350mm V2 Ghosting Gremlins

[stevereno30]

Printer Model

Voron 2.4 350mm

Extruder Type

Galileo

Cooling Type

Stealthburner

I have a 350 V2 with a seemingly impossible-to-solve bit of ghosting. It is immune to input shaping, doesn't seem to be related to the toolhead, and only appears on y-movements (meaning, I assume, the X axis is its origin). It is more pronounced when the toolhead is moving backwards toward y-max, but it also appears during movements toward y-min as well.
Some info on my Formbot kit circa 2021 machine:

• X axis rail is a CPC MGN12 equivalent with high preload
• Y axis rails are TRVIS MGN9 equivalents with medium preload.
• Y rails have backers (Formbot MGN9 rails)
• Vitali CNC Tap with high preload TRVIS MGN9
• I am running Nitehawk as umbilical
• Toolhead is a Stealthburner with Galileo 2
• Hotend is a Dragon or a V6 or an X1c (these were printed with the X1c, though all perform the same in these tests)
• ADXL graphs look good (see below)
• Gantry idlers and pulleys are genuine Gates, as are the belts
• Gantry bearings are BOM RainDew parts
• Pins mod on all gantry parts
• Front idlers are VoronBFI

Some things I've tried without any effect:

• Printing with 2hump_ei at 60hz, 50hz, and 40hz per Klipper guide
• Swapping Tap for a stock V2 X carriage
• Cleaning rails thoroughly and greasing with EP2
• Swapping all gantry rails for ones with lower preload
• Swapping A/B motors (using BOM OMCs currently, but also tried Moons')
• Changing belts
• Changing front idlers

All images below use a 35mm tall STL oriented so it is printed with mostly y moves that looks like this:

All prints were printed with accel set to 1500 and speed set to 52 unless the tuning tower demanded different settings.

Image of the resultant print	Tuning tower command utilized (purpose of test)
	TUNING_TOWER COMMAND=M220 PARAMETER=S START=50 STEP_DELTA=10 STEP_HEIGHT=1 (run the machine at increasing speeds, starting at 50% speed, topping out at 350%)
	TUNING_TOWER COMMAND=SET_VELOCITY_LIMIT PARAMETER=SQUARE_CORNER_VELOCITY START=1 STEP_DELTA=1 STEP_HEIGHT=3 (run the machine with increasing SQV settings, starting at 1mm/s, ending at 11)
	TUNING_TOWER COMMAND=SET_VELOCITY_LIMIT PARAMETER=ACCEL START=500 STEP_DELTA=500 STEP_HEIGHT=3 (run the machine with increasing acceleration, starting at 500 mm/s2 and ending at 5,500)
	TUNING_TOWER COMMAND=SET_INPUT_SHAPER PARAMETER=SHAPER_FREQ_X START=30 STEP_DELTA=5 STEP_HEIGHT=3 (run the machine with shifting x-axis ei input shaping frequency, starting at 30 Hz, ending at 85)

 

[NoGuru]

Belts look good, all the graphs look decent. the polar angle is a little off but over all things look good.
Have you tried with MZV or 3hump to see how it turns out?

 

[stevereno30]

I have not tried 3hump, but I can (printing now). MZV is what I've used for most of the printer's life, but I switched to ei to try to solve this issue. I have tried tweaking the polar angle, but nothing makes a difference. Swapping the A motor to the B spot and vice versa has no effect on the polar angle graphs, so it is somehow related to where the motors are, not what the motors are.

 

[stevereno30]

Here's 3 hump.

 

[NoGuru]

So a slight improvement?

Two things that I might try is, change stepper drivers. If you are on 2209's maybe some 2240's.
Another option is tuning motors/steppers. I use Autotune for simplicity. https://github.com/andrewmcgr/klipper_tmc_autotune

 

[stevereno30]

Yeah I use tmc autotune currently. I'll try 2240s to see if they make any difference. Do you happen to know if there is a way to run a Voron with ancient DRV8825 drivers? I have those lying around and would rather try something I have before throwing more money at my problem...I'm certainly approaching $200 of thrown away money trying to solve this issue.

 

[NoGuru]

Yeah I use tmc autotune currently. I'll try 2240s to see if they make any difference. Do you happen to know if there is a way to run a Voron with ancient DRV8825 drivers? I have those lying around and would rather try something I have before throwing more money at my problem...I'm certainly approaching $200 of thrown away money trying to solve this issue.

I don't think Klipper is compatible with DRV8825

I understand not wanting to put any more money at it. I ended up swapping out all of my idlers for Gates to help with this and it was pretty expensive and time consuming.

 

[stevereno30]

I have Gates idlers and pulleys, though I don't think they made any difference relative to what came in the original Formbot kit :/

Thinking through the list of things I've tried, I think drivers and extrusions/bolts are the only thing that have not been swapped for something different on my gantry.

 

[stevereno30]

So the ghosting is not from the x-axis, even though it only appears on y moves. I verified this by printing a 35x35x0.4 "cube" so print moves would only be along a single axis. I ran the print four times: one that only had X moves, one that only had Y moves, one that only had A moves, and one that only had B moves. The ghosting appeared on everything but the X moves print. Because of how thin the prints are, I was able to photo them with light passing through. Since the A and B prints were unremarkable, here is a pic of the X (left) and Y prints:


The X part does have some VFAs, but the Y part has ghosting on the right and weird diagonal artifacts on the left. I assume both the ghosting and the diagonal artifacts have the same origin. There is apparently some springiness in my printer along the Y axis. Rather than my toolhead resonating as it ends/starts x-motion, my printer is bouncing or slip sticking during Y moves.

So... any ideas?

 

[stevereno30]

This is what I think is happening...

 

[newauh]

I have the same configuration and print the same results as you, did you finally solve the problem?

 

[stevereno30]

Unfortunately no. I never tried NoGuru's suggestion of using different TMC drivers, so that could be it. I've come to an uneasy peace that my machine, despite my best efforts, cannot print PIF quality parts. Every once in a while I start caring again, but my obsession is in remission right now.