Skewed prints after Asura Distortion fix

#1

Hi all
So I recently grabbed 2 heaters for my moai and decided to run both the advanced calibration and Asura Calibration stl on both machines. After inputting both and doing several test prints, I’ve noticed that my prints are way more squished and performing poorly. Prints that were fine with the default setting on Asura were decent, but I have some models with parts that need interlocking pieces, and figured I really want to negate as much of the distortion as I can. In doing the advanced calibration I noticed that the cross test was a bit off so I manually adjusted that to be centered on both machines, mostly in the “X” value they seemed off.
Has anyone had great success with that process of using the Asura calibration stl and doing the basic math to set the numbers in the document?

0 Likes

#2

Sorry to hear your issue, here is a survey we use to help us get as much info as we can from user so that we could provide best solution. Even if you might have provide some info already, please answer the question for the completeness.

support question:

Room temperature (best to be >25C):

What type of resin:

Build plate type: (Standard 1-piece / easy to level)

Vat: (Silicon vat / FEP vat)

What type Vat: (Peopoly Original vat, Recoated Peopoly vat, custom vat made by user)

Leveling measurement:

A, B, C, D (example 10.9mm)

Firmware version:

System Setting is default per Peopoly recommendation? (Yes/No)

Laser Power:

Z-reset position:

PM Motor Speed:

Z Follow Speed:

Software used to slice?

If using Asura,what version and profiles

If using Cura, what version and Profiles:

Infill %:

0 Likes

#3

also attach some pictures of the print, that would help a lot

0 Likes

#4

All settings are standard as per the manual.
Room temp is set by 2 Moai heaters inside each unit.

Peopoly clear resin

Standard build plates

Silicone vat in both units

2 Peopoly vats

Measurements for both units were between 10.78 and 10.81 for all corners

Firware is latest 1.18

Settings were default per recommendations

Power of 57

Software was Asura using 2 custom profiles based on the Asura calibration stl

Z reset was standard

Pm motor is standard

Z follow is standard.

The only thing I did was to do the advanced calibration where when I printed the circle I noticed the laser was a little off on both the x and y in BOTH machines. That was rectified, then ran the Asura calibration stl, did the measurements and properly inputted those into Asura as two custom profiles.

I uploaded three pictures. One is of my Frankenstein’s Monster successfully printed. The other is the result after Asura calibration and the model is very squished on its x axis. The third pic is another toy where you can visibly see the cylindrical shape is no where near circular but oval. They all were aligned properly with their highest sides place near the red zone in Asura meaning that the distrotion was mostly based on the x axis.

0 Likes

#5

It looks like you may have made some error in your measuring and calculations for the calibration cross. Did you print the cross again after loading the new calibration file to verify the new numbers were working?

0 Likes

#6

I need to do that, being reprint the X Y cross again. I will do that and report back.

0 Likes

#7

Do you measure the calibration cross on the platform still or post peeling it off of it and being cleaned?

0 Likes

#8

I measured with the cross attached to the build plate. Just washed it with some alcohol to remove the excess resin.

0 Likes

#9

I ended up updating to Asura 2.21 last night and basically reset to 100% default settings all around. I ran the advanced calibration on both machines and lined everything up I even used the same spare vat in both machines when doing this to ensure consistency. Ensured the galvos were fully in their slots, the y tends to slip down. Made sure potentiometers were good too. Ran the Asura calibration cross at 100um using the base distortion setting supplied with Asura as my baseline. I have attached the pic of those two prints on my build plates. The x positive sides seem very close to, if not going over the edge, yes the model is centered in Asura also. I’ll measure these then reprint tonight. Having two machines down at the moment is really hurting my productivity, really hoping I can get back to decent prints asap.

Also should I be concerned that the calibration circles and cross calibration files were centered and traced the circle just fine, give or take some extremely small deviations, yet these 2 prints did not print out anywhere near center?

0 Likes

#10

you should export without any kind of calibration, so you need to set it to none first. If you are trying to create your own calibration file, you may as well just delete the Peopoly one. If you printed these with their calibration file selected, you need to do it again unfortunately.

Don’t forget to reset your Moai settings too, so X and Y back to 900/900 etc.

0 Likes

#11

Ah darn. Ok I’ll have to redo it tonight. I felt like I read the process over and over. Maybe that isn’t documented too well or it could just be me missing it. Thanks though.

0 Likes

#12

@peopolysupport Here is a new mystery I have noticed and this potentially could be one of multiple things I’m seeing happen. When I exported the base asura calibration cross with NO profile, it looks as it should in the asura viewport. If I then chose the same model and export to a different profile, for example my left machine ( i have 2 moai), and then I hit “center” in asura, to assure nothing has moved. the entire stl changes shape… Now I would assume that Asura would never change how the model should look, because if I suddenly chose a different profile and I do not hit “center” on the transform plane, I technically would be exporting something incorrect regardless of my calibration if I’m not mistaken. I’m adding several pics. First is of the model with no profile exported. The second is when you export with the moai default calibration, you can see a size difference, third and final you can see my custom one and how different this one is. Asura seems to be changing the model internally and displaying it incorrectly, I’m beyond confused atm as I do not see why any software would change the scale physically yet the numbers for scale remain at 100%. Finally I have a pic showing the differences using the difference filter from photoshop, you can see there is obvious shifting, which is the correction from the file, but why would Asura generate and warp the model in the viewport? It should just stay 1:1 nothing different to avoid confusion.

0 Likes

#13

Here is a gif screen cap of what I’m seeing on my end, it’s easier to see it in video format that still images. Based on this I would assume I need to unload and reimport the model each time to ensure I do not have a warped file going to my printers. This all might be visual trickery but sadly I do not want to waste anymore costly resin until this is solved. :frowning:

http://recordit.co/FBh1STA6D2

0 Likes

#14

set the calibration to none and print that file
about the problem in the gif you sent, it seems that you export and center on the same model, making the commands starts to overlaps on it self.
I’m not sure why the model got changed but for now you don’t have to keep adding center command since the model is already centered

0 Likes

#15

@JackyBoy I used your resin saver file to print my 2 machines calibration crosses, thank you for that. I did as said by yourself and @peopoly, printed both with NO calibration file, measured those, then did the calibration math for both machines, having another person check all my math on all numbers. Now I printed both new calibration crosses with the proper profiles on the correct machines and almost all of my numbers except for the initial “61”'s have enlarged by almost a constant 2mm so for example:

my -X 51 measured -51.31, pretty close. I did the math 51+(51.31-51)/2 = 51.155. When I reprinted and remeasured that value is now 53.32!?!

This seems totally off, I’ve probably printed way to many of these calibration crosses at this point but I honestly am at a loss. The end goal is to get closer to the true numbers of, 61,51,41,31,21,11 if I’m not mistaken. Yet my calibration circles and cross hairs are very very close to center. This is a huge mystery that I don’t have hours and hours of lost print time to dedicate to, are there any suggestions? Do I need to reflash the firmware on BOTH machines for some reason? Both machines are set to x and y size of 900, leading me to wonder why my measurements suddenly have 2mm more on almost all sides.

1 Like

#16

@timsep

one the program shifting as you keep loading and center. that is probably a bug. It may be best to just load one, run the calibration and save gcode and exit the program before doing another one. I will have the programmer to check memory clearing as I suspect there is some bugs there.

I am surprised to see a spike in difference after calibration. It has been pretty accurate for the users who wanted to run their own calibration. Can you share more detail data of you calibration, the file, the uncalibrated results and the calibrated results.

0 Likes

#17

@peopoly Please have somebody edit the Asura “Distortion Adjustment function” entry on the Wiki to specify that “none” should be selected in the “Calibration” field when exporting the first “asura-calib-print-Moai.stl”. As of Asura 2.21 it defaults to using the included calibration file (at least it did for me). Without any mention to select “none” calibration for the first print, I used the default settings not knowing this is not what we are supposed to do. I have printed 3 parts with no improvement in accuracy only shifting degrees of deviation (I believe because of the fact I didn’t start with the proper setting). I am now starting over (after finding this post), and printing a “none” calibrated part (my 4th 2.5 hour print).

Edit: I just wanted to post an update with my results after the PROPER Asura calibration process. Using the Peopoly Standard Black resin with default firmware settings (excluding laser power of 57 to help over-curing to build plate) and ambient temp between 21-24C after the first calculated compensations were made all measurement points were within 50μm of model geometry. This is vary reassuring as high accuracy is of upmost importance for my work, and I was getting worried with my results up until this point, and from what I have see from some others members on this forum.

I plan to run one more round of calculation/compensations tomorrow to see how close to perfect I can get it. After I get it up and running (as well as have time available) I plan to model a 3D Calibration cube that I will make available when it’s finished.

1 Like

#18

I have done a few calibration crosses trying to figure this out after reading and re-reading the calibration instructions. I didn’t select “none” until right now. I sure hope I can get this thing to be more accurate. I have been trying since November when I got this thing to get an accurate print. They are always off by about 0.2-0.3 mm. For jewellery, this is the difference between a usable print and not. I need accuracy.

0 Likes

#19

Was it ever solved as to why the calibration cross was not printing centered?

Mine do this exact same thing. After using cross-center.gcode, I believe I have the laser as close to center as I can possible get it. In Asura the model is in the middle, on the print it nearly falls off the edge of the plate.

0 Likes

#20

I never solved it. I actually used the circle calibration tests to make sure everything was correct and now just use a “no calibration” setting in Asura. I haven’t had time to post numbers as I am too much into production mode now. I actually had to purchase a different brand machine to assist with correct tolerances and calibration.

0 Likes