Annoying z banding issue prusa i3, possible fix?

Hey everyone, sorry if this has been discussed before but im having some issues with z banding with my i3, i have tried printed and alloy couplers, replaced the 5mm threaded rod with straight stainless steal rods and ensured the the motor shaft and threaded rods arent touching. All these fixes havent been able to sort the issue. After looking through many forums it seems that some have had success in ridding these awful bands by switching from 5mm threaded to 8mm threaded, there reasoning was that the 5mm threaded allows to much flex where as 8mm threaded rod provided more stability (aslong as it was straight). Has anyone tried this? i would be willing to print out the parts and test it if someone has either tried this with success or at least agrees with there reasoning.

The other idea i had came from seeing the excellent print quailty from the UP 3d printers that use belt driven z axis instead of threaded rod. Im not sure how difficult it would be to convert a prusa i3 to have belt driven z axis but it might be something to try

I use brake pipe male and female screwed couplings to join my z motors to the 5mm leadscrew, plus spent time lining up the shaft to the motor. This reduced the banding, although didnt eliminate. I am ging to reduce from a .5 nozzle to .25 nozzle and reduce the lair height from .35 to .15 to .2 for the .25 nozzel, this should further clean up the print.
It should be possible to convert to belt drive, although you would have to keep the motors on to stop the z axis dropping between moves. Clearly the motors would have to be re positioned, and the belt route would have to be thought through to avoid the x axis.

i have tried silicon pipe before but it was too flexible, i have a 0.35mm j-head printing at 0.2mm layer heights and as you can see from the attached pic the errors are highly visible

what type of extruder are you using? I had similar problems and it turns out it was the result of a slightly not round big gear on my gregs type extruder mixed with slight over extrusion.

Im using a direct drive extruder with a mk7 drive gear, ive re checked the esteps and it seems okay however its difficult to get it exact, if i told it to extrude 10mm and it extruded 9.72 is that exceptable or is it to far out of range?

Like you said you have Z banding the solutions you're talking about are for Z wobble totally different. Z banding can be caused by a lot of different things. Can be caused by thread pitch and layer height steps not coming out exact and it automatically makes an adjustment when it can. I'm not really up on this but there are posts out there to set your layer height so it comes out exact amount of steps with the z I would try that first

Edited 2 time(s). Last edit at 02/19/2014 07:52PM by cnc dick.

Are your z-axis motors disabled when not in use?

// Disables axis when it's not being used.
#define DISABLE_X false
#define DISABLE_Y false
#define DISABLE_Z false
#define DISABLE_E false // For all extruders


What are your z steps/mm?

Assuming M5 with 200 step motors and 16x micro-stepping, should be 4000 steps/mm. For a layer height of 0.2 that's 800 micro steps.

As an experiment, you should consider switching your z-axis to full-step mode and setting your Z steps/mm to 250. For a layer height of 0.2 that's 50 full steps.


What I'm getting at is that at anything above 8x micro-stepping is operating outside the stepper motor's positioning ability. Typically steppers are rated at +- 5% accuracy, so for a 200-step motor the maximum resolution you can reliably achieve is 2000 steps.

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

Hi akhlut

Yeah my z-axis motors are disabled when not in use, below are my current settings for that and steps

// Disables axis when it's not being used.
#define DISABLE_X false
#define DISABLE_Y false
#define DISABLE_Z false
#define DISABLE_E false // For all extruders

#define DEFAULT_AXIS_STEPS_PER_UNIT {80,80,4000,108}

So what you're saying is that at 16x micro-stepping the higher resolution can cause inaccuracy? How do you change to full steps when using ramps 1.4? i would have thought that going from 800 microsteps per layer(at a layer height of 0.2mm) all the way down to 50 full steps would dramatically reduce the resolution and would therefore make the layers worse? although i suppose the less steps then the less variation there can be

Well, it's not that it would cause inaccuracy, its that the additional resolution is meaningless. Think about an old truck with manual steering - with one of those steering wheels that you can turn through 90 degrees and it has no impact on the direction of the truck. A friend of mine had an old Bronco that you could do this with (fun times). You can turn the wheel 90 degrees in a single motion to get the steering to respond or you can turn the wheel in 2 increments of 45 degrees or in 4 increments of 22.5 degrees - both of those processes are wasting resolution because the accuracy of the system is low. Similarly, once you're operating outside the accuracy of the stepper, higher resolution is meaningless. It's a case of more is not better.

This wikipedia entry is helpful.

---

Actually, your Z is enabled when not in use (what you want to have) if '#define DISABLE_Z true' then it would be disabled.

Actually, first and foremost - have you calibrated X and Y? 80 steps/mm on each seems like such an unlikely number.

Depending on the stepper drivers you have you just change the jumpers underneath the stepper. But before you remove the stepper driver make sure you power everything down. Unplug the USB and pull the power block.

Once you have the jumpers set for full step re-install the stepper driver, power block and USB.

Open Marlin and change:

#define DEFAULT_AXIS_STEPS_PER_UNIT {80,80,4000,108}
to
#define DEFAULT_AXIS_STEPS_PER_UNIT {80,80,250,108}

And while you're in there you'll want to change these z-values to something less aggressive.

#define DEFAULT_MAX_ACCELERATION {9000,9000,100,10000}
to
#define DEFAULT_MAX_ACCELERATION {9000,9000,10,10000}

#define DEFAULT_ZJERK 0.4
to
#define DEFAULT_ZJERK 0.2

Recompile and upload.

Your Z-axis will sound...different.

Also, make sure you are choosing an ideal layer height! Anything that is a multiple of 0.02 will work. So 0.2 is fine, 0.16 is fine, but 0.15 is not fine.

if you really want layer increments of 0.01, then switch to M6 threaded rod.

See if that helps.

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

Ah i understand what you mean now, thanks for the example, im using a ramps 1.4 with a4988 stepper drivers, after looking on the net i found out that to switch to full steps i needed to remove all three jumpers from under the z axis stepper motor, i removed them and made the changes to marlin as you stated, connected the printer to repetier host and attempts to move the z axis, there was a very high pitched wine but no movement, i tried moving it again and along with the high pitched wine the left motor didnt move but the right motor moved a tiny bit, not sure what the issue was, but fearing that i had damaged something while removing the jumpers i went back to my original setup and everything worked/moved as it should have so im not sure what the issue was.

In relation to calibrating the X and Y axis' i used the reprap calculator, im using gt2 belt with a 20 tooth pulley which comes out at 80, when i print a 20mm test cube it measures 20.04mm x 20.06mm so i figured it was okay

Im using 0.2mm layer height as a base line for all my testing so i have a constant, its interesting that you say m6 can move in layer increments of 0.01 and m5 can only move 0.02, might be worth me upgrading to m6 at a later date,

For now im stuck onto what to change / try next

You might also want to check the linear bearings on your Z axis. I just replaced my metal ones with PLA bushings and now I'm getting Z ribbing like crazy (which I wasn't getting before). I'm going to switch back.

Hi sheck626, im currently using lm8uu's for the Z axis, i was considering printing some ABS bearings to try but after hearing your issues with printed bearings i might get it a miss

That high-pitched whine is the stepper motors being over-driven. Lower your Z-axis motion rates to about 1/10th of what they were.

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

If you have EEPROM enabled (it is by default enabled in most Marlin branches and Repetier) then the new steps per mm will be overridden by the old ones in the EEPROM. This would cause the high pitch squeal as it tries to take 16x the amount of steps per second than you want. You will either need to disable EEPROM and/or reset it to the factory settings after upload of the updated firmware.

M500: stores paramters in EEPROM
M501: reads parameters from EEPROM

If you need to reset them after you changed them temporarily
M502: reverts to the default "factory settings".

You still need to store them in EEPROM afterwards if you want to.
M503: Print settings

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

Thanks Sublime, it was the eeprom settings that was the issue, i disabled it and was able to run at full steps

Sadly akhlut changing to full steps didnt help, in fact it was slighty worse,i was hoping it would solve the issue but will have to try something else,

Any suggestions? at this point i will try anything, even if it is abit extreme

After running through my options i feel like the only real option that should in theory completely solve this issue is by switching the Z axis to be belt driven, sadly my design skills are abit lacking so this might be a very long shot

Again you have Z banding the belt drive for the Z was to help with Z wobble plus the set up in the firmware is going to be more in-depth than fixing your problem which is in the firmware

I am all for belt Z axis and cable Z axis but I would first try changing slicing software. Try Cura or Kisslicer and see if the results are any different.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

How well-calibrated is your extruder?

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

Hey everyone, thanks for all your comments / suggestions, heres what i have tried / trying at the moment

Sublime , I am currently printing a test print using cura instead of repetier host as you suggested and i will take a picture of the results side by side for a direct comparison

akhlut i believe extruder is calibrated alright, i have measured the filament and it is on average 1.75mm (ranges between 1.74 and 1.76). When i tell it to extrude 10mm of filament it actually extrudes approx 9.72mm (its difficult to get it exactly spot on when get down to those tolarances)

cnc dick are you suggesting that it could be something in the settings / firmware that is the issue opposed to the physical setup of the machine?

I feel that perhaps if i provide more info on my setup then perhaps someone can spot something that ive either overlooked or perhaps have done wrong

Im using a Prusa i3 Eintein Version
0.35mm J-head with direct drive using mk7 gear with 30mm cooling fan
GT2 belts for X and Y with a 20 tooth pulley and M5 Threaded rod for Z
Ramps 1.4 / mega 2560 / a4988 stepper drivers
Nema 17 stepper motors
Currently using printed abs couplers to attach M5 threaded to the motor shaft for the Z axis as weirdly it prints better with those than with the aluminium flexible couplers
mk2 heated
30amp power supply
lm8uu bearings for X and Z axis and the 10mm versions for the Y axis

Heres part of my firmware
#define NUM_AXIS 4 // The axis order in all axis related arrays is X, Y, Z, E
#define HOMING_FEEDRATE {50*60, 50*60, 2*60, 0} // set the homing speeds (mm/min)

// default settings

#define DEFAULT_AXIS_STEPS_PER_UNIT {80,80,4000,108} // default steps per unit for Ultimaker
#define DEFAULT_MAX_FEEDRATE {250, 250, 2, 22} // (mm/sec)
#define DEFAULT_MAX_ACCELERATION {1000,1000,5,1000} // X, Y, Z, E maximum start speed for accelerated moves. E default values are good for skeinforge 40+, for older versions raise them a lot.

#define DEFAULT_ACCELERATION 500 // X, Y, Z and E max acceleration in mm/s^2 for printing moves
#define DEFAULT_RETRACT_ACCELERATION 500 // X, Y, Z and E max acceleration in mm/s^2 for retracts

// Offset of the extruders (uncomment if using more than one and relying on firmware to position when changing).
// The offset has to be X=0, Y=0 for the extruder 0 hotend (default extruder).
// For the other hotends it is their distance from the extruder 0 hotend.
// #define EXTRUDER_OFFSET_X {0.0, 20.00} // (in mm) for each extruder, offset of the hotend on the X axis
// #define EXTRUDER_OFFSET_Y {0.0, 5.00} // (in mm) for each extruder, offset of the hotend on the Y axis

// The speed change that does not require acceleration (i.e. the software might assume it can be done instantaneously)
#define DEFAULT_XYJERK 20.0 // (mm/sec)
#define DEFAULT_ZJERK 0.4 // (mm/sec)
#define DEFAULT_EJERK 5.0 // (mm/sec)

Hope this info helps shed abit of light on the issue, any suggestions / observations welcome, i can post my full firmware if needed

Will post the comparison pictures as soon as the job is complete

Hi all,

Heres the comparison pictures, They are both printed at 0.2mm layer heights

The one on the left is printed using Cura and the one on the right is printed using Repetier Host

[imagebin.org]

I looked through all of the settings, the only one that sticks out to me is the E.
It is not hard to set the E steps to be closer.
in your case
you expect 10mm but get 9.75 with a setting of 108

the math is
10 / 9.75 = 1.0256
1.0256 x 108 = 110.7692steps
you can keep doing the math until you get exactly 10mm
I normally use a little more then 10mm when checking.

the bad part of this is it does not solve your problem, because you would end up extruding more, causing (I would expect) more banding.

I am fighting a simular problem also, but am about to do a whole upgrade to my repstrap, to get rid of all of my weak points that I cobbled together to get it to work, and make my bot more solid.

The other thing you could look at is the extruder, it is posible that the orifice is bigger then what you think it is, I also ran in to this recently. my .35 was actually closer to a .5, much to my dismay. I'm not sure if it came that way or if I damaged it somehow, but adjusting my settings to .45 and changing my layer height to match made a noticable improvement. You can check this by extruding in to free air at about the same speed print with, and measuring the output diamerter. Here is a calculator [nathan7.eu] that will give you the expected diamerter of your output for your orifice.

Not that I am a fan of playing with the extruder multiplier in slic3r or the Flow rate in Cura but it might be worth it to see if adjusting it lower( say .95) would reuduce the banding.

I think my issue comes from imperial threads, and getting the proper layer height set. I am close enough that it is not so bad, and I have been too busy printing my upgrade parts to play around with it much. also thinking of going to 8mm threaded rod to make life easier, if I can soucre some locally.

Quote
Horse Renoir
Hi all,

Heres the comparison pictures, They are both printed at 0.2mm layer heights

The one on the left is printed using Cura and the one on the right is printed using Repetier Host

[imagebin.org]

The difference is interesting but clearly not the problem.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

Sublime, Thank you for the pololu tutorial... I found it MOST applicable.....


I think the issue with threaded rods is that the cross section of the outermost surfaces is staggered relative to the absolute center. So by installing it on to the nut, you have (at the nut) a concentric assembly, but the exposed threaded portion in cross section at any given height (rod vertical) is slightly eccentric.(not like I am one to throw stones). Find a picture of a bolt, magnify it and look at it closely. What you will see is a triangular wave, with its crest and trough oscillating 180 deg as the bolt turns. This is the cause of banding.When using a collar to fit the rod to the motor over the screws, this issue becomes apparent. So my solution which seems to be working in early testing is to machine a hole using a bit the size of the motor shaft on a lathe. turning down the exterior, 3 set screws are installed to fasten the screw firmly to the motor shaft. The upper end is then turned down leaving a smooth surface to set close fit in a bearing. While I have yet to print anything, the Z motion seems to be quite smooth and even. Now I am using a rather heavy bolt (12mm I think - been a while) For a smaller bolt you could turn the first 5/8 inch or so of threads to exactly match the motor shaft diameter, and use a coupler.

Okay, first and foremost id like to thank you guys for sticking with me this far,

I think i understand the issue abit better, since the threaded rod is designed to hold nuts etc and not for linear movement the threads arent as precise and the defects are translated into the print. Acme leadscrews are designed for cnc use and the threads are more precise / smoother and should in theory provide better prints, at least this is my understanding of it.

What im struggling with is that ive seen better looking prints on other repraps and other prusa i3 and they are using standard threaded rod. After looking through many forums this banding issue seem more common with the i3, possibly due to the fact it uses m5 instead of m8

As my design skills need alot of work im opting not to switch to a belt driven Z axis at this stage, after searching through thingiverse i found this [www.thingiverse.com] which is supposed to get rid of z wobble.

Im also very keen one switching to m8 threaded rod, i found this [www.thingiverse.com] which in addition to switching from m5 to m8 it also adds anti backlash with the use of extra nuts with springs. If i use this design then i wouldnt restrain the Z Threaded rod at the top as ive heard it cant make the prints alot worse

Ive also found where i can buy some 8mm acme leadscrews at £15 per length, but the only nuts ive found to use with it are £12.50 per nut so it seems alot of expense, that being said, i would buy them if it would get rid of my z banding

Im interested in hearing what you guys think to the ideas above, feel free to say if you think the ideas wont work or perhaps you have an idea that i havent thought of

Thanks again for your time

If you do decide to go to M8, choose M8 fine pitch (1mm/turn) rather than the standard (1.25mm/turn).

Those anti-wobble bits could work, but I have concerns about their actual efficacy. With those you're trying to decouple the X and Z axes. Unfortunately, you have the threaded rod that can't seen to help but touch the X-end.

Maybe something like these [www.thingiverse.com], coupled with those M8-sized X-ends might work well.

Honestly I wouldn't worry about anti-backlash nuts, they were only a source of problems for me.

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

Hi akhlut,
An interesting suggestion with using m8 fine pitch, i wasnt aware they made a 1mm pitch one, is there a reason why this is better than the standard 1.25mm pitch? after looking it up on the net i came across this quote '' The terms “coarse” and “fine” have (in this context) no relation to the manufacturing quality of the thread'. So it would seem that the quality between the two different pitches are the same.

Im not sure how easy m8 fine pitch is to get hold of in different countries but the cheapest one i can find in the UK is £17.50 for 300mm, thats more expensive that the trapezoidal leadscrews, the issue with using the leadscrews is that it costs £12.50 per nut to go with it

I have to admit im abit confused with the concept and reasoning behind the decoupling the x and z axis, i will try printing the Z Screw Isolator and the anti wobble bits and see if they help at all, if they dont help im seriously thinking of spending the money and upgrading to trapezoidal leadscrews, i havent come across other i3's using them (although im sure there will be) so sourcing the stl's for the x ends might prove an issue

Check this video out.

[www.youtube.com]

And these threads:
[forums.reprap.org]
[www.printrbottalk.com]
[www.evernote.com]
[billysprojects.blogspot.com]

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]

Hi akhlut,
Thanks for those links, they really helped, i understand now (at least i think i do lol)

The z isolator basiclly acts as a platform for the x ends to sit on, so any wobble in the threads will allow the platform to move side to side but since the x ends are only sitting on the isolators the wobble isnt tranferred to the x axis, so the only movement on the x axis is the z movement, at least thats how i think it works

Im going to print the z isolators for both the 5mm and 8mm versions along with the 8mm versions of the x ends, i already have 8mm threaded rod and couplers in both solid and flexible and some silicon tubing to so i have multiple variations i can try.

The only thing im unsure about at this stage is the nut that sits in the X ends, when i add the isolator (which has its own nuts) do i remove the nuts that are currently in the x ends? if i leave them in wont it defeat the object of fitting the isolator?

Also im going to take pictures of the prints with the different setups im going to try in the hopes it can help others in the future, is the test cube print okay or is there something else that would help more?

Yup - the nuts in the x-ends come out, and the isolators do all the work pushing the axis up and down.

I'd just keep the cube for consistency.

---

- akhlut

Just remember - Iterate, Iterate, Iterate!

[myhomelessmind.blogspot.com]