Strange Ringing artifacts on coreXY

kuhnikuehnast

Would be great if it was that simple... As I only wanted to use high quality parts I installed Misumi long bearings... Bending isn't a probleme I think.

Steppings:
Already tried switching off interpolation and tried from full step to 1/32. But I will give a second try later (just out for dinner)

makerdise

@kuhnikuehnast - I am unsure if we did test the M350 X16 Y16 while setting the I flag to 0 explicit...

@T3P3Tony - for sure we did test a lot and discussed a bunch of common and uncommon things. It seems to be more complex in this case. I was focused on the mechanical subsystem since we did some tests today. As written in the OP, the belts, pulleys were changed, so they are out. We measured the rods, the movement was ~0.05mm. There's no measurable bending or movement of the frame. So we continued debugging within the XY frame, by measuring if the Y carriages does bend in x direction while moving x, and they didn't. The only measurable movement within a linear x move(as seen in the video) is into the y direction (+-0.2mm = 0.4mm). You can clearly see this sinus wave like pattern with a ball pen mounted on the x carriage painting on paper. In conclusion, we have to find all factors that can cause kind of a stuttering in the motors. As the motors were changed, it can be related to soft- or hardware issues.

I thought about an issue with M669 which was introduced in RRFW 1.1.9 - are there potential collisions with M667 or known issues related to CoreXY?

Maybe @dc42 can join the discussion to bring in some light into the darkness

mrehorstdmd

@makerdise

Yeah, I have the belt lifted Z axis. I recently put a Volcano and 0.8 mm nozzle back on it to print some structural parts and ran into some Z axis banding problems that I am starting to think are due to the gears in the worm drive that lifts the Z axis. I'll probably do a post on it to see what other think is going on.

Older prints made with a standard heater block and 0.4 mm nozzle in 0.2 mm layers don't seem to have the problem, or have it to a much lesser, more acceptable extent.

I have a CubexDuo with a relatively large diameter filament drive gear that produces little ridges in the walls of prints and seems to be related to the microstepping causing the motor to pulse enough to create the ridges. For some reason, changing the motor current changes the spacing/angle of the ridges.

makerdise

@mrehorstdmd - this is a real nice project. I like the worm drive, in my research I only found a very few ideas to stop the unwanted z movement and this is the best implementation.

OMG, another complex issue For sure, feel free to post it by pm or here. Did you thought about a PID related issue? Is it possible that the block itself is bending cause of this and the vertical alignment? Did you ever tune the PID's after changing the block? Are you using the volcano sock?

Do you have any detailed information on the motor pulsing, this seems to be interesting...

whosrdaddy

Question, what firmware are you currently using?

mrehorstdmd

@makerdise

The machine still has the smoothieboard in it, so I'll post here after I get it switched to the Duet if the problem persists. I was preparing to install the Duet and got side-tracked by the print quality problems that suddenly appeared when I switched to 1) volcano and 0.8 mm nozzle, 2) 0.4 mm layers, and 3) white ABS.

I'm hoping it isn't the worm drive...

kuhnikuehnast

Do not cap my thread

But back to topic:
Here a picture of my printer:
0_1526420009376_20180515_231101 (Large).jpg

And here the 2 tests:
on the left:
M350 Z16 E16 I1
M350 X16 Y16 I0

on the right:
M350 X64 Y64 Z16 E16 I1

Both look pretty identical...
0_1526420058702_20180515_232835 (Large).jpg

The 0.1 mm layer height is still printing.

makerdise

@kuhnikuehnast will arrange a test with unmounted motors and a flag on my recommendation...

hurzhurz

@kuhnikuehnast I had the same problem on my HEVO.
You may have a look at this thread:
https://forum.duet3d.com/topic/4246/vertical-banding

There is mentioned that the problem probably lies in a to small current and especially to few steps per mm.

I had 1.8° steppers with a maximum rated current of 1.2A from an aliexpress hevo kit.
Because of the thread, I then switched to these 0.9° 1.7A steppers:
https://www.think3dprint3d.com/motors?product_id=122

That helped a lot and practically solved the problem for me!

I drive them with 1.2A, 16x microstepping and 160steps/mm.

dc42

What is the pitch of the banding, and how many motor full steps and pulley teeth does it correspond to?

kuhnikuehnast

@dc42 The pitch seems to be 2 mm (full "sine wave") which corresponds to 1 tooth (GT2-Belts).
as I am using 1.8° motors :

80 steps per mm at 1/16 microstepping
so 5 steps per mm at fullstep
2 mm banding--> 10 full steps.

I yesterday did another test:
disconnetcted the stepper motors and connected the spools with jumper wire. (to produce a good holding torque). then I carefully tried to move the x-axis back and forward only using one finger and measured the maximum distance with a gauge-> this is +/- 0,1 mm, so 0,2 mm in total. And exactly the movement shown in the videos I posted...

cgcafer

Hi,
I guess you've checked it over and over again until now. but I want to ask you one more time. how are the spaces between the rod and the bearings(bushing) mechanically? and if you are using elastic coupling and the load is too much, you may have problems.

kuhnikuehnast

@cgcafer don't get what you exactly mean? The rods and the linear bearings? The linear bearings (Misumi) are directly placed on the rods. The linear bearings themselves are hold with a really thick printed part. You can't measure any flex while printing.

Or do you think of the Idler? There is a screw that holds them in place. As I first changed the Idlers (after ~ 1 month of test prints) you couldn't see any damage on the screw..

robm

@kuhnikuehnast,
As per comments and details in the vertical banding thread, I've had this problem for a while. I found some improvement increasing the steps per mm by using smaller pulleys and 0.9 degree stepper motors, and by vibration isolating the steppers, but I would not say that I have eliminated the problem.

You noted that you have changed the pulley and idler wheels, have you tried different tooth count pulley wheels?

This page under 'Stepper clipping' references this work on a similar pattern, but this is traced to different motor driver chips and stepper motors than either of us are probably using.

Mine is a DIY Prusa i3 design, some claim that these are more susceptible to motor vibration than other designs; possibly the Kossel I am building will be free of the problem?

dc42

@kuhnikuehnast said in Strange Ringing artifacts on coreXY:

@dc42 The pitch seems to be 2 mm (full "sine wave") which corresponds to 1 tooth (GT2-Belts).
as I am using 1.8° motors :

80 steps per mm at 1/16 microstepping

so 5 steps per mm at fullstep

2 mm banding--> 10 full steps.

I yesterday did another test:
disconnetcted the stepper motors and connected the spools with jumper wire. (to produce a good holding torque). then I carefully tried to move the x-axis back and forward only using one finger and measured the maximum distance with a gauge-> this is +/- 0,1 mm, so 0,2 mm in total. And exactly the movement shown in the videos I posted...

In that case I think the problem is definitely related to belt teeth and not to the motors or the firmware.

Phaedrux

If possible, try what I did to solve this for my DBot. Replace the rear toothed idlers with smooth ones and give the belt a flip.

kuhnikuehnast

Did some further tests now:
I flipped the belts, so all Idlers are run on the smooth side of the belt (and are changed to smooth Idlers with no teeth) except the Pulleys of the Stepper motor. Therefore I had to flip the belt. This reduced the ringing a lot. It is still there but almost gone. Here a photo of the new belt-lineup (only for one motor otherwise my painting skills would be to confusing )
0_1526899461325_20180521_123135 (Large).jpg
I also had to loose the belt tension a lot to reduce the ringing even more. So it is still visible but you can't feel it any more...
0_1526905575813_20180521_131937 (Large).jpg

Also performed a second test:

changed to 0.9° stepper motors to increase the steps/mm. (and also tested 16 teeth Pulleys instead of 20. So I could compare the print quality from 200 steps/mm with 80 steps/mm… But there was no difference.

Next, I will try to replace the belt by a fishing line. So there won't be any toothed pulleys in the hole system any more. (the line just wrapped around the motor shaft 5 times to ensure enough friction). I know, that this isn't a solution, but it may help to find out whether the ringing is really caused by the belts...?

It is really strange!I think I just masked the problem with this conversion, but I did not fix the cause…
Almost everybody else uses the Hypercube evolution with a regular belt line-up...?

kuhnikuehnast

So, next test finished:
The problem seems really to be due to the toothed belt... Using fishing line (the line wrapped around the motor shaft 5 times) the ringing was completely gone! (Despite by a damn poor accuracy )

Now that the culprit has been located, the question remains: Why do I have a ringing problem while others don't? And what else can I try to solve the problem?

Phaedrux

It's not just you having ringing though. I'm going to be testing some supposedly genuine gates belts soon to see if that helps, but I wonder if this isn't more of a common problem than you'd think.

kuhnikuehnast

Sorry, didn't get it right what you meant...? So do you think more people than me suffer from this problem or do you think it is a specific common problem on my printer?