How to build a CNC 5th axis for a small milling machine. 

Thanks !

This Could be an interesting project from start to end , a full cnc .

Thanks ! I'll take a look at the model. It's a bit odd, as the assembly looks fine in Fusion 360. I'll let you know when it's fixed.

@@girvaw It might be easier to share the a360 file. There still seems to be an issue with the Grabcad model. It seems to be missing components that the model references.

Thanks Yes, I did measure the c-axis flex, but forgot to record the footage! The flex was minimal even with quite a bit of pressure being applied.

Bonjour, Normalement oui. En revanche, avec l'opération "Swarf", on peut attendre qu'après le positionnement des axes x, y et a, on verrait uniquement le mouvement des axes z et c. Cela dit, Autodesk a beaucoup de savoir-faire dans ce domaine et la qualité des surfaces est évidente. J'ai pris des petites passes parce que c'était le première essai. Normalement, on prendrait des passes plus profonde et plus rapidement. Cela dit, c'est une petite fraiseuse avec une faible puissance (500W). Je n'attendais pas de problèmes avec l'usinage de l'acier sur le 5eme axe - il est super rigide.

Hi. The servos used are harmonic drive-based. Direct-drive stepper motors would not be suitable for this type of application. Perhaps a stepper with a toothed-belt reduction might be feasible.

@@girvaw Sir, I am asking about the stepper motor on the Y-axis 11:18 , not the servos. What is the configuration of the stepper motors?

@@mmnyako Steppers are all NEMA 23s. The drivers are set to 1/8 micro stepping. I don't have torque figures for the steppers. Supply voltage is 30v.

Thanks. It's quite straightforward with LinuxCNC? Parallel port pins are allocated for each axis. The servo motors of the A & C axes are in stepper control mode, so they take step and direction signals directly from the parallel port. It all works really well.

@@girvaw Got you - i was referring to using this as a machine that either drip feeds commands through a software program or X and Y co-ordinates created by a software program then fed into the machines controller via a USB stick.... I guess you manually programming each step and pausing between each command change between the mill and then the added 4th and 5th axis? To get this to work simultaneously would you be able to achieve this?

@@TurboDirectSA the toolpaths are generated from the 3d modelling software - Fusion 360 in this case. All of the axes can move simultaneously. This is dependent on the gcode generated by the Fusion 360 post processor in question and the CNC controller. See Tool Centre Point control (TCP) for further reading. The 4th & 5th axes could be driven using their native API, however in a CNC context this would not likely be practical.

@@girvaw So are you using a single controller board to manage all 5 axis at the same time? Like a Fanuc control or Siemens? I understand Fusion360 will output the code which i assume you are either drip feeding to the machines controller (which controller are you using?) or you plug a memory card into the controller - but either way you using a controller of sorts - what are you using for this?

@@TurboDirectSA axes X, Y & Z have their own stepper drivers which are driven by LinuxCNC via parallel port pins. A & C don't need drivers, as they have TTL-level inputs that can be driven directly from the parallel port. See LinuxCNC docs for an explanation of how this works.

Hi. I am using a paid version of Fusion 360 now. However, when I built the 5th axis, I was on the (free) startup version. As I mention in the video, the startup version I was using only had access to a subset of Fusion's multi-axis operations. For the full suite of operations, the manufacturing extension module is required. This is unfortunately rather expensive. Good luck with your project.

Thanks for your feedback! Firstly, I should note that I have an interest in Robot Articulation. The servo motors used in this project are Mercury M30 models. They are currently only being released to early adopters. Unit prices are TBD, but the aim is to have them very competitively priced. If you wish to stay informed, contact Robot Articulation via the contact page on their website. If you're in a hurry, there are alternatives: - Dynamixel Pro servos have cycloidal bearing outputs, so should have very little backlash. I'm not sure how much axial and radial forces they can tolerate. They are however expensive. I've used both AX and MX series Dynamixels in the past. I found that the metal-geared MX versions had significant backlash and were nowhere near rigid enough for this type of application. - Harmonic Drive LLC produces various models of motor+controller+harmonic-drive in a single unit, but again they appear to be quite expensive. Discrete harmonic drives could also be used, but the combination of a motor+harmonic-drive would probably necessitate a much taller a-axis centre of rotation. - Another alternative is to use steppers with some type of belt-drive mechanism. At the end of the day, harmonic-drive based solutions are ideal for this type of application due to both the lack of backlash and the rigidity provided by the heavy-duty crossed roller bearing output. Hope this helps.

you can find a 80mm bldc drive with 30Nm and harmonic drive for $400 on aliexpress. 30RPM.

Thanks ! As far as I'm aware, xyzac-trt doesn't support TCP out of the box. There's some commentary on this on the LinuxCNC forum here: forum.linuxcnc.org/10-advanced-configuration/31813-tcp-5-axis-kinematics?start=120#139881 This is a subject I'm going to be looking at in the near future, so keep an eye on my channel.

Thanks ! I'll let you know when the 3d model is fixed and re-uploaded to Grabcad.

Hi. There are no brakes on any of the axes.

Thanks. The constant movement of the y (and X) axes is a result of the toolpaths generated from Fusion 360's multi-axis Swarf operation. I don't know the rational behind this. That said, the surface finishes produced by these operations were very good.

@@girvaw Thanks. Can you read the gcode? It would be interesting to see if x and y were being driven as we see. Another question if you don't mind. Can the 4th and 5th axis be manually backdriven? If so how much? I'm retrfitting a small lathe at the moment and I only use servos. These are easy to backdrive which is a downside compared to steppers which have max torque when stationary. Not that I have ever had any problems. my existing lathe has servo driven spindle and I cut gears and accuracy is not an issue.

I'll have a look at the gcode when I have a moment. Regarding the back-drivability of the servos, with no power applied, the back driving force is specified as 5 Nm. See robotarticulation.com/pages/product_mercury_m30.html Backdriving when running would be very unlikely, as the servos are closed loop with high stall torques. They do however have a torque control mode, which would allow this, but this mode would not be used in a CNC application.

@@robguyatt9602 Here's a snippet of the Swarf operation's gcode. As you can see, there's even an oscillating movement of the A axis. It would be interesting to understand the mechanics behind this (seemingly) unusual tool path. N100 X-0.286 Z25.793 A45.002 C147.609 N105 X-0.285 Y-68.827 Z25.79 A45.001 C150.008 N110 X-0.277 Y-68.826 Z25.791 A44.997 C152.416 N115 X-0.262 Y-68.825 Z25.79 A44.996 C154.835 N120 X-0.237 Y-68.826 Z25.785 A44.997 C157.271 N125 X-0.195 Y-68.827 Z25.778 A45. C159.733 N130 X-0.106 Y-68.828 Z25.774 C162.205 N135 X-0.133 Y-68.829 Z25.769 A45.002 C164.559 N140 X-0.157 Y-68.828 Z25.767 A45.001 C166.916 N145 X-0.18 Y-68.827 A44.998 C169.277 N150 X-0.199 Y-68.825 A44.996 C171.642 N155 X-0.217 Y-68.826 Z25.762 A44.997 C174.011 N160 X-0.231 Y-68.827 Z25.755 A45. C176.385 N165 X-0.242 Y-68.828 Z25.75 A45.002 C178.763 N170 X-0.249 Z25.745 A45.003 C181.148 N175 X-0.252 Z25.742 C183.539 N180 X-0.25 Z25.738 C185.938 N185 X-0.243 Z25.736 A45.002 C188.346 N190 X-0.227 Z25.735 A45. C190.766 N195 X-0.202 Y-68.825 Z25.738 A44.994 C193.202 N200 X-0.16 Y-68.826 Z25.734 A44.995 C195.664 N205 X-0.071 Z25.729 A44.997 C198.136 N210 X-0.098 Z25.726 A44.996 C200.49

@@girvaw That's crazy code. I don't know enough about it to suggest why. Perhaps a not so good post processor? The fact that the servo has a backdrive torque almost the same as the rated torque says to me it is pretty dam good.

Bonjour Bryce. You can find the servo motors at www.robotarticulation.com/

@@girvaw thanks mate I’ll check em out. 😊

They are both M30 models.

@@girvaw What was the approx price of the motors? I'm looking into making my own and I really like your approach.

The 3D modelling was done in Fusion 360.

@@girvaw and Mach3? or?

@@Arjan-swets Linux CNC

Hi. No, I just made it for use in my home workshop. It has turned out to be very useful for a variety of tasks.

@@girvaw could you make me one I pay for it as long as I can afford the price. Thanks

@@walkbehindtrimmer_attachment Hi. Thanks for the proposition! Unfortunately it would not be viable to produce these commercially with my lightweight machines. In addition, hand scraping is very time-consuming so a surface grinder would be required.

@@girvaw that will be ok and not a big issue - I can manage to surface my wast-board to be even and flat - me me one and I pay for it