Vertical banding
-
Tested x8 microstepping and reduced the spectra line tension a bit, no banding improvements seen.
Print speed does not seem to effect the banding spacing.
3D benchy with vary apparent vertical banding:
3D Benchy bow:
A strange pattern on the compound curves of the Benchy.
Printed a 100 by 280mm tall tower in vase mode.
Here you can see some banding still on the side walls the bands are much farther apart but still vary noticeable to sight and feel.
Think at this point I should try another type of stepper motor or even go to 0.8 step degree motors.
Stepper in use:
KH4234-b95101
3.41v
1.1A
3.1 OHM
4.4 mH
1.8 degreeEdit: Fixed the photos, thanks dc42 for the pointers:)
-
Is it not just the infill touching the walls?
-
I edited your dropbox links to make them work. Put them between img tags (there is a button for doing this) and change ?dl=0 at the end of each link to ?raw=1.
I am no expert on CoreXY mechanics, but I suspect that friction in the mechanics is causing the head to move jerkily. Does the banding pattern on the square frame print change if you rotate it 45 degrees?
-
I have possibly the same issue which I believe is due to radial vibration in/from the stepper motors. See https://www.duet3d.com/forum/thread.php?id=4580 for my more specific query in my next efforts to deal with it.
My printer is a DIY Prusa i3 design built from OpenBuilds (Ooznest) V-slot (4020 in Z and 6020 for X & Y) so should be reasonably solid.
I find:-
Frequency of pattern along perimeter is constant regardless of print speed, at about 25Hz on my system calculated from gcode analysis. Shows up really well at 30mm/s, spread out thus less noticeable at higher speeds (so frustrating that slowing down does not improve print quality!). This seems to rule out e.g. pattern from hob teeth on filament or bumps in the grooves/wheels etc. which wold lay down the same pattern regardless of speed.
-
Vibration app on phone shows peak around 25Hz with phone sitting on bed during printing or gcode exercise
-
Amplitude of vibration (pattern intensity on perimeter and gram measurement on phone app) varies directly with belt tension (lower tension = less pattern)
-
Lots of people have this problem, refer to as moire, ripple, zebra stripes, vertical lines (where it is not ringing which always confuses the threads), with varied solutions making me feel like none of them are really getting to the real answer; some hacks involving reducing stepper motor voltage or hardware mods on driver boards
-
Have in both X and Y axes but they are configured same with similar belt tension; not sure about Z and extruder.
-
As with Kezat testing has shifted to single wall usually spiral vase prints so not infill showing through.
-
Shows up really well with shiny black PETG (admittedly I am burning up the famously crap filament begged off another forum member and blogster to figure this out :-)) ) but examination of old prints shows to varying degrees going back to ABS and my pre-Duet, pre-OpenBuild system (So not DC42's problem, just hoping the Duet capabilities can help me solve it!)
So that's what I've been playing with the last week or so…. Sorry to hijack Kezat's thread if his problem features don't match mine, but the photo looks the same.
-
-
@fma:
Is it not just the infill touching the walls?
Both the hollow test box and the tube does not have any infill, test box is done with 4 extrusion loops for a solid infill free wall.
As for the benchy it has a low infill so its only touching in inside in 2 or 3 locations not the 8 or so times it would take to cause that.I edited your dropbox links to make them work. Put them between img tags (there is a button for doing this) and change ?dl=0 at the end of each link to ?raw=1.
I am no expert on CoreXY mechanics, but I suspect that friction in the mechanics is causing the head to move jerkily. Does the banding pattern on the square frame print change if you rotate it 45 degrees?
Thanks for the tip on the images, most helpfull.
The system is extremely friction free, with no motors attached to the carriage lifting one corner of the printer only 20mm will cause the carriage to slide to the low side. Even with the motors connected and spectra line tight it still has the least friction of all my printers.
As noted above rotating the hollow box 45 degrees does not seem to change the ripple, if anything it could be slightly worse.
-
Tested 0.9 degree stepper motors for the X and Y, I cant see any changes.
Left is 1.8 steppers right is 0.9 steppers, black sharpy mark is the front and marks the top of the part.
The banding has the same spacing with both a 1.8 and 0.9 degree stepper motor
Top object is 1.8 deg and lower one is 0.9 deg.
Making a little progress I think, it looks like slow speeds make it much more visible and also reduce any corner ringing overlapping the banding. I changed the model a little to have a wedge shape at each end, I wanted to see if the bands would take the shape of the triangle, as you can see below it does not. It looks like the bumps stay perfectly vertical. I then paused the print and shifted the X zero point 0.3mm to the right and again a little bit later 0.6mm to the right to again see if the banding location should shift. The horizontal lines are where i did the shift, as you can see it still lines up perfectly vertically.
I'm not sure what this tells me, I would think this could be the full step location on the motor but the 1.8 and 0.9 comparison seems to disagree with that. I don't think this is going to have anything to do with the extruder because the full step location is not going to be in perfectly the same place every time.
More testing is needed.
-
Hmmm…
It looks like the I get about 14 bands per 10mm.
My steppers are set to 16 microstepping (ignoring interpolation which is on atm).
My steps per MM are set to 45.35.45.35\16=2.83 full 1x steps per mm
2.83*10mm=28.34
That is really close my vertical bands multiplied by 2. Could this just be a coincidence, anyone have ideas whats going on?Edit: Perhaps the reason 28.34 matches 14 if divided by2 is due to the core XY mechanics?
-
Looking closer now that I can see the banding better at slower speeds the top object is printed on the bed rotated 45 degrees and lower one is normal 0 degrees. By my count the lower object has 14 bands and the top has 10 bands per 10mm, could this be due to the 1:1.41 ratio of a coreXY?
I feel like I need some one who is good with numbers to put all the above information together, my thinking at this point is the cause of this is the way the duet board is driving coreXY mechanics.
-
Hi
I'm no expert in this type of printer but do mechanical designs 2000h a year and will try to help.Since the pattern is so uniform the extruder can be excluded. It's not over under extrusion.
You have rails all around the machine so you know that the linear motion is without the pattern. You also know that the patterns are exactly straight in the Z direction –> the fault is in your XY plane. The carriage needs to wobble in the Y direction along the X linear rail. The pattern is very tight and I find it a bit strange because my guess would been that there is a pulley or two that are not concentric. But to get such a tight pattern means that the pulley is small or it has several high spots. I don't know exactly how an eccentric pulley would effect the print but I'm guessing exactly like your fault. The harder you tighten the wires the more motion is created by the elongation of the path.
It is very unlikely that it has to do with the card. It runs smooth as silk and even if it didn't the error is no were near enough to create such a pattern. It would require microsteps out of sync pulsating back and forth.
How have you made all the pulleys?
//P
-
Hmmm…
It looks like the I get about 14 bands per 10mm.print a rectangular object aligned with bed so can identify axes, print at two different speeds. If same as my problem at 2x speed (faster) above will get 7 bands per 10mm.
-
Hmmm…
It looks like the I get about 14 bands per 10mm.print a rectangular object aligned with bed so can identify axes, print at two different speeds. If same as my problem at 2x speed (faster) above will get 7 bands per 10mm.
I'm getting the same spacing at any speed, speed only seems to effect how visible the bands are, I suspect due to the mass of the extruder/gantry dampening out the vibration. The only change i can get in the spacing is by printing the test object at a 45 degree angle then its a constant 10 per 10mm with i suspect a linear transition to that point as the angle of the wall is changed from 0 to 45.
Hi
I'm no expert in this type of printer but do mechanical designs 2000h a year and will try to help.Since the pattern is so uniform the extruder can be excluded. It's not over under extrusion.
You have rails all around the machine so you know that the linear motion is without the pattern. You also know that the patterns are exactly straight in the Z direction –> the fault is in your XY plane. The carriage needs to wobble in the Y direction along the X linear rail. The pattern is very tight and I find it a bit strange because my guess would been that there is a pulley or two that are not concentric. But to get such a tight pattern means that the pulley is small or it has several high spots. I don't know exactly how an eccentric pulley would effect the print but I'm guessing exactly like your fault. The harder you tighten the wires the more motion is created by the elongation of the path.
It is very unlikely that it has to do with the card. It runs smooth as silk and even if it didn't the error is no were near enough to create such a pattern. It would require microsteps out of sync pulsating back and forth.
How have you made all the pulleys?
//P
Definitely, I agree its in the XY movement, not sure if its clear but this pattern is seen in both the wall in the X and the Y direction, so all sides of my little hollow test box have this pattern.
All XY 8 steel idler bearings (u groove) are of good quality and 11.33mm diameter where the line rides. The spectra line spool on the motors is 3d printed and 22mm diameter, while its not flawlessly round its not bad and any error caused i think seen over 10s of mm and not 14 in 10 mm segment.
While I love the Duet I am using the less common coreXY motion system and this issue can be hidden with slightly faster print speeds so I feels that its still on the table as a factor.
-
I'm getting the same spacing at any speed, speed only seems to effect how visible the bands are, I suspect due to the mass of the extruder/gantry dampening out the vibration. The only change i can get in the spacing is by printing the test object at a 45 degree angle then its a constant 10 per 10mm with i suspect a linear transition to that point as the angle of the wall is changed from 0 to 45.
…
All XY 8 steel idler bearings (u groove) are of good quality and 11.33mm diameter where the line rides. The spectra line spool on the motors is 3d printed and 22mm diameter, while its not flawlessly round its not bad and any error caused i think seen over 10s of mm and not 14 in 10 mm segment.Interesting, thank you for that. So next question is what do you have that is doing 14 repeats in however long 10mm segment takes? If the pattern is constant with linear travel of filament or carriage then it should all be geared together? Otherwise like on my issue you can calculate a per second frequency from the gcode.
Tooth marks from your extruder hob? As you note, pulley wheels and the like do not seem to cycle as quickly as this pattern.
-
To see if it is an extruder issue, you can change the wall width, so the extruder will turn faster/slower per mm of wall, and so banding step should change.
-
By running the machine faster you introduce more randomness to what the cords are doing and the speed makes jerks less impacting. I am still 90% convinced that there is some sort of concentrically issue. It is not easy to get artifacts as repeating as you have managed in other ways. The ball bearings should be good. And If mounted correctly, by fixating the centerpiece down to the base of the mount, there can't be any movement to them. If the bearings can't be bolted down add a shim above, middle and under.
Do you have the same bands on you pulleys? That can explain the problem. I think that if you spend the money on nice rails you should try to get the pulleys turned in a lathe or minimum put a metal sleeve above your printed part.
Can it be the patterns on the cords from how it is woven? I would say you have to many contact points with the cord to make it the most likely scenario.
You might also want to have some air in between the cords on the pulley. If the new cord clings to it's neighbor the effective diameter increases. but it would most likely not give this repeatability to your pattern.
The satisfaction once you solve this will be good
-
@fma:
To see if it is an extruder issue, you can change the wall width, so the extruder will turn faster/slower per mm of wall, and so banding step should change.
But I doubt it is an extruder issue: bands would not be vertical, and angle should change depending on the perimeter length.
A friend had such issue; see entry #24:
-
You have an unusually low steps/mm. My Scara printer also had low steps/mm and I could see on it that the arms were not moving smoothly even when doing raw motor moves that don't involve segmentation.
The cause was that not all microsteps are equal. Manufacturers of Nema 17 stepper motors make a greater or lesser extent to make them more equal, but this comes at the cost of reduced holding torque.
So if the spacing of the banding matches the full step interval, that's probably the cause. Solutions include:
1. Don't set the motor current too low. At low currents the detent torque causes the microsteps to be less even. Try 80% of the motor rated current.
2. Use 0.9deg motors. That's what I did.
3. Use smaller spools or possibly gearing to increase the steps/mm.
HTH David
-
-
@fma:
To see if it is an extruder issue, you can change the wall width, so the extruder will turn faster/slower per mm of wall, and so banding step should change.
While I think the extruder is all but ruled out at this point as I cant see how any extruder defect would re align up perfectly each pass, this is a good suggestion for a easy test that I will try when i get home from work to make 100% sure.
By running the machine faster you introduce more randomness to what the cords are doing and the speed makes jerks less impacting. I am still 90% convinced that there is some sort of concentrically issue. It is not easy to get artifacts as repeating as you have managed in other ways. The ball bearings should be good. And If mounted correctly, by fixating the centerpiece down to the base of the mount, there can't be any movement to them. If the bearings can't be bolted down add a shim above, middle and under.
The metal idler bearings have a close tolerance to the 3D printer mount with 3 washers to prevent any friction. The clamping force is only light so the bearings could move slightly but as the angle of the spectra line on the bearings never changes i'm not sure how this would cause a issue as the bearing would be pulled to once side of the bolt and stay at that location. Feel free to correct me on that point.
Do you have the same bands on you pulleys? That can explain the problem. I think that if you spend the money on nice rails you should try to get the pulleys turned in a lathe or minimum put a metal sleeve above your printed part.
The pulleys don't have this banding nope. I totally might get the pulleys turned on a lathe if i can get the other mechanics sorted.
Can it be the patterns on the cords from how it is woven? I would say you have to many contact points with the cord to make it the most likely scenario.
Hmm i did not think of that one, here i was thinking i did not have to worry about any belt pitch or to turning over belts to run the belt flat side on the idlers. I will look in to this, this would be the worse case issue as it could scrap the entire design.
You might also want to have some air in between the cords on the pulley. If the new cord clings to it's neighbor the effective diameter increases. but it would most likely not give this repeatability to your pattern.
The satisfaction once you solve this will be good
I could space them a little father in the area im printing to check this perhaps. Yea, i really want to get this sorted, it could be ignored by finding the sweat spot on the print speed and such.
-
You have an unusually low steps/mm.
HTH DavidYes, the spools are very big. But are motors that bad? It would have to be approx. a full 16th microstep out of sync. @David did it solve your issue?
Yes motors can be that bad, some are better than others. Even full steps can vary in size by up to 5%. Changing to 0.9deg motors helped me a lot, but didn't completely solve it.
-
You have an unusually low steps/mm. My Scara printer also had low steps/mm and I could see on it that the arms were not moving smoothly even when doing raw motor moves that don't involve segmentation.
The cause was that not all microsteps are equal. Manufacturers of Nema 17 stepper motors make a greater or lesser extent to make them more equal, but this comes at the cost of reduced holding torque.
So if the spacing of the banding matches the full step interval, that's probably the cause. Solutions include:
1. Don't set the motor current too low. At low currents the detent torque causes the microsteps to be less even. Try 80% of the motor rated current.
2. Use 0.9deg motors. That's what I did.
3. Use smaller spools or possibly gearing to increase the steps/mm.
HTH David
I agree the steps are a bit low, while i can reduce the pulley size slightly I'm limited because it needs to spool the line from the diagonal bed distance ie bigger the bed the more line needs to be able to be spooled up. Also the smaller the pulley the more distance is covered by the line in the Z direction changing the line tension slightly.
1. For most of my testing the current is a bit low at 550mA, my motors are rated to 1.1 A. I did a test with 1.2A motor current with no reduction on the banding.
2. I did give 0.9 motors a go (see photos above) the band spaceing stayed the same and did not seem to reduce the issue any.
3. The spool size is limited by the bed size but with some changes and closer tolerances i might be able to reduce it a little but.
