Rotary Axis - Trials and Tribulations

I am currently working on adding a rotary axis to my Burly.
The intent of this topic is to share my trials and tribulations and possibly pick up some advice along the way. There are other threads on the topic, but nothing particularly recent that it felt appropriate to tag on to.

The Why:
Really more of why not. Personally I enjoy building capability more then exercising it.
One of my kids is interested in making several magic wands, which was adequate motivation for me to give this a try.

Prior Art:
The main projects I have leveraged / remixed are:

There are others I reviewed and have probably taken ideas from, but I like the leo69 collet and decided to do a gear drive instead of belt reduction.

MPCNC Interface:

  • I am running dual end stop Marlin on a Ramps board.
  • I simply disconnected the X axis steppers and plugged the rotary stepper into one of them.
  • I use M92 to change the steps / mm for the axis to reflect the number necessary based on the work circumference.
  • I lock the x axis in place over the rotary axis using clamps from the Primo part release.
  • My workflow likes z-zero at the center of the work piece, so I touch off the OD and use G92 to set the z position at the radius.

Progress:

  1. First step - print leo69 parts:
    The rod I want to use to make the wands is 1.25" diameter, so I remixed one of the collets for this size. I printed that and the collet body & nut.
    Learning:
    The parts printed great, fit together well and work very nicely, but this thing is huge for my purposes.

  2. Rescaled leo69 to 80% & add integrated herringbone drive gear
    Arbitrarily chose 12T / 50T reduction.
    Learning:
    Much closer to size of my intended work.
    Runout caused periodic backlash in gear mesh.
    Backlash didn’t cause a significant issue in the quality of my test cut, but I lost steps on the rotary axis, so my 4.16:1 reduction may not be adequate.
    Overall height of axis is tight with my 3" z travel. May consider a drop section in my table, but it works for now.

  3. Tailstock support

  • Tried a simple under rod bearing (simplified version of something like this), but it didn’t feel like it was going to provide enough lateral end control with tooling pressure.
  • I printed a second leo69 collet and support block and omitted the drive stepper. This worked fine, but is not practical for change over with the rod inserted into both ends.

Software / workflow:

  • The wand file I tried came from thingiverse, but I modified it in Meshmixer to have a little more girth.
  • Aspire has a great “Wrapped Job Setup” Gadget. I was able to import an stl and create toolpaths that worked fine with the standard post processor.
  • My access to Aspire is through a work account and I’d rather not use it for personal projects. It is well worth the investment for someone that is going to use it regularly, but a bit expensive for me to justify for my home use. So far, I consider my use a training exercise with potential future work application :wink:, but for moving forward, I wrote a crude python script that does the “unwrapping” function so that CAM package that can import an stl surface can be used to toolpath.

Current status:

  • Rescaled leo69 design to 50% (minimum practical size for my work piece).
  • Added 4 bolt pattern for attaching an 8mm flange coupling.
  • Purchase geared planetary stepper & designed a mount. Settled on a 20:1 one from stepperonline, as it was on clearance and fairly reasonable.
  • Designed, but haven’t built a double bearing tailstock similar to ZenziWerken’s with a relatively easy to adjust extending live center.

Next Steps:

  • Hopefully my latest config will yield a relatively clean physical part at which point I will share an actual picture.
  • If there is interest in any of my files I will share freely as time permits.
10 Likes

This is awesome. I plan to do the same thing eventually.

@caffgeek Don’t do exactly the same thing:

I think I just ruined my planetary stepper.
I was all excited to attach a picture of progress tonight, but alas the apropos title of my initial post.

Everything appeared to be going great. I was dialing feeds and speeds and trying different roughing patterns: (raster Y & raster A).

Then, it appeared to stall on a rotary axis move at one point and I tried to restart.
After that, on rotation moves the planetary stepper was making bad noises and moving erratically.
I am pretty sure I screwed something up internally.

I can’t say for sure, but I think the tailstock mount I made may be putting too much axial load on the planetary shaft. I didn’t think I cranked down on it too much, but I don’t know what they are using for bearing. It is also possible that something just came loose inside and a sun gear is slipping on a shaft.
Regardless, no new wand tonight.

I guess its back to the drawing board on either the tailstock or drive system. I think a hybrid of my first and second tailstock concepts has some merit, as does a belt drive (maybe that’s why belts are common).

On the bright side, I get to take apart the planetary gear drive and see how they implemented that.
Good thing I’m doing this for the learning opportunity…

2 Likes

I’ll be following your progress to steal the ultimate design when you find it!

2 Likes

Cool project. I’ll be watching to see what I can learn from your progress as well.

I do have some doubts about the wand project though. My understanding is that wands require a core of some sort of unobtainium type material. Phoenix feather, Unicorn tail hair, sort of thing. Lacking a proper core, I would expect limited functionality at best…8^)

4 Likes

I have heard that llamas are the new unicorns. Maybe just some llama fur?

1 Like

My doggie is pretty special - even if not a magical creature himself. I’ll have to see about incorporating some of his fur.

The actual magic is supposed to come from encapsulated smoke.
The first incarnation of the project was a 3D printed case for an arduino nano. It had 3 buttons which fired an IR LED with remote control codes. It had an MPU6050 connected, but he never got around to writing code to use it.

For this revision, he designed a PCB around an ESP8266 that added a battery charging circuit and haptic vibration motor. He has some functioning code to process motion from the MPU and “recognize” a spell from a spell book, and a crude webserver that displays it, but is still quite a ways away from his ultimate feature wish list. What you can get prototyped on a PCB for next to nothing out of places like JLCPCB is absolutely astounding!

The current package is about the same as the original nano based version but has a lot more punch and a lot fewer wires to deal with. I think it will fit in a 3/4 or 7/8" bored hole in the wand handle. Figuring out how to bore the hole straight so I can CNC the outside with a reasonable wall thickness is an anticipated challenge.

However, on to the CNC hardware development topic at hand.
Much to the prophetic title, I have good news and bad news.

Lets start with the good:

  1. Stepperonline customer support is awesome. They provided a very quick response to my inquiry about my motor woes. They also asked some probing questions about how it was being used / driven.
  2. My motor is resolved and I made it through a full cut (picture to come either here or in Things I’ve Made).

The bad - there is no cure for stupid.

  • I had calculated and updated my step/mm for the axis which went up quite a bit, but didn’t decrease the associated max feedrate.
  • I didn’t pay enough attention to the rapid feed-rate speeds.
  • I didn’t come through the gcode to find the offensive feed rate command where the motion problem started.

It turns out there was a gcode line for a pure radial move at a speed pretty close to the threshold limit of the pulse speed the arduino can generate or the stepper can handle.
I have done something like this (tried to overspeed steppers) before and should have been able to figure it out on my own, but the intermittent movement was new to me.

So, while I don’t seem to necessarily learn from my mistakes, fortunately the max feedrate settings in Marlin do work and putting a reasonable limit will help prevent a subset of my future blunders.

Oh well, at least I am back in business :slight_smile:

2 Likes

Lots of good info here thanks.

You have a CNC…You could mill your own PCB. 8^)

Actually if you mill it then it’s not really a “Printed” circuit board (i.e. it’s not photo-etched), but maybe an MCB (milled circuit board).

We did mill the first prototype board to test some of the components / circuit ideas.

Getting a really nice V-carve sign requires some effort to get the work flat and level, and then the bit height has to be just right. Milling PCB is way harder. Fortunately they are pretty small, but the copper clad boards I bought weren’t super flat and trying to register double-sided traces accurately enough was a challenge. I should troll and see if someone has some tips to get the kinks worked out.

I also did give a go at etching, but getting a good mask wasn’t a ton of fun either.

I am not sure how they do it, but JLCPB is ridiculously cheap and the quality is amazing. The shipping charges have gone up astronomically since the Chinese government stopped subsidizing but the boards are essentially free (~$2) with $30 shipping. His first order about 10 months ago was like $10 with delivery.

Anyway, now we have to figure out how to get the electronics inside.
He’s got the board down to 19mm wide, so I bored a 3/4" hole in a dowel and got it pretty straight and it fits nice.
I am not sure if it would be better to CNC profile the shape before or after the core hold is drilled.
All along I figured we would end up with some kind of end cap, but the more I look at the thing the more I think I want to split it near the junction between the handle and the stick part.
But there again, I’m not sure if it would be better to:

  • CNC as one piece
  • CNC as two separate pieces
  • Split stock in two, but then assemble and cut as one piece. They could be assembled after the electronics are installed and permanently bonded together before the finish shaping, or just temporarily if I can come up with a way to assemble / disassemble securely without glue.

Or maybe we’ll just loose interest…

1 Like