Yeah I get it tho, I had to slap myself a few times before I purchased some of the stuff to take measurements that would cost me more time and money then an extra print with different specs 
Something I just thought of: there are several settings with the slicing software that can have a pretty big affect on dimensional accuracy. The ones I can think of that I have changed in Cura to get better dimensions when needed are…
- Wall Ordering
- Wall Line Count
- Extra Skin Wall Count
- Infill Before Walls
- Slicing Tolerance
Also, from the way you described how you are measuring the cylinder, if your tool is measuring the first layer, it may be picking up the “squish” inaccuracy caused by the need for the filament to be slightly smashed into the bed in order for it to adhere properly. This “initial layer” almost always has a slightly different dimension than the layers above it. In theory, it is feasible that inaccuracy could also change at any given point based on numerous other variables, such as the direction the print head is moving at that time, the texture/friction of the bed, the filament in use, the print cooling fan output, and probably others I cannot think of.
Do you get that same dimensional inaccuracy if you measure the center of the cylinder?
1 Like
Yes you bring up some good points. I have also truncated some of my results and explanations.
Measurements are equivalent (+/- .02mm ish…) no mater the z height the measurement was taken at along the cylinder.
If you spin the cylinder in between your fingers its obvious there is two “flat” spots 180 deg apart from each other totally from bottom to top. Across these flats has always been my min measurement and across the points at the end of the “flats” has always been my max measurement.
I Purposely do not measure the first few layers. even though i have very low elephants foot as is. (this took a while to dial out. But e-steps on the extruder showed the most improvement.)
the linier caliper i have been using has about 2mm off set between the back of the lcd screen main case and the lower part of the jaws. so when set flat on a reference surface the jaws measure a point about 2mm off the surface. Im also using the larger flat surfaces deeper in the jaws of the caliper then the sharp ends.
I have the same results in multiple slicers. Well Luban and cura at least debating setting up prusa just for this test.
Wall ordering: have tested both ways but that was before i was logging data for this issue. Didn’t seam to make a difference.
Wall line count: haven’t changed it. currently @ 3
Extra Skin wall count: from what i understand of the help section this would only effect the upper and lower layers. Roundness error shows up on all layers.
Infill before walls: i have another test that has no infill, more of a thick walled cylinder Still shows same issues.
Slicing Tolerance: Something i want to test out. But with these print in place tolerance tests having the slicer pick the inside, out side, or center of a line doesn’t matter if it is done to both lines equally? Mines currently set to middle.
I have G90 arc on.
Max path deviation is at 0.05 Path Tolerance 0.5%
I have turned the slicer resolution way up.
Wouldn’t the direction of the print only matter if I had any measurable backlash?
I don’t see how the roundness of the upper z areas would be effected by the bed surface. If the first layer is adhered stoutly then upper levels shouldn’t have issues.
Basically I cant pass any of the simple tolerance tests. which is annoying seeing the mechanical results I get from this machine.
1 Like
I am a new user to 3d printing and with the A350T printing a circular 2 part model requiring low tolerances as they are to fit inside one another. Upon completion of the second piece I discovered the internal piece would not fit inside the outer housing. After measurement, with calipers only, it became evident that the circular model was out by 1.5 to 2mm which prevents the model pieces from fitting together. So it seems there is some sort of common issue with circular dimensional accuracy that may require some expert attention. I did purchase this higher end 3D printer expecting precision for prototypes I am printing for an engineer. Maybe snapmaker developers can take a look at what seems to be a common issue.
Teaching tech Calibration.
https://teachingtechyt.github.io/calibration.html
After you have walked through
1: Extruder E-steps
2. Temperature Tuning
3. xyz Calibration
Those should get the machine “Physically” correct or better.
Then you need to work on the profiles.
I would recommend getting those all in order and getting a really nice calibration cube consistently.
Then work on the roundness.
Your out of roundness is an order of magnitude bigger then what i am seeing so i think XYZ calibration would be a really good start for you.
Let us know.
Keep in mind that the Snapmaker is really NOT a “higher end” 3D printer. That seems to be a common misconception based on its price tag. Those kinds of machines will be in the $1000+ price range for 3D printing ONLY. The price of the SM is because you are also getting laser and CNC functions with it, so it would be better to consider the value of each function to be about 1/3 of the total price. A lot of people have run into disappointment with the SM because they initially view this 3-in-1 machine and it’s price to be comparable to dedicated 3D printers of the same price. It’s definitely not, and nor is it plug-n-play like many people expect it to be like some other machines as well. Its nature of being modular and assembly-required design dictate that it does require quite a bit of calibration to work accurately and reliably. Though there are still some issues which need to be worked out, most of those seem to be specific to individual machines.
Agreed in most ways. but even at its price i would say its an order of magnitude less expensive then a good 3d printer.
But man it can be frustrating sometimes.
Mxbrnr
Is there any build plate flatness info that would help your theory that i could provide?
1 Like
Ahhh so true, I guess I did over look the 3 in 1 fact for sure but I did feel that the assembly was quite precise so expected higher quality than the Ender 3 V2 which has actually been an incredible little printer. I also bought the A350T for the build plate size. I actually have not even tried the other two functions yet in my endeavor to get a 3D project completed. Thank you for your input.
Regards, Perry
1 Like
Thanks for your suggestions. I will go through some calibrations again. I did find that filament changes can require E step calibration but I have to be honest on my lack of xyz calibration. Is there a procedure on the forum for this. Thanks
Regards, Perry
I think I have the most common variables figured out and how to test them, just a really tedious process now to make sure the tests are reasonably thorough, and done in the right order.
Have you noticed or thought of something specific that you would like to make sure gets looked at? The list of things I am currently set to test are…
- Platform flatness delta when installed
- Stock probe measurement accuracy
- Heated bed delta when both not installed and installed
- Delta change based on print sheet/build plate orientation
- Delta error induced by removal/re-install of heated bed
- Delta error induced by removal/re-install of print sheet
- Delta change through tramming the Z-sliders/x-rail according the manual vs using 1-2-3 blocks
- Best delta based on various heated bed screw torques
- Best delta based on various heated bed screw tightening patterns
- Difference in delta based on heated bed cold assembly and cold calibration vs hot assembly and hot calibration (and several variants thereof)
- Affect on delta by using a glass build plate
- Delta correction via shimming
I am probably a little over halfway through the testing; the longest ones are the tightening patterns and hot vs cold tests.
1 Like
Check out that link,
nothing really on this forum for xyz calibration.
WOW!! just a few data points.
I was just wondering if you wanted my plates height map or anything.
1 Like
Don’t get discouraged. The machine and the “full” setup process can be intimidating, but once you get it figured out, it’s a lot of fun.
If you do your e-steps calibration a certain way, it will actually be a one-time deal, and does not have to be adjusted again. A lot of misinformation has been passed around on that missing quite a few key notes. Here is a write up of how I do it…
E-Step Calibration.docx (12.9 KB)
2 Likes
Sure! There are a few ways I can use it.
Removing the hot end was not in the other procedure I found. I look forward to giving your steps a try for sure. Thinking about the calibrations, I actually had to change a hot end because the PETG had it plugged up almost solid even with heat, not sure why it did because I had printed out about half a spool. I did complete the E-step calibration with the new hot end and had installed PLA+ as the filament. The calibration seemed to go so much better than with the PETG but maybe the nozzle was partially plugged at that calibration. Any way your information is much appreciated and I am pleased to know that E-step calibration is not effected by a different filament.
1 Like
By removing the hot end for the e-steps cal, it removes the flow restriction from affecting the calibration and giving a faulty accuracy. It is one of the very commonly missed yet important steps that most guides never mention. Using F80 (extrusion speed in mm/min) instead of the F100 or F300 that many guides suggest also makes the calibration much more accurate because it is closer to the typical speed that the filament will extrude at during printing. Speed is one thing that does affect e-steps a bit. If you run the new vibration compensation firmware and will often push your machine to the higher print speeds over 60mm/s, you will probably want to adjust your e-steps to match. I have a math formula for figuring out the best speed to use for the e-steps based on your typical print speed.
The hot ends have a PTFE liner in their heat break that is notorious for melting and clogging the hot end when using temps around those needed for PETG. It seems to be really hit or miss as to which liners will melt. I had one melt on me after only about 25hrs of printing PETG, and another that has over 300hrs on it and is still in perfect shape. There are a few options for either changing out the liner for a higher quality one, or changing out the heat break for a bi-metal one that changes the location of the liner so it is less prone to melting. Retraction settings in the slicer seem to have the biggest affect on hot end clogs, as the default settings in Luban are too high and end up pulling melted filament back into a cold section of the heat break where it will solidify and can basically never get melted again. That default setting is one of those poor oversights that SM should have paid more attention to, especially since it takes all of 5 sec to adjust.
1 Like
Mxbrnr and Vettester
Yeah removing the hot end to do a e-step calibration is nice. But not something you can do easily with the new dual extruder head. If i never mentioned it in the past, all this cal work has been done with the new dual head. While i have printed dual color and with support material (wow). I haven’t done a full calibration on it either. Still working the kinks out as a single extruder.
Mxbrnr the new plate leveling with the dual head extruder is… interesting… somethings I like somethings I don’t. It does the z wave pass with the inductor. and then touches off and pushes up the nozzle until the retracted sensor is tripped. (leaving little marks on your plate and causing some flex)
But i have to say so far its been touch button, come back after leveling pass and hit print. And its been successful 95% of time, I’m impressed with it to be honest.
I saw on your other posts you added the after market inductor add on. with how I assume the artisan dose its leveling and how the 2.0 does with the dual head it would be interesting (I may be able to do this) to do a level with the old upgraded print head and the new dual head and a dial indicator all on the same day with the same heat. ill have to get octoprint up and running to do that though.
1 Like
The leadscrews from that period were not particularly accurate. And even then you’ve measured only a small section; do not assume other sections have the same average pitch. This can be compensated for in software if there’s measurement data available on the actual screw; SM has no such compensation.
Here’s a link to how Mach4 deals with the problem of lead screw tolerance. Mach 4 Screw Mapping | MachMotion I don’t expect SM to ever get anywhere close to this.
I’m having the same issue with cylindrical prints. At first I thought it might have been because of some extra extrusion at the z seam, so was starting to do some tests to improve that when I noticed something a bit odd.
Making a 20mm diameter test cylinder with just 2 walls has shown that just after the z seam it does indeed get thicker, but it is the same the opposite side of the cylinder as well. Not only that but the 2 walls actually separate slightly. As best as I can tell from measurements, the outer wall bulges out a little and the inner bulges in a little.
You can see the bulges and the gap in the walls at the top left and bottom right of this print. My wall thicknesses are normally close to perfect, and between these bulges they are near enough spot on 0.8mm. But just before the bulge they drop down to 0.6mm and at the gap they are 1.1mm.
My next guess was this could be down to the xy calibration, but I’m a little confused about that too (see here Confused about xy calibration)