Question Cylindrical parts not quite "round"

[billgeek]

I'm not certain if this is hardware, firmware or slicer related, but I've been running my Trident for about six months now. I thought the print quality is pretty decent, but I've since purchased a Bambu Lab P1S and this made me realize what "round" actually means.
Circular objects on the Trident tend to come out as "segmented" lines.

I followed and dialed in my printer using the excellent guide by Ellis for tuning the machine and am using BambuStudio to slice my model.
I've tried this with "Arc fitting" enabled or disabled on the Trident, but the result remains the same. (I needed to add "gcode_arcs" to the config file on my Trident and set the resolution to 0.25mm - any lower and I get "timing" issues)

This first picture is from the Trident. Notice the "wobble" on the lower half below the knurled grip. I took this picture in the worst possible light to exaggerate the issue:



This next picture is from the P1S. Aside from a few lines appearing slightly darker, the walls on the lower half are perfectly circular. (This picture was also taken in the worst possible lighting conditions)



I also own the Creality K1 and it looks the absolute worst, with the lines creating a texture on the cylinder:




So I guess the question is: What could be the cause of this, and how can I go about addressing the issue? I fully realize that the Bambu is a proprietary machine and all that, but at the end of the day it's "just another" CoreXY machine.

 

[Ranayna]

Doesn't this essentially come down to the steppers?
Steppers have a finite resolution, round can not really be round, ever.

Now i am really curious how the Bambu gets this roundness.

 

[NoGuru]

This is VFA and comes down to tuning. Being Machine or slicer.
The K1 picture looks like to much belt tension to me. Each like looks like it can match the belt teeth.

As far as your Trident goes, you can try slowing down a little in speed and accel. Or try tuning the motors and steppers more.

 

[Dave32]

Steppers have a finite resolution, round can not really be round, ever.

3D printers with the Polar motion system would disagree!

 

[Dave32]

The Trident print looks more glossy, as if it was printed at a higher temp? Printing at lower temperatures will make the filament more viscous - which provides a natural "smoothing" to the extrusion flow.

Also, check your model to see how the facets look. It's a bit counterintuitive, but the better a printer is tuned, the more accurately it can reproduce the original model - INCLUDING its faults.

 

[ChrisA]

Doesn't this essentially come down to the steppers?
Steppers have a finite resolution, round can not really be round, ever.

Actually stepper don't have a finite resolution. They can be driven in a continuously, smooth, non-stepped mode. After all, a stepper is nothing more then a two-phase BLDC motor with many pole pairs. That said, you =are not going to be able to drive them in a stepless mode with the electronics and softwarte typicaly found inside a Voron printer.

The other thing I'd look into is the idlers. Are you running the toothed side of the belt against a flat pulley? This is a design error and should be fixed. I'd bet you Bamboo printer does not have the same error. Gates makes toothed idler pulleys and also smooth pulleys that have a better profile than stacked flanged bearings.

Bamboo has raised the bar simply by doing everything right and by not limiting itself to commodity and plastic parts.

Running a stepper in contiinous mode requires quite a lot of hardware. Each motor needs a good rotray encoders and a typical MCU can drive only one motor befor it runs out os "compute power". Youcan Google "FOC" otr "Field Oriented COntrol" for details. I've got a demo running on my desk. An RP2040 does fine, an ESP32 does better. A mid range STM32 would work too. Then you need a $15 sensor attached to the motor shaft. Ho, and you need to measure current in both phases.

The other op=tion for continously motion is industrial servo motrs as found on higher-end machine tools. But these are $200 motors, even in the tiny sizes we needs and the controllers are $100+ per motor.

The next thing we need to do is get away from STL files and use CAD files like STEP. These file contain the perfect precision mathmatical models of the shape, not a bunch of straight-line triangles.

And while on the subject, we need to print with the Z-motors running and not just on slice planes. Look at how 3 and 6 axis CNC machines work.

The future is exciting. There is so much more to be done. Today in 2024, the low-hanging fruit is mostly gone, getting much better will take a LOT more work and cost.

But for now? Try setting the number of microsteps as high as you can and print slower. And do NOT accept a 0.25 mm error. That will put flat sides on your round parts. If the computer can't keep up, go slower.

 

[ChrisA]

Also, check your model to see how the facets look. It's a bit counterintuitive, but the better a printer is tuned, the more accurately it can reproduce the original model - INCLUDING its faults.

I think the Bambo was printing the exact same STL file as the Trident. The Bamboo printer is doing a better job of it. The question is "Why?" what is it doing that the Voron printers are not doing. It might be informative to look at the g-code.

 

[billgeek]

Thanks for all the great responses! To address some of them individually:

This is VFA and comes down to tuning. Being Machine or slicer.
The K1 picture looks like to much belt tension to me. Each like looks like it can match the belt teeth.

As far as your Trident goes, you can try slowing down a little in speed and accel. Or try tuning the motors and steppers more.

I've seen VFA's mentioned on the K1 across the web and suspected the same, though I see no VFA's on rectangular objects only cylindrical.
The consensus online to beat VFA's on these machines appear to be increasing print speed, though I'm using Silk PLA so I'm pretty limited to the speed here.
Currently all machines print at 100mm/s outer wall and 150mm/s everything else. (support, infill, inner wall, etc...) I'll slow this down to see if it helps.

I will definitely check belt tension on the K1 machines as I didn't do this when I first set them up.

Finally, regarding the belt teeth: (and the following comment by ChrisA)

The other thing I'd look into is the idlers. Are you running the toothed side of the belt against a flat pulley? This is a design error and should be fixed. I'd bet you Bamboo printer does not have the same error. Gates makes toothed idler pulleys and also smooth pulleys that have a better profile than stacked flanged bearings.

The Bambu and the K1's both have the toothed side of the belt running on flat pulleys, (I was actually quite surprised re: the Bambu being built this way) whereas the Trident has geared pulleys as per the build spec. (Not Gates pulleys, mind you - just generic ones)
I will print a set of XY joints for the K1 this weekend and see if this helps. I have more of the generic pulleys lying around which I'll use to test, but if this improves even slightly I will definitely acquire some of the Gates pulleys.

The Trident print looks more glossy, as if it was printed at a higher temp? Printing at lower temperatures will make the filament more viscous - which provides a natural "smoothing" to the extrusion flow.

Also, check your model to see how the facets look. It's a bit counterintuitive, but the better a printer is tuned, the more accurately it can reproduce the original model - INCLUDING its faults.

All the prints are done at the same temperature across all machines, (though I'm aware that thermistors have a degree of error so temps are probably not identical) but I will definitely start playing with the temperature on this. The gloss is the desired finish as these products are being sold. I'll start printing some "temperature cylinders" to see if the quality improves, as well as what level of gloss I can get.

As for your last sentence and this comment:

The next thing we need to do is get away from STL files and use CAD files like STEP. These file contain the perfect precision mathmatical models of the shape, not a bunch of straight-line triangles.

I only ever use STEP files exported from Fusion 360, both in Bambu Studio as well as Orca Slicer. I realized early on that STL files do not have the desired smoothing or accuracy when compared to STEP. I'm 99.9% confident that I only use STEP, but I will confirm this 100% with the next print.

So to summarize my homework sheet for the next few days:

• Check and adjust belt tension on the K1 printers
• Slow down the prints on the K1's and Trident (and also speed up on the K1 to test the theory of increasing speed for VFA's)
• Print a new XY joint for one of the K1 machines that can use toothed pulleys instead of the smooth pulleys (I'll try one machine to see if this improves)
• "Calibrate" temperatures on the Trident
• Confirm 100% if I'm using STEP files vs STL files (Though I'm 99.9% confident this is already the case; doesn't hurt to check!)

Thanks again for all the responses, I'll post back with my results for future readers.