A new Z-axis nut - metal on metal w/ built in anti-backlash

discussion of design changes / improvements / suggestions
ejs
Posts: 84
Joined: Sat Apr 28, 2012 7:42 pm
Location: Pittsburgh, PA, USA
Contact:

A new Z-axis nut - metal on metal w/ built in anti-backlash

Post by ejs » Sat May 25, 2013 4:27 pm

Before I jump in to the details, here's the finished product:

Image

One of my frustrations with the kit, and one that has been talked about at length, is the Z axis. The nut and threaded rod was binding up on the first run through the build. I spent more hours than should be required trying to get the rod to thread through cleanly. The supplied rod isn't exactly straight, maybe a four degree arc through the length of it, which compounds the issue.

Add to this concerns that using something like WD40 risks penetrating and destroying acetal and I was just frustrated. Every now and again I hear the rod squeaking against the threads in the nut or I note that Z steps go missing once I push a vertical feed rate approaching 50 mm/min. The missed steps seem to happen more often during retraction than then during a plunge operation. This could easily be a function of additional material being scraped off inside the threads when the axis is under load. We know that material can be removed fairly easily from the nut since running the rod through with a drill to loosen it up is the prescribed action for a new nut that's too tight.

I love my 'Oko, but this is unacceptable for the operations I'm doing. I wanted a solution where the metal thread were pressing against metal so I (1) didn't have to worry about material being scraped away that may jam the assembly up, make things too loose, or both; and (2) if I hear a squeak I want to be able to grab whatever canned product I have at hand to be able to grease things up. I don't want to have to worry about the pitfalls of a polymer-metal interface..

So, the quest was on.

The starting point of the solution was to first consider an all metal Z-nut and the appropriate machining that would be required. After about a half hour that just seemed silly--really, all that needed to be metal was the threads for the rod. The rest of the acetal (Delrin) nut was polymer and was sturdy enough for the purpose of attachment to the gantry plate.

The real first part of the solution was in seeking out a metal insert for the 8 mm rod alone. I came to realize the best solution has been presented by the flat pack furniture industry. Cross nuts are the little round bits that you sink into, say, a half inch hole in the side of a horizontal member in a flat pack bed frame. You then put a bolt through a separate vertical member and it seats and threads into the nut. This allows great tension since the bolt is torqued to keep the wood together but is relying on a metal cross piece to distribute the force over the width of the plank. Google cross nut and you'll instantly recognize what I'm on about.

This seemed that it could be a complete solution, but I knew there was a danger. There was the probability that the threads of the cross nut were going to be more narrow than the tapped acetal, meaning that there will be slop, or backlash, in the assembly. After getting a sample of the nuts I realized this was the case. But even before they were in hand, I was sketching out a solution.

The specifics turned out to be backlash of 10 to 12 ten-thousandths (or just over 1/1000) when threaded in the insert. Good enough for a hobby CNC kit? Maybe. Good enough not to be improved on? Hardly.

The goal: a single piece block that would keep the nuts in place and eliminate backlash without the addition of more components. No extra metal or springs. No external tensioners.

Thinking back on my repertoire of dismantled products, I recalled the design of the rubber band shooter that I so coveted as a kid. I had to take it apart before understanding that the plastic the mechanism was made of was actually acting as a leaf spring to reset the trigger while keeping part counts down.

Below demonstrates the three generations the design progressed through. The obvious addition included a second nut--a necessity with the anti-slop design. Pressure between each threaded insert would serve to eliminate backlash. The design element that would provide this tension was to be a hollow between the two cross nuts. By compressing this space before threading the rod through, the polymer side walls would want to flex back to their original shape, and so becomes an anti-backlash nut.

Of course, this would take some testing.


v0.1

Image

This was the first crack at, well, cutting anything on my machine. Going at the polymer with too much bravado created this mess. It ultimately gooped up the WEN spindle to the point of locking up and fried the spindle's circuit board. The walls on either side of the hollow are meant to have just enough material that they will bow outward all spring-like. This iteration, besides being a mess, has a few other issues. First, the fit for the metal insert nuts is oblong--a reprecussion to trying to cut linearly too quickly and not having a second motor on the Y-axis.

The side walls are entirely too thick to flex.


v0.2

Image

The side walls on the second try are curved outward and are about two-thirds as thick as the prior. I thought the curve may help promote the outward bow I was looking for. Under a full squeeze, there is slight compression of the hollow, but not enough to really matter. Since M8 is 1.25 mm per turn, I need one to two threads worth of compression to make this work. This gap needs to exert a few pounds of pressure, but not too much otherwise we end up with friction issues similar to what we have with the original nut.

The aesthetic of the curve is nice. Still not enough flex here, either.


v0.25

Image

I took the same idea and then thinned out the walls to half of v0.2's thickness. I added a bit of thickness to the material at the center of the sidewall knowing that the center point would bear the brunt of the flex. My head was in the right place, but cutting speeds and tolerances were not. You can see that the right side split under pressure, though the left can stand nearly a millimeter of compression.


v0.3 is the Reactive Z-nut v1.0

Image

Just a note: One thing you likely noticed on all of these are the spaces for bolts and hex nuts at the top of the polymer body. I may eventually move to a place where these are tapped like on the original nut, but that would require added trial and error which I'd rather deal with at some later date. The spaces fit their nuts perfectly. A set of two 16mm M5 bolts and corresponding hex nuts is all that's needed to secure the assembly in place. The hole spacing is still the same.


As for the nut, I saw that with the minimal tension I wanted to produce over a few millimeters of travel would have to be spread out over more linear surface than I was allowing for in the designs so far. The ultimate solution came by my mimicking the rubber band shooter's leaf spring, but by serpentining it into a 2.5D imitation of a wrapped compression spring.

This was the winner. It allows flex over almost twice as much material as v0.2 without adding any height. Ditching the round profile and adding ears next to the lower cross nut allows points for easily gripping both sides when smooshing things together. Once assembled, we have an effective anti-backlash solution for the Z-axis with minimum fuss and no need to modify gantry plates. If you look at the image, you see that the assembly on the threaded rod is compressed by 2 millimeters compared to the empty body beside it. These were quick cuts, and my most recent version looks a great deal more like the render at the top of this post.

This was exactly what I aimed to engineer and am happy to have done it proper. I now give a quick squirt of WD below and above the new z-nut assembly and let the motion of threaded rod transport the lubricant to the inserts. Bully.
Build it better than it was built for you. And give permission for the next guy to do the same. That's how Open Source works.
ShapeOko # 497: http://bit.ly/reactshop producing the Buildlog CNC Stepper Shield

WillAdams
Posts: 8488
Joined: Mon Apr 09, 2012 6:11 pm
Location: Pennsylvania --- south of the Turnpike, East of US-15
Contact:

Re: A new Z-axis nut - metal on metal w/ built in anti-backl

Post by WillAdams » Sun May 26, 2013 12:44 am

Brilliant!

I wonder if it could still be included in the next iteration.

I added same links to this in the wiki. Any chance of a file?

Wouldn't it be possible to do this for an Acme screw as well?
Shapeoko 3XL #0006 w/Makita RT0701 Router w/0.125″ and ¼″ Elaire precision collets
Nomad 883 Pro #596 (bamboo)

ejs
Posts: 84
Joined: Sat Apr 28, 2012 7:42 pm
Location: Pittsburgh, PA, USA
Contact:

Re: A new Z-axis nut - metal on metal w/ built in anti-backl

Post by ejs » Sun May 26, 2013 1:49 pm

Well thanks, Will. I'm very happy with the functionality on this one.

I did try to find Acme equivalents of the cross nuts with no luck. When I realized how much this would improve the standard setup, I wasn't too broken up about it. I could do some redesign on this to incorporate Acme hex nuts instead--though one advantage to the setup is that there is about 20 mm of receiving thread, and I would likely need three Acme hex nuts to do the same. The more threads the less pressure on each individual turn. Acme would, of course, also drastically raise the cost of this upgrade. I was looking for something cheaper than the Acme upgrade that would still improve the overall function of the machine. But an Acme version is still worth considering. As an aside, I also found an arrow straight, 200 mm bolt (threaded along the entire length!) that will be replacing the allthread.

I would love to see this as a permanent upgrade to the machine. Thank you so much for the wiki edits. I'll either post an installation video or mock up installation in Sketchup this week and add it to the documentation. The hole size still needs a tiny bit of tweaking to make as tight a fit as I would like, but the file will be forthcoming. This is the start of a much more extensive z-axis upgrade setup, including shields, a new motor platform, and carriers for limit switches in addition to the new nut. All things that I either wish had been provided earlier or things that have caused a headache at some point since starting with my 'Oko. Stay tuned!
Build it better than it was built for you. And give permission for the next guy to do the same. That's how Open Source works.
ShapeOko # 497: http://bit.ly/reactshop producing the Buildlog CNC Stepper Shield

twforeman
Posts: 1347
Joined: Tue Jan 29, 2013 4:51 pm
Location: Minneapolis, MN
Contact:

Re: A new Z-axis nut - metal on metal w/ built in anti-backl

Post by twforeman » Mon May 27, 2013 2:00 am

Hey, that's really cool! I like this solution. Nice job.
Ender 3 3D Printer
ShapeOko v3 serial #0004 - upgrade thread
All of my ShapeOko related blog posts

OKShapeO
Posts: 7
Joined: Sat Oct 05, 2013 10:32 am

Re: A new Z-axis nut - metal on metal w/ built in anti-backl

Post by OKShapeO » Tue Oct 08, 2013 6:47 pm

Just came across this - really nice! Did it ever get taken further eg drawings? Seems to have just gone to sleep...

WillAdams
Posts: 8488
Joined: Mon Apr 09, 2012 6:11 pm
Location: Pennsylvania --- south of the Turnpike, East of US-15
Contact:

Re: A new Z-axis nut - metal on metal w/ built in anti-backl

Post by WillAdams » Fri Feb 14, 2014 4:58 pm

AFAICT, ejs never did post files for this --- did anyone try it? I'm wondering if it wouldn't be worth re-creating as one way to address the difficulties which seem to keep cropping up w/ the Z-axis on the SO2.
Shapeoko 3XL #0006 w/Makita RT0701 Router w/0.125″ and ¼″ Elaire precision collets
Nomad 883 Pro #596 (bamboo)

WillAdams
Posts: 8488
Joined: Mon Apr 09, 2012 6:11 pm
Location: Pennsylvania --- south of the Turnpike, East of US-15
Contact:

Re: A new Z-axis nut - metal on metal w/ built in anti-backl

Post by WillAdams » Tue Feb 18, 2014 7:10 pm

Okay, I went ahead and drew up a file for this:
ejs_zaxis_nut-svg.txt
(5.62 KiB) Downloaded 123 times
(drawn up in Adobe Illustrator, opened the .ai file in InkScape, saved as .svg, then renamed .txt)

[img]ejs_zaxis_nut.png[/img]

Based on comments, and the 20mm spacing for MakerSlide (does that need to be adjusted for the SO2?) and 8mm thread dowel nuts seeming to have a typical diameter of 12mm. YMMV, offer void where prohibited by law, &c.
Attachments
ejs_zaxis_nut.png
ejs_zaxis_nut.png (14 KiB) Viewed 5752 times
Shapeoko 3XL #0006 w/Makita RT0701 Router w/0.125″ and ¼″ Elaire precision collets
Nomad 883 Pro #596 (bamboo)

cvoinescu
Posts: 4442
Joined: Thu Jul 19, 2012 6:50 pm
Location: Camberley, UK
Contact:

Re: A new Z-axis nut - metal on metal w/ built in anti-backl

Post by cvoinescu » Tue Feb 18, 2014 7:33 pm

Will, the flexible part looks too thin. For anti-backlash action, you want a stiff spring -- something that can apply more force to the lower nut than the largest expected force in the "up" direction. You have the weight of the spindle to help, of course, but the "up" force can be greater than that; the rest of the "up" force must be resisted by the spring, or you lose the anti-backlash feature.
Proud owner of ShapeOko #709, eShapeOko #0, and of store.amberspyglass.co.uk

WillAdams
Posts: 8488
Joined: Mon Apr 09, 2012 6:11 pm
Location: Pennsylvania --- south of the Turnpike, East of US-15
Contact:

Re: A new Z-axis nut - metal on metal w/ built in anti-backl

Post by WillAdams » Tue Feb 18, 2014 7:57 pm

Okay, I inset the inner path by 1mm to double the thickness:
ejs_zaxis_nut-2mm-svg.txt
(5.22 KiB) Downloaded 149 times
--- I guess one could make further adjustments in MakerCAM when calculating the pocket / profile.

I'll try to pick up some nuts and see if HDPE is suitable for it or no --- if anyone cuts it, I'd be interested to know how it works.
Shapeoko 3XL #0006 w/Makita RT0701 Router w/0.125″ and ¼″ Elaire precision collets
Nomad 883 Pro #596 (bamboo)

LouisV
Posts: 316
Joined: Fri Dec 13, 2013 3:41 am

Re: A new Z-axis nut - metal on metal w/ built in anti-backl

Post by LouisV » Fri Mar 07, 2014 4:52 pm

I would be very interested in an Acme version of this. If anyone makes a successful one that works well I will gladly buy a few from them.

Post Reply