Sunday 18 August 2019

Fusion 360 post processor - create a post processor for Centroid with probing??

This must the the definitive rabbit hole. I've been aiming to attempt a modification to the existing post processor that is provided for use with the Centroid controller. Fusion 360 now implements a fairly comprehensive range of workpiece probing functions for use within a machining operation. The issue is that the existing Centroid post processor doesn't incorporate the required content to handle those features. Most of the main, generic posts have (by definition) a large user base, so one way or another, the effort has been applied to create the required modifications. Centroid on the other hand is more of a minority sport and as far as I can tell, there are no plans afoot to do the required work.

I never expected it to be a simple matter but the more I get into it, the more convoluted the challenge reveals itself to be. Hence the Lewis Carol reference.

The scope:
Here's how I saw the task, around 2 weeks ago, as posted on the Centroid forum:

This may be naive on my part but I'm looking at the possibility of creating a modified post processor to enable Fusion 360 probing. There are several Fanuc-based posts available that support Fusion probing and I have been digging in to the details to understand what I'd need to do to create a Centroid equivalent. Looks to me as if there are several elements to it:

* Replace the basic probe move macros (G38.2, G38.3 etc) with equivalent Centroid macros. Can I simply use M115 and/or M116? Can I access / edit M115 / M116 if they aren't quite right?
* Modify the higher level Fusion probe macros (slot, boss, corner etc) that call them. That sounds less challenging unless I misunderstand. I have been given Fusion-Fanuc versions that implement the required functions.
* Re-reference any called system variables from the Fanuc numbering to the Centroid numbering
* Copy over / modify the post processor probing content into the Centroid post to create a "Centroid with Fusion probing" post.

From my current position (of massive naivety?) it's looking almost achievable. One critical element though is replacing the Fanuc system variables with the equivalent(?) Centroid ones, so they can be called. In Fanuc Custom B, the variables are typically in the 5000 range. From the list of variables in the Centroid probing folder, I see a section at the top named "return values and modals". This looks promising. Could somebody help to confirm / clarify for me what they mean? My questions are in {brackets}.

#34000 ; Probing feedrate {this seems fairly obvious}
#34001 ; Clearing feedrate {is this the retract rate back to the clearance height?}
#34002 ; PLC bits start ;50001 for cnc11 6001 for cnc10 {so Acorn starts at 50001?}
#34003 ; Probe PLC bit variable {indicates the PLC probe input and whether it's NC or NO??}
#34004 ; Probe PLC bit tripped state {the probe input state?}
#34005 ; M115/116 probe input number {eg 7 for INP7? - not sure how M115 differs from the PLC input}
#34006 ; Last X probe point ; also single axis probe return values {at the moment of tripping}
#34007 ; Last Y probe point ; also single axis probe return values
#34008 ; Last Z probe point ; also single axis probe return values
#34009 ; Probe diameter {presumably this is read from the tool table in CNC12?}
#34011 ; comparison tolerance {what is this for?}
#34012 ; Last boundary crossed value (M227) {I can't see any reference to M227 in the manual)

Am I on the right track? Are Centroid about to release a Fusion probing post? Is there any obvious roadblock I'm not seeing?

Finally, are the values for feed and distance in the default metric / imperial units? Questions, questions...

So I started out by printing out the various macros and post processors that seemed to be most relevant to the case i hand:
  • Fanuc generic post (which incorporates probing functions)
  • Tormach / PathPilot (LinuxCNC) post (ditto)
  • Centroid latest post (as discussed, no probing functions at all)
Given that Centroid have recommended the use of Notepad++ for some time now (and continue to do so) - and Autodesk initially recommended it too, I started out using it to edit, print and notate these puzzles. It's certainly better than a std, simple text editor such as Notepad but doesn't come close to the more modern alternatives such as Brackets or (more compellingly) Microsoft's Visual Studio Code, which is both free and available with Fusion-specific support for the Javascript code that the Fusion post is written in.

Post editing options - Visual Studio Code:
The required post processor editing stuff is linked to in the Autodesk HSM website, including the Fusion-specific extension. The easiest way to install the extension (in my view) is via the VSC environment itself, by searching in the extensions marketplace for "HSM" and grabbing the post editor extension from there. On my workshop PC I found I had to go into extensions and enable some of the options for running the loaded post when an intermediate file is loaded.

There's some VSC-related guff here too. For Javascript-specific info, there's a VSC page here.

If you prefer a video overview of VSC and its use here, this is the link.

The vital resource for anyone believing they are about to attempt edits to a post processor is the 180 page Post Processor Training Guide from the HSM team

Editing posts with VSC and the HSM extension:
Once you have set up the JS extension provided by Autodesk, you need to install a dedicated post processor that supports VSC. This generates an "intermediate" file (with a ".CNC" extension) that can be used by the JS editor. 

In VSC, this file appears in the left hand column in the "CNC SELECTOR" panel under the "CUSTOM" area. Alternatively, there is a range of standard intermediate files preloaded there which can be used to test out post processor edits. Along with the active post processor (shown in the middle panel), you can generate g-code output in the right panel (press Ctl-Alt-G). 

This is where things start to get really horny. If you select a line in the g-code, the editor will highlight the line in the post that generated it.

This is a lot smarter than Notepad++. However, there's simply no avoiding the need to digest the Post Processor Training Guide, all 180 pages of it. That'll shut me up for a while. It's almost enough to drive me to working on the bathroom again....

Parametric (re)model for motor housing

What up, Fatty?
So, one or two improvements came to mind last time round. I've been busy building an ensuite bathroom in the house, so had little in the way of serious workshop time. 

A couple of weeks ago I set to on the motor / drive housing (yet again), hopefully for the last time. There's only so much you can bugger about but when you sit back and look at what I've created so far, although it looks pretty nice, when you ask why certain features are the way they are - and more to the point - what purpose they serve, I must admit there are one or two legacy features that complicate the design and its production while adding no obvious functional benefit. That's one of the problems with humans.


The housing is currently dimensioned to suit the Leadshine closed loop stepper I bought several years ago (iSS57-20). I'm aware that there is no clear standard for the flange and register design, despite all the NEMA etc stuff. So the chances are that most other motors of this approximate size would require different dimensions. I have also be buggering about with different pulley and bearing sizes and belt lengths. So why not use "User Parameters" to control the dimensions in Fusion 360? That way, it should be possible to change some of those variables without having to dive deep into the design history to find where they are defined. Obviously there will be a practical limit on the range of variation possible but nonetheless I should be able to accommodate some.

At this point, The Stupid Fat Bloke got into the CAD seat and predictably got rather carried away, trying to parameterise(?) almost every possible dimension near the thing, based on a single master sketch. Obviously that was never going to work, due to the sheer number of lines and constraints at play. This became evident when it came to creating extrusions etc from the master sketch. And as for the idea of creating offset shapes from chains of segments, forget it! Some happy(?) medium exists between a single master sketch and a whole series of sketch / extrude / sketch / extrude operations.

Here's a "middle way" - a fair degree of parametric dimensions for the core model, which will be extruded and then modified later:

And an initial parameters table:

A bit more buggering about...

...and this is what I've got now. Sort of simpler in many areas.

Top view:

It's mostly parameter driven and has lost quite a few silly (unnecessary) features. I need to finalise the key measurements such as the distance of the ballscrew from the spindle (currently 50mm but almost certainly an imperial 50.8mm?) and the motor dimensions. A couple of features still missing such as a mans of fixing the lid. Otherwise, seems like a reasonable result for now.

I've also reduced the bearing size to a 4200 (30x14x10 double row), which gives me more modesty space between the bearing and the quill, while still having plenty of margin on the max loadings etc. Combined with an 18t driven pulley and a 10t driving (motor) pulley, I've reduced the belt length. This in turn reduces the size of the housing slightly. 

Monday 5 August 2019

10t pulley arrives - trial assembly

Finally (a couple of weeks ago) the new 10t XL (3/8" toothed belt) pulley arrived from China. This has a large (relative to the actual pulley) diameter boss which features 2 grub screws. This solves the question of how to lock the pulley to the motor without having to resort to Loctite or similar. I wouldn't have a big issue with Loctite apart from the challenge of trying to remove it if the need ever arose. Heat is the answer but that might not be so good for the rotor and bearings, so best avoided.

Here they are. 8mm bore fits the motor shaft and the max diameter is small enough for it to fit through the hole (= slot) in the housing.

So now I can see how well it all goes together. I'm planning on having some adjustment on the radial position of the ballscrew relative to the 16mm hole in the head casting but in the event, I made some of the dimensions close to the nominal and will need to do some finessing before it fits together as I intended.

Pressed the new bearing into position - wasn't too tight a fit and I could alsmost imagine being able to press it out again if required. I'm wishing I'd gone for the sealed version of these bearings, as they are at risk of attracting dust. Obviously I left the procurement to The Stupid Fat Bloke, with predictable consequences.

I can now see (access) the bearing outer race, so it would be possible to turn up a drift if / when I need to remove the bearing later, rather than try to drive it out using the inner race which would most likely bugger the bearing entirely..

The 2 mounting holes have been opened out to 6.8mm. Given that these are for M6 caphead screws, they should never have been 6.0mm to start with. Furthermore, as I plan to allow a fair degree of slop in these fixings for final alignment of the ballscrew, they should end up closer to 7.0 or even 7.5mm diameter.

Now I can trial fit the ballscrew, sleeve and ballscrew retainer nut and see how it plays with the head casting. My design concept would allow for the motor / bracket / ballscrew / bearing / belts to be pre-assembled together, inserted into the head casting and then fastened in place with these 2 machine screws. The aforementioned slop would then allow the assembly to be aligned correctly with the ballnut so that the ballscrew doesn't end up going out of vertical as the quill moves.

Finessing required:

  • Open up the holes for the M6 fixings, as described above. Possibly also the counterbores for the capheads.
  • Reduce the diameter of the ballscrew down to 15mm - up to a height of 15mm above the casting. This will allow some radial slop on the ballscrew within its bore, without which it will be difficult / impossible to insert the ballscrew and its assembly into the head. Obviously I need to avoid compromising the "active" part of the ballscrew where the balls will do their thing but I know I have some margin here. I'll need to do some measurements and sums to figure out what can be achieved safely.
  • Reduce the OD of the ballscrew where it sits within the casting. Until now I've done this by fitting a thin walled sleeve but it would be better simply to reduce the shaft. As well as reducing the part count, it would have the added benefit of allowing a beefier shoulder for the bearing to abut against. The sleeve is only required to support the bearing preload.
  • Provide a (turned) concentric feature at the "top" of the ballscrew. This would be nonfunctional but would be useful when it comes to aligning the ballscrew. There is already an M8 threaded hole for the ballnut keeper at that end.
  • Provide a bit of clearance between the housing and the head casting. Otherwise, we won't be able to take advantage of the slop. Currently it sits right against the vertical face
  • Draw up a drilling jig for placing the two M6 tapped holes in the head. Getting these halfway right is somewhat helpful.
  • Make the ballnut bracket narrower where it pokes through the head casting onto the quill. I have left minimal clearance in my desire to reduce any rotational movement of the quill. I measure about 16.10mm across the bracket feature and closer to 16.00mm across the slot. I'm guessing it's designed to clear a 5/8" feed stop, which would be about 15.87mm or so.
  • Seems pretty clear that the ballscrew is still positioned too far from the quill. That's down to the bracket dimensioning. I'll need to check how the ballscrew / ballnut / yoke line up with the bore of the head casting to see if the yoke is OK or the problem is confined to the housing....
  • Yoke (ballnut bracket) - round off the top to maximise the allowable vertical travel. A square body will stop 8mm from the top of a 16mm slot with rounded ends. In fact the quill stops with the fixing screw centre ~13mm from the top of the slot, which coincides with the top of the bracket reaching the top of the slot. I simply need to round off the bracket feature to a radius of ~8mm.
Should be pretty quick and simple, then.

Final assembly and test of the spindle nose adaptor - RESULT!!

After the recent distraction caused by the 3D scanner, resurrecting the 3D printer and buggering about with the throttle bodies for my Honda...