Still getting Z wobble with ACME lead screw

When I rebuilt my printer, I decided I would use lead screws instead of normal threaded rods because I wanted to get rid of the Z wobble artifacts. It didn't help - they still appear, although to a lesser degree. The problem, I think, is that I went the cheap route. I bought plain steel 1/4-16 acme threaded rod with matching hex nuts and used a normal prusa-style set of X ends, complete with the internal compression spring. I wanted to go with a trapezoidal screw, but I was talked out of it

Now I'm trying to figure the best path froward from what I have. I'm thinking of printing some new X ends to accommodate a 1/4-16 flanged acme nut, but I keep thinking that I would be better off finding something in the trapezoidal family? I see an 8mm screw on terrawatt industries web site that I almost bought earlier (and now wish I had).

Does anybody have any alternate sources for trapezoidal screws and nuts? I looked at roton, but the smallest trapezoidal they have is 10mm. I guess that's not really a problem but I never considered a screw that large.

Anybody have any other advice?

maybe try these, they make the z axis move up only, by absorbing the wobbz of the zzzzz [www.thingiverse.com]

jbernardis, for the record, ACME lead screw IS trapezoidal form-factor. In fact, it's the first trapezoidal form-factor lead screw and the "trapezoidal" refers to the thread profile, nothing more. Don't let the EU rhetoric and dogma about metric vs SAE fool you. The ACME thread was invented in the US in the late 1800s and was later made in a metric form-factor. Some here choose to ignore the fact that trapezoidal is an American invention and that "trapezoidal" doesn't just include metric trapezoidal.

If you're in the North American continent, McMaster distributes precision 1/4-16 ACME lead screw in carbon, heat-treated 4140 and 304 stainless steel.

JB,

First, read this thread. Also consider moving to full-step layer heights if possible.

I'm using the McMaster 1/4-16 ACME screws. The came pre-bent from McMaster. Very nice of them. I highly recommend not purchasing screws from them (at least 1/4" anyway). I've bought (3) 3' screws from McMaster, each one pre-bent. Live and learn. If you've got 8mm smooth rods I wouldn't go up to a leadscrew larger than 8mm.

Misumi has an 8mm Trapezoidal screw and a brass flange to fit it. Part #'s MTSRA8-200-S10-Q6 or MTSR8-200 and MTSFR8 (the 200 is a guess of how long a screw you need). Then you'd need to find X-ends that would fit it (or design your own).

Like you mentioned, Roton does have Trapezoidal leadscrews, but not down to 8mm. If you beef up your smooth Z rods to 10mm you'd be able to go that route. But that's an expensive proposition...new rods, bearings and X-ends. Not to mention the $60 minimum. Meh.

Since you'd like to stick with your existing setup (much cheaper) there are a few things you can do.

First, print out these or download the one I've attached (the source file is there as well if you want to modify it). They will decouple the X-axis from the bent ACME screw by allowing the screw to move freely in the X-end pocket.

Second, remove any spring and second nut to eliminate backlash.

(Optional) Third, drill out the X-end with a 1/2" bit to remove the ribs that keep the nut from slipping through the X-end. The screw may be fetching up on these ribs.

Also, what couplings are you using? They may make all the difference in the world.

Before you re-install the X-end place a couple of nylon washers that you've lubed up between the X-Z decoupler and the X-end itself. Make sure the washers are faced the correct way - so that the rounded faces of the washers face one another. Better yet slip a teflon washer between these two if you can find one.

Here's a picture of my X-ends with the X-Z decoupler to illustrate.

Before I installed this setup my parts showed terrible wobble. Afterwards they look like this.

Not absolutely perfect, but I can live with it.

Good luck! Hopefully you'll get away with the cheap fix and not have to go to Trapezoidal screws.

Edited 1 time(s). Last edit at 03/31/2013 09:29PM by akhlut.

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

Thanks for the input everyone. I realize that acme and trapezoidal are the same thing, I was just using the term trapezoidal to imply metric. I've read somewhere that the math works our better for metric, but that is probably overblown.

I think I'm going to try ahklut's ideas. I will probably need to adapt the design since I have 10mm smooth rods, and while I'm at it I'm going to design new x ends and completely replace the nut channels with a simple circular hole.

Right now I'm away for a few days, but it looks like I have a project t when I get back home. Also, while I have the axis disassembled, I'm going to try nophead's belt flip trick.

ACME is trapezoidal, and Trapezoidal is trapezoidal.

But ACME is not Trapezoidal.

Just like all jacuzzi's are not Jacuzzis. See the diff?

One is an adjective while one is a noun.

Unless of course you'd like to argue with the good people at Roton, Misumi, Thompson, Nook, etc. You know, the people who make ACME and Trapezoidal leadscrews to standards (you know, EU hippie stuff).


xiando Wrote:
-------------------------------------------------------
> jbernardis, for the record, ACME lead screw IS
> trapezoidal form-factor. In fact, it's the first
> trapezoidal form-factor lead screw and the
> "trapezoidal" refers to the thread profile,
> nothing more. Don't let the EU rhetoric and dogma
> about metric vs SAE fool you. The ACME thread was
> invented in the US in the late 1800s and was later
> made in a metric form-factor. Some here choose to
> ignore the fact that trapezoidal is an American
> invention and that "trapezoidal" doesn't just
> include metric trapezoidal.
>
> If you're in the North American continent,
> McMaster distributes precision 1/4-16 ACME lead
> screw in carbon, heat-treated 4140 and 304
> stainless steel.

Edited 1 time(s). Last edit at 03/31/2013 10:04PM by akhlut.

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

Then you are in luck. I have these generated for several configurations. Too bad I don't post to thingiverse anymore.

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

BTW I never answered the question about my couplers. I am using 1/4 inch to 5mm aluminum couplers purchased from trinity labs. Also, with the design of my printer, the motors (and couplers) are at the bottom.

I was using the same ones. They weren't really doing it for me. Switched to a set of hubs with a plastic spline in between. Been much better ever since, but they're pretty pricey ($16 each). But don't switch out until you whittle the issues down to the couplers. On my second machine I went with nophead's couplings and silicone tubing to save some $$$. Even with only the top two screws on each coupling tightened down they work really well. The silicone tubing allows for quite a bit of flex at the coupler. Second machine has essentially the same setup as the first, but no nylon washers, just teflon ones.

But I've got 10mm smooth Z rods on both machines, and I've got my Roton 10mm Trapezoidal screws and flange nuts standing by. Now I just have to build some maintenance time into my machine and stop printing things for people. Kinda afraid to though - once something is set and works I hate to mess with it, even if it's for an upgrade. Gonna have to bite the bullet eventually though.

Also, another advantage of those X-Z decouplers is that you'll never have a destructive Z-crash ever again.

Did you round your Z-axis e-steps to a whole number? That has helped a lot as well. It imparts error into the height of an object, but it's so small as to be nearly un-noticed.

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

I hadn't thought of rounding off the steps value. You're right - it imparts some error but makes it consistent layer to layer. I might give that a shot.

You're full of good ideas - thanks.

I, too am having problems with slight Z wobble.

My set-up:
M5 screws
Cf53 hardened 8mm shafts
Bushings working as thrust bushings at lower Z end
Custom couplers

After analysing what might be wrong, it pretty much came down to how my X axis shafts are installed. The supports are too narrow and placed in a way that lets the X ends rotate around the Z shafts, leading to Z wobble as the rotation shifts the X axis around.
This is an inherent problem with a vertical X on which the X shaft supports are not aligned with the Z shafts, I guess.

Pictures of my printer to give you an idea what I mean since this isn't a standard build:



Edited 1 time(s). Last edit at 04/01/2013 08:39AM by uGen.

uGen,

that is a heck of a repstrap!

Have you tried it without the thrust bearing? 1/4" aquarium tubing would work like a charm with that M5 threaded rod as a replacement flexible coupling. Prusa uses it on the i3 and his prints are pretty stunning. But I'd try without the thrust bearing first.

You don't have a sprung backlash nut on the Z-drive rod, right? If so you might want to try it without it.

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

"I'm using the McMaster 1/4-16 ACME screws. The came pre-bent from McMaster. Very nice of them. I highly recommend not purchasing screws from them (at least 1/4" anyway). I've bought (3) 3' screws from McMaster, each one pre-bent. Live and learn. If you've got 8mm smooth rods I wouldn't go up to a leadscrew larger than 8mm. "

Then you should have sent them back to McMaster, since they were obviously received in a failed condition, instead of snarking about your own lack of due-diligence as if it was their fault. How nice of you.

Just a quick note. I bought spider couplers from love joy and they helped a lot with wobble. I wouldn't use anything other than spider couplers now.

---

--------------| For Everything |--------------------------
Check it out here:
[reprapsquad.wordpress.com].
---------| For Everything Prototype Related |------
Now featuring comp case mods:
[RepRapLab.wordpress.com]

--------------| Find us at Twitter|------------------------
@REPRAPSQUAD (RS Main)
[mobile.twitter.com]
@REPRAPSQUADHQ (ProtoLab)
[mobile.twitter.com]

Bent is a relative term, isn't it? I don't have the tools to determine whether a screw is bent or not other than rolling it across a flat surface, and that is not a sure-fire method of determining if a screw is bent or not. The only way I have to tell is to cut them to length, install them and observe them. And I'm not bending them while they're being cut. Vice-held and cut with a porta-band at the vice jaws, not 6" away. Pretty sure I did all the due diligence I could without having access to a machine shop.

I don't think McMaster will straighten the screws that they ship out, or even verify that the screws they ship are straight at all. Hell, they don't even check what they ship. I ordered an M3 threaded rod the other day and got a length of 4-40. They are not QC oriented, but they are cheap. And no, I'm not going to spend $5 to ship it back to get $1 credited to my account.

I know for a fact that Roton offers screws that are certified to a straightness spec. Guess that's the difference between a supply house and a manufacturer.

And yeah, all (3) 1/4-16 screws were bent - 2 RH and 1 LH. And all screws were ordered at different times. I'm willing to bet it's down to their sloppy method of shipping "precision" parts.

My 10mm Roton screws came inside a tube that you could probably kill someone with if you really tried - the walls were about 3/8" thick with wooden plugs securing the screws in the tube. Professional.

And yeah, it was nice of me to buy from them. But I have learned my lesson.


xiando Wrote:
-------------------------------------------------------
> "I'm using the McMaster 1/4-16 ACME screws. The
> came pre-bent from McMaster. Very nice of them. I
> highly recommend not purchasing screws from them
> (at least 1/4" anyway). I've bought (3) 3' screws
> from McMaster, each one pre-bent. Live and learn.
> If you've got 8mm smooth rods I wouldn't go up to
> a leadscrew larger than 8mm. "
>
> Then you should have sent them back to McMaster,
> since they were obviously received in a failed
> condition, instead of snarking about your own lack
> of due-diligence as if it was their fault. How
> nice of you.

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

Thank you akhlut, that's my first build!
Well, it is not without its problems. The gantry is a little bit wobbly; I am already planning the next iteration to fix that and to add a folding mechanism to make it more portable.

Before I installed the M5's, I had some M8 screws installed (complete with spiral couplings; you should be able to see them on the second photo), with 608 bearings as supports. I tried the M8's without bearings, but the situation did not improve. I personally would not like to run the machine without thrust bushings because of the load imparted onto the motor bearings (does anyone have data about motor bearing degradation due to incorrect use of stepper motors?).
If the wobble doesn't go away in the successor build, I will try one or two runs without thrust bushings.

So that is what these plastic tubes are! I think my solution is flexible enough, though, as both the upper and lower sections are bifurcated and the middle section is pretty thin. Together with the elasticity of shrink wrap and zip ties, this coupling bends pretty easily.

Nope, my Z nuts look like this (photo of the old M8 version, but overall construction has not changed):


What I have found out in the mean time is that my Z bearings needed some re-tightening. Doing that immediately yielded better results, but the X End construction is still promoting wobble.
These are my bearings (used on all axes) by the way:

The motors have a pair of 5mm ball bearings, so adding an 8mm one doesn't do a lot. Particularly as threaded rod is not a good fit in the bearing because it is about 0.2mm smaller, so does not centre.

The datasheet gives maximum axial and radial loads. Since the Z motors move so little compared to X Y and E I don't think it would matter even if you exceeded the rating somewhat. You can also replace the bearings if they wear out but none of my motors show any sign of wear after years and when I scrap one machine I put them in the next.

---

[www.hydraraptor.blogspot.com]

I don't use the bearings to guide the screw, but to provide an axial support like for the ballscrews on "proper" CNC machines.

Unfortunately, the data provided by Wantai Stepper Motors is a little bit sparse, so I am substituting with a datasheet by Nanotec which tells me the rated allowable axial load is 7N. Distributed onto two motors plus support from tight fitting bearings (only applies to when the X axis is not moving, I guess), I might even be within specifications.
I will definitely weigh my X axis and try this out on my next build.
As always, nice to read some insight from you!

The bearings on the leadscrews of proper CNC machines provide both the axial and radial support. They have a precision turned shoulder that fits the bearing and threaded collar. Then the flexible coupling allows for slight misalignment of the motor shaft and absorbs the vibration from the motor.

I tried making the same set-up with an M8 lead screw with nuts each side of a 608 bearing and a printed flexible coupling. It didn't work very well because M8 nuts don't sit flat enough to make the screw perfectly perpendicular to the bearing. The sells Mendel gets away with it by not constraining the bearing fully. I don't know if that was on purpose or accidental.

So on Mendel90 I went with the simple arrangement of motors at the bottom with a coupling that is flexible only angularly. The motors I use are rated for 10N axially so the pair will support 2kg, which is more than the weight of the X axis and extruder.

Edited 1 time(s). Last edit at 04/02/2013 08:38PM by nophead.

---

[www.hydraraptor.blogspot.com]

I noticed that the nuts are not perpendicular to the bearings, too, but only used one nut on top of the bearing because there are no forces coming from below, pulling on the motor shaft.

What motors are you using? The 0.9° ones I am using for all axes are overkill for Z - 500 full steps for 1 mm!

I wonder if I'm having a similar issue on my X and Y axes. The pulley in both cases is a GT2 36 tooth pulley. Using the reprap calculator, this works out to 44.4444444... steps/mm. I would think I'd see a similar periodic error. I can try rounding to two decimal places to see if it helps. Alternatively I have some 20 tooth pulleys (for which the math works out to exactly 80 steps/mm.

I would hope that the rounding would be enough, because on these axes, the movement is not as regular as in the Z direction. In Z, for the most part, we are moving 1 layer height at a time, so we can optimize the math. Since X and Y are moving arbitrary distances, we cannot perform this optimization.

@uGen,
I use these: [www.zappautomation.co.uk]

---

[www.hydraraptor.blogspot.com]

uGen,

I'm probably using the same motors as you, 42BYGHM809's by Wantai.

I've got a pretty heavy X-axis - two 650mm 10mm steel rods + the usual bits. Not sure of total weight, but my steppers support it all and seem none the worse for wear.

But your X-axis does look quite heavy...lots of fittings and plastic

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

What slicer are you using?

If you haven't tried it yet go try kisslicer. I thought I had a z wobble and it turned out being my slicer.

slic3r 0.99 is pretty good, much better than 0.98

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

I switched to a full-step layer height AND I rounded the Z esteps value to two decimal places. There has been a significant increase in print quality. I've printed off the X-Z decouplers as well as a couple of redesigned X ends that do not have a channel for the nut - just a circular hole. I haven't installed the new parts yet because I do not yet have the nylon and teflon washers. I expect to get them this weekend and I will see the result.

As far as slicers go, I've tried kisslicer and I just don't like it yet. It shows promise, but it's still a little rough to me. I'm a skeinforge user and have it tuned nicely, but lately I've been happier and happier with slic3r. I'm currently using 0.99, and the only remaining issue for me is bridging.

See, I tried skeinforge and got lost in the settings...too much control and no idea wtf I was doing. Typical noob stuff. Maybe I should give it another go.

As far as Z esteps go I've got an integer value and the layers are quite good.

Do you have a before and after pic of the print quality change so far?

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

If you're happy with slic3r, I'd say stick with it. It's probably the future of slicers. I haven't seen an update to skeinforge in a year, and in a hobby moving as fast as this one, as far as I'm concerned, that's dead code. I only use it these days if slic3r doesn't work on a specific object.

@akhlut and nophead: Thanks, you two removed many of my doubts about using the motors without support. Guess my concerns were a little bit over the top, although it is good to have some safe wiggle room before one exceeds the specs of a part.
My X axis is in for some major re-design anyway, but it looks much more massive than it is right now because of the large, flat parts.

Why would I argue with Roton or Thompson, Akhlut the anti-american troll?

Trapezoidal is a form factor commonly referred by the oroginator's name of Acme, whihc is the standard form factor upon which metric was made. Therefore "metric trapezoidal" is simply a metric version of ACME form factor. Dip.


akhlut Wrote:
-------------------------------------------------------
> ACME is trapezoidal, and Trapezoidal is
> trapezoidal.
>
> But ACME is not Trapezoidal.
>
> Just like all jacuzzi's are not Jacuzzis. See the
> diff?
>
> One is an adjective while one is a noun.
>
> Unless of course you'd like to argue with the good
> people at Roton, Misumi, Thompson, Nook, etc. You
> know, the people who make ACME and Trapezoidal
> leadscrews to standards (you know, EU hippie
> stuff).
>
>
> xiando Wrote:
> --------------------------------------------------
> -----
> > jbernardis, for the record, ACME lead screw IS
> > trapezoidal form-factor. In fact, it's the
> first
> > trapezoidal form-factor lead screw and the
> > "trapezoidal" refers to the thread profile,
> > nothing more. Don't let the EU rhetoric and
> dogma
> > about metric vs SAE fool you. The ACME thread
> was
> > invented in the US in the late 1800s and was
> later
> > made in a metric form-factor. Some here choose
> to
> > ignore the fact that trapezoidal is an American
> > invention and that "trapezoidal" doesn't just
> > include metric trapezoidal.
> >
> > If you're in the North American continent,
> > McMaster distributes precision 1/4-16 ACME lead
> > screw in carbon, heat-treated 4140 and 304
> > stainless steel.