jim

Radial load: https://www.leadshineusa.com/UploadF...M%20Motors.pdf
Updated drive calculation: https://github.com/inatic/4-5th-axis...iming-belt.ods
Updated FreeCAD model: https://github.com/inatic/4-5th-axis...5th-axis.FCStd

Hi Awerby,

I'm afraid you're right, holes won't be accurate without preparing with another tool. I'll start the hole with a plunge bit or at least mill the surface so it's perpendicular to the hole, and then go in with the drill. This means switching tools for each beam but that's fine. Ripping and gluing would be possible as well but difficult for a big number of beams.

Greetings,
Jim