Banding on tall parts
-
I will try printing again and slow the speed down as it is printing. It doesn't seem to be related to the direction of the printing as the printer is set to print in the same direction the entire time.
The slicer may be reversing the print direction now and then. I suggest you watch which way it prints next time.
-
Ok, I just re-ran the print at a slower speed….hence why the results took so long.
First I can absolutely confirm that the printer does not change directions. What I can say however is that reducing the speed 50% reduces the frequency of the banding by 50%. It appears to be a time based wobble which leads me to two possibilities. 1) Normal power supply fan turn on/turn off. I need to measuring that,
or
- most likely the bed heating up and and cooling down is warping the bed/glass slightly and angling the print, which is why it is more apparent at top versus the bottom. It would make sense since slowing the print down 50% means that the bed warping would change reduce the z-height of the banding since it takes more time to increase the print height.
Is there a way to change the hysteresis for turning on and off the heated build plate?
My bed currently is a heater connected to the aluminum heat plate which is then adhesively attached to a borosilic glass plate (anycubic ultrabase attached to anycubics standard aluminum heater plate). Is there a better setup you would recommend for this?
-
have a look here:
https://duet3d.dozuki.com/Wiki/Gcode#Section_M307_Set_or_report_heating_process_parameters
you can change your bed heating cooling PID etc to it holds a steady temperature. If its slow enough of a cycle to see the banding over multiple layers then you should see that in the bed temperature graph as well unless your bed thermistor is poorly thermally coupled to the bed?
-
Not sure yet, but it seems odd that changing the printing speed would directly change the frequency of the banding. I changed the speed to 150% and the banding changed accordingly. What gave me the idea is that in a few other forums people have complained about banding it it seemed correlated to their bed.
The bed is standard from anycubic and has a PCB heater connected to a 3mm thick piece of aluminum and their ultrabase is another 3.5mm piece of glass with a special coating. The thermistor is on the opposite side of the pcb heater from the aluminum and is held against it by a piece of insulator. The temp. graph looks stable but I can't figure out how to adjust the scale or dump the temp data to graph it and see. Would be interesting to see the duration the heat is on versus off and see if it matches with the banding.
I changed the bed to PID from bang-bang and I will see if that makes a difference at all.
-
I assumed you were already using PID to control the bed, because if you ran heater tuning on it then PID would become the default. Using bang-bang to control the bed often causes Z banding. In your case I suspect that bed tilts a little with heat due to uneven heating or uneven support.
-
OK, it appears that the bed sits on plastic supports on the 2020 extrusions. I still get the banding with PID. I need to print some other stuff out and will get back to debugging this issue soon. The banding doesn't start for about 3-4" or at least isn't noticeable. One thing I read on some forums is the power supply fan can cause issues, but at this point I am done guessing. I will try printing without the heater on at all, and another with the temp increased 20 degrees to see if there is a difference.
-
So I figured I would continue to update this thread in case someone else runs into this sort of thing in the future. I started digging into this again and still not finding any solutions I changed the test print to be a simple 20mm cylinder that was 180mm tall. What I noticed is the the start/stop point which is supposed to be at the same point everytime (and same direction) is shifted each layer. As I begun searching for the layer shifting issue in a delta I found an older post in which someone references a better belt tension. In any event I have new carriages (and new linear rails...just to rule them out) in which the carriage is made out of metal (instead of the current plastic) and have a screw based belt tensioner, versus just a spring. I will see if that helps.
[http://forums.reprap.org/read.php?178,763936](link url)
-
Maybe try a different slicer that you can set the layer start point to be the same point on the cylinder and see if that has an effect on how the banding looks.
-
@t3p3tony sorry if I wasn't clear. I meant that it is set in Simplify3D to the same point and I can see it shifting in and out along with the banding. The bottom picture shows this if you look careful in the middle. Now based on the link I tried tightening the belt as best as I could based on the current carriage. The picture on the left is part of the cylinder prior to doing this and the right is after I tightened them. Definitely look better but still very apparent.
Note the left one is cut in half so I could see if the infill had the same pattern and it does. It is definitely a layer shifting issue.
-
@bpislife ahh ok, well.i guess this has also demonstrated that it's most likely a mechanical linkage problem as well. Is there other play other than loose belts in the printer) what can you observe if you try to move the effector by hand in various directions (not so hard that you force the motors to move) have a look and see if that play is different at different Z heights
-
Great idea. I did find some play in one of the magballs on the cooling fan. I just tighended it the best I could and reprinting the test now to see if that improved anything.
-
Ok, so I have now added angle braces to the side to keep the frame from twisting. It is amazing how much it was able to twist before versus after. It didn't change anything however. I also could not spot anything wrong at different heights. One thing I did notice is that after it homes, it drop by 5mm and the head shifts ever so slightly towards one tower. Now I can see one tower almost bounce and another almost drop. Interestingly enough its the two towers correlating to the layer shifting. Now I just have to find the cause for that little shift at the end.
-
@bpislife said in Banding on tall parts:
Ok, so I have now added angle braces to the side to keep the frame from twisting. It is amazing how much it was able to twist before versus after. It didn't change anything however. I also could not spot anything wrong at different heights. One thing I did notice is that after it homes, it drop by 5mm and the head shifts ever so slightly towards one tower.
That's normal, it just means that your endstop switches are at slightly different heights and corrections have been applied in the M666 command..
-
You're dashing my hopes DC your dashing my hopes.
Seriously thanks for pointing that out, the homedelta file moves to 0,0 at the end so there is that.I am going to continue on the path of replacing things and maybe I will find it. Either that or I will simply build a whole new printer from scratch.
-
Did you ever try it without the bed on? I see you mentioned it, but didn't come back and say whether you did or not.
What about printing a tall cylinder in vase mode? Print it pretty fast and it should give you the chance to watch things as the effector goes round and round.
-
@number40fan said in Banding on tall parts:
Did you ever try it without the bed on? I see you mentioned it, but didn't come back and say whether you did or not.
I did try with the bed off and it made no difference. I am going to try vase mode that is a great idea. Will let you know. I have new carriages and linear bearings coming as well.
I did notice the wheel at the top of the tower are all very loose, but I don't think that matters as it only allows the belt to roll and doesn't really effect things much. I did also notice the belt on one tower turning in and out as the printer went from lifting the carriage to lowering the carriage. Again with such a large band I suspect its not that directly.
-
The twisting of the belt has me worried. It shouldn't do that if under the correct amount of tension. Are the pulleys on top smooth or toothed?
-
@number40fan They are smooth. I tend to agree that something doesn't seem right. can you recommend a good pully for the belts on the top? It would explain why putting pressure on them helps,
-
Anything that is toothed. Do you have 20 or 16 toothed pulleys on the motors? Are the bolts passing through the upper pulleys 3mm or 5mm? I bought a pack of 5 or maybe 6 from Amazon that were toothed, 3mm through hole and even had a bearing for just a few bucks. Cheapest ones I could find, really, and they worked good.
If the pulleys on the motors are 20 toothed (and looking at your config, they probably are), you might consider getting 16 toothed all the way around. It will up the resolution from 80 to 100 steps/mm.
-
@number40fan said in Banding on tall parts:
, 3mm through hole and even had a bearing for just a few bucks. Cheapest ones I could find, really, and they worked good.
If the pulleys on the motors are 20 toothed (and looking at your config, they probably areNice, ok I will take a look. They are 20 teeth right now. The banding is very visible in vase mode but its very hard to pin point. I think that pull may be the issue as there is a ton of play. It is on a 3mm bolt and is free to slide forward and back. It also it does not have teeth so I can see the belt bounce with each movement. Keep in mind I don't think any of that is causing the banding pattern though it is hard to do. I am going to right a macro that just moves slow up and down over the area that bands. Does anyone know how to do "loops" in a macro or am I better off just copying and pasting a ton of gcode repeats manually.
