Vibration Compensation Firmware Open Beta for 2.0 Series (A/AT/F Models)

AT350. Upgraded from v 1.19.0 to 2.2.0. Luban 4.11.0. Using MacOS Sonoma 14.4.1…Using dual extractor and primary left nozzle 0.4.
Created material and print settings Speed Beta Normal, and Speed Beta Fast by copying the canned ones and modifying the new ones with new settings: Speed 120 & Acceleration 2500, and 125 & 3500. (most of the other setting calc from these…)

Large print filament runout issue. After 18 hours of nice printing, the filament ran out. I received the proper notification. I loaded new element (same type, color, and brand) and resumed printing. Almost immediately I heard very loud sounds and noticed the print head colliding on the model. I stopped the job, and took a picture. The print model looked very nice until resuming. The build plate was moved as shown in the photo.


I started another print job, and, as expected, the job did not adhere to the build plate. I re-leveled (only the automatic process not the other calibration steps), and a new print job on Luban. Print completed successfully.

I kept the same parameters in Luban and started another copy. Print failed with no adhesion. I re-leveled and started again. The print completed successfully. I started another copy and one corner lifted or curled up. Not adjusting anything, I did another copy and the print was excellent with virtually no stringing and no lifting up.



I did some small print jobs and resent them to the printer repeatedly. Prints came out excellent. After several times, some parts were good, but others failed. See photo. Whenever I encounter an issue, I go back and re-level the print head. (but only the 6x6 and auto z), haven’t had to do the other xy calibrations…

1 Like

@takeota @jenkins007 These two issues could be bugs. Could you please export the system Logs to us for troubleshooting?

To export system logs of the 2.0 models:

  1. Insert a USB flash drive into the Controller.
  2. On the Touchscreen, tap Settings > Export Log. Logs will be exported to your USB flash drive.
    Then please find all the log files (named with FW & SC) in the USB drive and send the files in this thread or to me via messages.
1 Like

I hope I can help!
I have been doing other prints and hope the log was not over-written for these items…

(Attachment SC_913.log is missing)

(Attachment SC_914.log is missing)

(Attachment SC_915.log is missing)

(Attachment SC_916.log is missing)

(Attachment SC_917.log is missing)

Firmware seems to work well here too, except for some of the noises the rails make (or could be the extruder), which is kind of odd since I did not change any slicer settings compared to the previous beta which did not cause these noises. Nothing too crazy though. Also, the vibration compensation was not enabled after flashing the new firmware and had to be enabled manually.

@DrSkiba Did you just play around with the frequencies to reduce the noise, or did following the ringing tower guide do the trick? So far I have not seen any ringing artifacts so I’d do it if it reduces the noise.

Yeah, for me, I did the ringing tower principally to reduce the rail noise. The supplied gcode provided by Jade is great for EI shaping (the beta firmware default). It really did make a huge difference for me doing that calibration, so I’d say it’s absolutely worth doing.

The other thing that helped with noise for my F350 was reducing the acceleration from the firmware default of 3500 to 2750. You can do that through the Luban serial console:

M204 P2750 R2750 T2750
M500

My F350 with the beta firmware is in a place now where I’m very happy with the noise for high speed printing. Do let me know how you get on and what works for you, I’ll be interested to see if there’s any further improvements I can try :slight_smile:

2 Likes

Great to hear, thanks for the response! For me the beta (even the first one) has been an absolute game changer. I used to have my heavily modded Ender 3 run some jobs because it was much quieter and much faster, now it’s just much quieter (even more so because it’s gathering dust).

When my currently running job is finished I will give the calibration a shot. I did realize that the “sand in the rails” noise could actually be CNC dust on my SBR16 rails… I’ll report back if I find something interesting :slight_smile:

EDIT: After performing the calibration, M493 was set to mode 11? Is that supposed to happen?

As Jade wrote is the supposed mode M493 S14 but hey, we can try every mode.

Here my insights so far:

Firmware seems a bit glitchy in the beginning, there were a few things which went wrong but did not repeat.

  • Filament runout, restart of the left extruder was about 5mm too high. Every runout in following prints where fine.
  • Ringin gcode of snapmaker had no temperature settings at the first job, it printed like this (with prevented cold extrusion)

I think the additional noise comes from my bracing kit and low tension on the steel strips, mainly I guess the steelstrip, in my case.

I did a lot of tuning and calibration so far but it’s really time consuming and I have not many spare time to use.

I would like to compare my machine calibration with others:

A350 + QS + BK (Quickswap + Bracing kit)
Dual Extruder

K-Factor:
M900 T0 K0.02 ;stock K0. 04
M900 T1 K0. 02
Junction Deviation:
M425 J0.02
Acceleration (not sure yet)
M201 X3000 Y3000
Vibration Compensation
M493 S14
M493 A19.62 ;(measured 6.5mm)
M493 B50.4 ;(measured 47.4mm)

As far as I understand it (and I’m sure Marlin mystics will correct me if I’m wrong), but I don’t believe the new Input Shaping mode uses the standard linear advance M900 settings. From the Marlin documentation, I believe Input Shaping has its own Linear Advance K values. From my experiments with altering K values on Input Shaping, it does not follow the same rules as the K values in linear advance (not sure if intentional or a bug with the Input Shaper, since it’s still marked as an experimental feature in Marlin). Any alterations I’ve made have really goosed printing.

Has modifying the M900 settings made a difference for your prints, or have you applied settings that have previously worked in a different firmware?

During the bed leveling process, vibration compensation will be disabled. Once leveling is complete, the input shaper setting will be restored to the one that you had set before running bed leveling. Unexpected changes to the input shaper setting are not anticipated.

1 Like

So far working well with an A350T, dual extruder (using left side),quick swap and bracing. I have noticed when doing the ringing tower I need to up my flow rate to 140%+. I have calibrated the e value but everything else is default for the new firmware. If I don’t significantly in crease the flow rate I get an awful print, is this expected or would this be more about the filament / a clogged nozzle etc.?

can you please share the snapmaker luban project for the shark(A350T/F350)? I would like to compare the settings.

@Jade the ringing tower gcode contains the following:

M493 S0
M203 X120 Y120
M201 X4500 Y4500
M204 S4500 R4500
M493 S11 P1 K0.0008

It disables input shaping, adjust some settings and then sets input shaping to S11 (not sure why). It is not changed later in the gcode. When the guide is followed, the required frequency is calculated and set using M493 Ax and M493 Bx. Then the user is instructed to execute M500 to save settings (including M493 S11) to EEPROM.

If the user doesn’t manually set M493 S14 it will remain at S11.

(sorry for the confusion, with “calibration” I meant determining the frequency using the ringing tower)

2 Likes

Oh, nice find!

Ooh, that is a good find @Mayco. I always thought the one Jade supplied was a S14 (EI) configured ringing tower. I don’t think the frequency will be majorly different with the different input shaping model, but I’ll print off another ringing tower with S14 set to validate. :slight_smile:

Edit: No difference to my ringing tower results when comparing a S14 printed tower to a S11. Frequency artefacts seem to be the same for those wondering.

1 Like

Hmm, I’d suggest to measure your E value again, as that does look like under extrusion.

You haven’t altered any pressure advance/linear advance settings on the firmware either have you (e.g., supplied a K value to M493)? That’s the only other place I’ve seen that kind of artefact appearing in prints.

Thanks for replying, just the E value was calibrated - I swapped the left hot end to the right and tried with a newer filament which improved things considerably, I get a much better vibration tower with the default flow (100%) now - I still got under extrusion though on the outside Y wall where the head stops and starts again at the corner. I’m guessing retraction maybe an issue here.

Doesn’t seem to be a thing for other people though?

it would be nice to implement fan speed for the enclosure, since it doesn’t need to be on full all the time. By having the option to reduce the fan speed it makes it much more quiet. I can do it manually but that not for everyone.

@jbnse Do you need this file?

@Mayco @xchrisd @DrSkiba Good find! After Input Shaper Tuning, the input shaper was initially set to S11, referring to Marlin. The shaper corresponding to S11 is ZVD. ZVD is more sensitive to frequency changes, which can make the model surface easier to see the difference. Alternatively, you can also change it to S14 according to your preference. Refer to the image below for the updated instructions.

2 Likes

@dandyfop It appears that the filament extrusion resistance is somewhat high. You might consider raising the printing temperature to address this issue. Another potential factor could be wear on the extruder gear. If you wish to modify the K value, you can send the code of M493 P1 K0.001.