D-bot wobble !! :-(

Hi everybody,

I have a impossible wobble!!
My mechanic seems good but still have wobble.
With a comparator mounted on the carriage I can see the default but can't find it.
My config:
MKS Sbase 1.3 16 micro steps, 20 teeth pulley on the motor, 60 teeth on the leadscrews, 1.5mm pitch leadscrews, gt2 belt.

The motor current is ok,
I tried different motors,
I tried 2 motors at same time with current at 2.2A
I tried different pulley config,
I tried different position for the belt
(I ordered bigger belt: 10mm gt2 if the belt stretch too much)
I tried full steps, 32 micro steps 16 micro steps, 1/2 steps.
I changed Z wheels guide to linear rails (really stronger setup :-) )

Nothing fixed the issue

If you have an idea :-) to help me

Thanks

To me it seems, the belt tensioner screw is a bit underrated. You should make a bracket, where the idler is supported from top and bottom.

Can you wiggle the corners of the bed by hand, do you have play in the linear rails?

Edited 1 time(s). Last edit at 05/25/2018 02:34AM by o_lampe.

no play in linear rails, I can move the bed with strong pressure front, the link between bed and linear rail is printed (I'll mill this part in alluminium soon) but I already have this issue with 3 leadscrews and wheells instead of linear rails :-(

I changed the belt tensionner no change :-(

I see two possible causes:
1. The pulleys on the leadscrews aren't centered. When you move the bed, do you hear a constant "eeeeeeee" or is it more like "uiuiuiuiui"?
2. The leadscrews are crap and the pitch isn't distributed equally around the circumfence.

Maybe the leadscrews but good supplyer (https://www.motedis.fr/shop/index.php)

[www.youtube.com]

Several questions come to mind:

1) nozzle size?
2) layer thickness?
3) print speed?
4) DRV8825 drivers? They are known to have issues missing usteps at low speeds.
5) two motors belted together to drive two screws? Have you tried one motor driving both screws?
6) calibrated extruder?
7) layer thickness a full step multiple?

I have been chasing a similar problem in my printer, UMMD, for 3 weeks. I first noticed it was a problem when I put a volcano heater block with a 0.8 mm nozzle on the machine. I spent a week trying to tune up the extruder, tweaking all the retraction settings, etc. and making dozens of test prints. I also thought maybe it's a Z axis problem- UMMD has a belt lifted Z axis. While the construction is as solid as yours (kudos, by the way), the Z axis has one critical part- in UMMD it's a worm gear mechanism that drives the belts that lift the bed. Any flaws in the gears (or in the screws, in your machine) will directly translate to Z axis positioning errors and banding.

The problem is determining whether the banding is caused by a Z axis problem, an extrusion problem, an XY mechanism problem, a filament problem or some combo of any or all of those. I tried to think of all sorts of tests to isolate each of those and could not come up with any. If you print two different objects on the bed at once, a Z axis problem should result in the same banding in the prints, assuming everything else is constant, but if there's some backlash in the XY mechanism, or maybe the filament moves follows different paths on the drive gear, you may get different looking banding. If you print two identical parts at the same time you'd expect the banding to match, but it doesn't tell you if it's from the Z axis for the same reasons. If you print a part, remove if from the bed and print it again, you'd expect identical prints if nothing changes, but there's always room for some error to creep in.

I tried all sorts of things to fix the problem in UMMD and the thing that had the biggest impact was switching hot-ends and heater blocks. Old prints, made using a standard heater block and 0.4 mm nozzle looked great. New prints made using the volcano block and a 0.8 mm nozzle looked bad. Here are a couple old prints, lighted badly to make flaws visible- be sure to zoom in:
Old print, 0.4 mm nozzle, 0.18 mm layers, ABS
Old print, 0.4 mm nozzle, 0.18mm layers, PLA
Both of these prints look great under normal conditions and you have to look carefully and critically to see the flaws.

Here's a recent print made using the volcano with a 0.8 mm nozzle, with 0.4 mm layers, ABS printed at a leisurely 30 mm/sec. I know, it looks over extruded, but I spent a week tuning the damned thing and if I turn the extrusion down, it creates other problems that are more severe. I used Cura and Slic3r and just could not get either to give me quality prints with this nozzle and heater block.

I changed the belts and pulleys in the Z axis from 36 tooth HTD-3M to 60 tooth GT2, put a twist in the belts so that the smooth back of the belts would ride on the smooth faces of the top pulleys, which changed steps/mm from 55.55555 to 50. I took apart the XY stage, cleaned all the linear guides and lubed them, cleaned off the drive pulleys, retensioned the belts. Pulled the extruder and cleaned everything, examined the drive gear under a microscope looking for wear (there appears to be some but it looks uniform and I never looked at it when it was new, so I'm not sure what an unworn gear looks like), and once it was all back together, found no real improvement. Then I decided to take out the volcano block and 0.8 mm nozzle and go back to the original set-up. While all this was going on I received a new hot-end that I had ordered via Ali Express, so I put it on the machine. It has a "standard" sized heater block and a 0.4 mm nozzle. The stainless steel heat break has a thicker neck than the E3D part and the whole thing fit more tightly into the Titan extruder than the V6 that was in there- I always felt like the V6 with the volcano was a little wobbly.

I tried a few test prints, here's a typical result- 0.4 mm nozzle, 0.2 mm layers, 60 mm/sec. Under normal lighting, this print is great, and the ringing (from tweaks I made to the junction deviation) looks much worse than the banding, but I was still curious about random banding. I compared this print to good prints from other machines at the makerspace- SoM, a new Prusa i3, a CubeX and a makerbot clone. All the prints exhibited some of this random banding in the Z axis and the new print from UMMD actually looked equal or better than the others.

I have come to the conclusion that a small amount of banding is probably unavoidable in FDM printing. I think there are numerous sources of imprecision that add up to the banding. Some is going to be Z-axis, some extrusion, and some inevitable backlash in XY. As far as extrusion goes, once the plastic leaves the nozzle, it keeps flowing (even if only for a fraction of a second) until it cools enough to harden. With the volcano depositing comparatively large amounts of plastic, I think it takes longer to cool and so will tend to slump more after it leaves the nozzle, resulting in worse banding. If anyone has made very high quality prints with a volcano and large nozzle, please show me that I'm wrong.

Edited 2 time(s). Last edit at 05/25/2018 08:27PM by the_digital_dentist.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

1) all the size :-): pictures 0.4mm

2)all layer thikness: I tried a lot of different layer thikness, reprap prusa calculator told me 0.18mm don't really change anyting

3) speed here 60mm/s, tried 50mm/s and lower and higher but same result

4)Yes DRV8825 :-( ordered diode smoother just to try if it improve Z

5) I tried 1 motor and 2 motors to increase the torque. rifht current for 1 motor 1.5 A (set to 1.8) and current for 2 motors 2.2A the max of the driver. same result

6) yes calibration OK ask 100mm the extruder give me 100mm

I tried other leadscrews in stock (8mm pitch 4 ways) the issue is almost the same

I'm reading your long post :-)

OK I follow you on your Z axis setup on your specific post.

Your design inspire me for mine ;-) thanks

This banding issue is a crap

I'll try soon ballscrews with the same setup. later I'll design Z belt design to test all of this but without worm gear !

Same as you I tried a lot of things to find the issue and isolate it !! I'll ty some of your test on my setup to find the GOLD setting :-)

Just a couple of thoughts,

I read something a while back about bang-band control on the heatbed causing some strange issues or it could be thermal expansion knocking something out, maybe print without heatbed

Have you tried printing in vase mode might show a differance, could also try a different slicer

Already tried without heatbed same result, for vase mode same result.
I changed my leadscrews for balscrews it's better but wobble is still here just lower (not the same occurrence in Z because different pitch)

ballscrews + vase mode, cura or slic3r same result, 0.6 mm nozzle 0.18mm layer height

I don't see "classic" wobble anymore. But it looks like inconsistent Z-movement. Some perimeters are thicker all around than others.
0.18mm layer height is a bit low for 0.6 nozzle. For vase mode you can double it up.

What about the z-stepper wires? They are pretty long, maybe it helps to twist them to avoid crosstalking?


Edited 1 time(s). Last edit at 05/27/2018 02:53AM by o_lampe.

God idea for stepper wires, layer height too, usually I use 0.2mm minimum with 0.6 nozzle bit prusa calculator give me 0.18, I'll try with smaller nozzle or bigger layer.

On the left of the picture 0.4mm nozzle layer height 0.18mm, cura, spiral vase



On the video with comparator, 100 microns movement ask by software, the linearity is quite good repetability too but could be better

[youtu.be]

Looking a lot better,

I would go to 0.4mm nozzle and around 0.2mm layer depending on picth of ballscrew and gearing.

Do a thin wall calibration too.

I swapped over to slic3r prusa edition due to poor finish on the outside

Gearing, 20 on motor, 60 on leadscrews, pitch 4mm, prusa calculator told me 0,18mm optimal layer but I can try 0,2mm. Thin calibration is done on this print, 0.4mm nozzle -> around 0.4mm wall

The Prusa calculator is suggesting layer thickness based on full step resolution in the Z axis.

With your pulleys and screws, the motor turns 3 times (600 full steps) to move the bed 4 mm (one turn of the screw). 600 full steps / 4 mm = 150 full steps/mm. The calculator says to use 0.18 mm layers: 0.18 mm x 150 full steps/mm = 27 full steps, but 0.2mm x 150 full steps/mm = 30 full steps. So either should be OK. As long as you're using full step multiple layer thicknesses you should be fine.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

I received today my 10mm GT2 closed belt and 10mm pulleys instead of the 6mm one, I'll check ASAP if the tension is better and the flexibility lower, to increase the linearity cheched with the comparator

I hope the new belt is shorter too? You can better add a pressure idler close to the stepper to improve belt-wrap, but keep the belt as short as possible.

Edited 1 time(s). Last edit at 05/30/2018 02:15AM by o_lampe.

No same length, it was planning to use 3 leadscrews with this belt, I designed the new linear Z a week after the order, a new shorter one is planed later.

Damned :-(

The bottom part looks better than the rest. Do you have any probs with the Z- linear rails? ( flat spot on bearing )

the Z rails are smooth as possible don't feel something when I move by hand :-(

Don't understand the difference between bottom and the rest of the part. yesterday I checked the quality at the begenning of the print and 10 hours later bad result !! frustrating :-O !!

Now I installed 10mm belt, test ongoing

I wonder if your antibacklash nuts are contributing to the problem. The weight of the bed will normally keeps the nut's threads resting on the top of the lead screw's threads- it doesn't matter whether the bed is moving up or down. Antibacklash nuts are only needed is horizontal motion systems that don't have the force of gravity ensuring that the screws and nuts are engaged, or in vertical systems that use more than 1G acceleration. The antibacklash nuts may be creating some backlash if they don't allow the lower surface of the nut's threads to rest on the upper surface of the screw's threads.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

the other standard leadscrews (8mm pitch and 1.5mm pitch) are without anti-backlash nut, with the same issue.

Test with 10mm belt:

In this picture, I noticed both leadscrews are on the same side of the linear rail. "In front" of the rail, but not in line.

If you hadn't have the same problems with all the leadscrews you've tried, I wouldn't have paid further attention.
But you might solve the problem by placing the right leadscrew "behind" the linear rail. Or place both leadscrews in line with the rails.



Edited 2 time(s). Last edit at 05/31/2018 01:19PM by o_lampe.

In your video of the 100 um steps, the error increases to about +30 um as you get to the 1 mm point, and by the time you get down to the 2 mm point, the error has corrected itself. That looks a lot like it might be a periodic error (can't tell unless you test it at different positions on the Z axis). Try moving the bed down 50-100 mm, and bolt the dial gauge to the printer's frame and do the same test again. See if the error is still there.

When I was trying to make a cheapo worm drive to run the Z axis in my printer, I measured the step sizes over two complete rotations of the worm gear and plotted the result:


And this is what the prints looked like:


After I put the Rino back into the machine, I started having random banding problems.

Going way back to the beginning of the thread, it looks like you have a volcano heater block on the hot-end. What is your extruder/hot-end set up?

When I was chasing random banding in the Z axis, the change that started my troubles was putting a volcano heater block and 0.8 mm nozzle in the V6 hot-end and the change that fixed it was changing the hot-end to one that fits in the Titan extruder more securely (a hexagon knock-off from China) with a "normal" heater block and 0.4 mm nozzle. In my case, I am pretty sure most of the problem was caused by the difficulty I had in tuning the extrusion with the larger nozzle and maybe a little was contributed by somewhat insecure mounting of the hot-end in the extruder. I've had the Volcano in and out of the machine a few times and have never managed to tune the extrusion enough to get satisfactory prints.

Maybe you can try a "normal" heater block with a "normal" sized nozzle. It looks like you've tried almost everything else...

The hexagon knockoff has been printing in my machine for about a week and it seems to be behaving very well. I have printed PLA, ABS, and PETG with it so far and have had no issues.

I just ordered a metal version of the Titan via Ali-express. It has some good features, but I can see flaws in its design. It looks like a couple simple modifications will fix the flaws.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Couple more thoughts,

Have you got a geared extruder with printed gears , might not be round

Have you got a flat tip for your clock ?, You could put the clock on the leadscrew move z up and down and see if it runs true

To the digital dentist:

I'm already working on this kind of measurent :-) your thought call me ;-).
I will try too to make measurement with manual pisitionning of the motor: 1 full manual rotation of the motor -> 1 measure of the position of the bed (to completely remove the uncertainty of the motor driver.

Yes the volcano is hard to seting up :-( still working on the PID and extrusion rate are really different depend of the filament type and the diameter of the nozzle, need to make an excel sheet for each combinaison.
I'll try another heater block with 0.4nozzle (try the same part with 0.4mm nozzle on volcano -> change the hotend and PID and try again.

To O lamp:

You think the position of the leadscrews will change the linearity of the movement you whant to place the leadscrew in front position aligned with the linear rail ? with this configuration you can have tilt on the bed ? (OK the linear rail could handle the torque) in the actual position there is the minimum effort on the carriage

To adambrum:

I use titan aero extruder with volcano hotend and panckage stepper. (I think I could try bigger motor on extruder to be sure about extrusion flow linearity)

Do you have the top of the leadscrew supported, could be worth removing it, the speeds we move at its not going to whip