Voron Test cube dimensional accuracy?

[Beez1965]

Newbie here, my first printer is up and running V2.4rC, LDO kit 350 mm.

Lots of googling has not yielded useful results.  Perhaps I'm not good with generating search terminology, sorry.

Here's the issue, I print the Voron test cube, many of them.  The X and Y dimensional accuracy are pretty spectacular, IMO, they come out about ten microns oversize in both axes.  Pretty repeatably.  The Z dimension though is 250 to 300 microns over.  I'm being pretty careful not to measure over any warts and I've tried lightly sanding the top and bottom on 400 and 600 grit to remove same.  1% oversize seems excessive to me.  Can this be improved?  I've gone thorough tons of tuning, the stuff in the Voron build documentation, calibrations built into SuperSlicer, and the Ellis pages.

Any advice would be GREATLY appreciated.

Thanks!

Chris

[claudermilk]

Which printer model? The Z axis operates differently on each one--that's really the primary difference between a V2.4 and a Trident.

[Beez1965]

V2.4r2, LDO kit 350 mm

[Beez1965]

Interestingly, it seems like the Z dimension of the early cubes I was creating was pretty spot on, but I've printed nine of them over the months I've been building this machine and the Z has gotten worse.  I should have put a date code on the cubes and tracked the changes...

[claudermilk]

All I can think of is something with the Z belts? I am not personally familiar with the V2.4. All my printers have a lead screw Z axis (Trident, V0, Prusa Mini), so once the setup is done, it's done.

[VDZ]

Make sure you are slicing your cube with a layer hight that divides evenly into the hight you are trying to print. 

[Beez1965]

16 hours ago, VDZ said:

Make sure you are slicing your cube with a layer hight that divides evenly into the hight you are trying to print. 

Oh, now THAT'S interesting...

I'll look...  Yeah, 0.2, although the variable layer height may be interfering?

I'll try setting the limits for min and max on that to .2 as well.  Using SuperSlicer 2.4

[Beez1965]

So, after changing the layer height max and min settings, both to .2, the result is the same, although the printer, via the lcd, claimed the top layer was at 30.4, which I thought was interesting.  Usually it says 30.0 as expected.  

[Beez1965]

Maybe I should try a different slicer, or an update to SS?

[YaaJ]

On 12/18/2023 at 12:29 AM, VDZ said:

 

You should upload the gcode generated by the slicer. What is the last Z in the G-Code ?
(had a hard time with such an issue, until I had a look to what was sent to the printer...)

[Beez1965]

Oh, interesting, I'm not savvy with gcode yet, although I used to be a C programmer, I've been a commercial pilot for a couple decades since so a bit rusty.  Let's see...

 

 

Voron_Design_Cube_v7.gcode

[Beez1965]

So, skimming through the file, it seems the top layer is printed at 30.0.

I do see some jumps up to 30.4, but always preceeded by G1 E-.85 or so, which I assume is retraction, then an XY reposition, and deretraction.

As far as my naive eyes can tell, nothing should be printing above 30.0mm.

I have run a dial indicator on all axes, and it indicates proper motion, although it's only 10um per division and 5.5mm throw, it's the best I got on hand atm.   

[YaaJ]

Yes last "printing" G1 Z = 30 mm, and you're using z hop (doesn't change the overall height)

Could be intesting to print something taller, and see if it is proportional ; for example a 60mm or 120mm cylinder. Is the error proportional ? Cyclic ?

At this point, my 2.4 doesn't print. But on my bed slinger, belt driven on X and Y, don't have the "default" values. I had to adjust rotation_distance on both ; (I use a 50mm dial indicator, my poor man's lathe DRO, and it shows cyclic variations : more or less +- 0.1mm) ; also, steppers, pulleys and belts are not servo driven precision grinded ballscrews ! +-0.1mm is in the ballpark. On my side, I don't like the paper trick for Z offset adjustments. It's a starting point : I adjust Z offset using a micrometer to measure the brim or skirt thickness. If the paper is considered zero (origin), the part will be 0.1mm too tall. Add +-0.1mm for the machine, some overextrusion, roughnress and ironing on the top layer, a textured bed, the pulleys eccentricities, and we easily get 0.2 mm. Could also vary depending on the bed and gantry expansion while printing. For example the "taco" bed if printing in the middle.

The Voron 0.2 (Z screw) just gave a 35.1 part for 35.0 expected... I don't care. If a more precise height is required, I play with rotation_distance or compensate in CAD (do it often...), or adjust scaling in the Slicer. A printers repeatability usually being amazingly good, the second iteration is the final part. I do it all the time.

Aren't you chasing unicorns ?

 

Edited December 20, 2023 by YaaJ

[Beez1965]

 

Aren't you chasing unicorns ? https://cdn.jsdelivr.net/gh/twitter/[email protected]/assets/72x72/1f609.png

 

Maybe I am, but I do have a machinist background as well, so, I'll continue a bit longer.  Also, since I'm getting about 20 micron precision in X and Y, and I did three times out of 12, get that same accuracy from Z...

 

That first layer and paper calibration might be an issue, the paper I'm using is only about 70um though and i'm off around 250 to 300.  You have paper .1mm thick?  Or that was just an example?  What's the goal of that paper test anyway, to get the nozzle microscopically close to the bed without hitting?

I'll print a taller thing and check.  ATM I'm thinking it's either something I did in calibration (Z belt tension), or something I changed in the slicer.  Learning so much so fast there are just tons of variables.  Def triggering my OCADD at the moment. 

 

Thanks for helping!

 

[YaaJ]

Yes, my paper is 0.1. I always found the X and Y dimensions more precise than Z (for the reasons I listed, and am probably missing many others !)
Maybe you could print a few 1 or 2 walls cylinders ? Would be interesting to measure 20, 40, 60, 80, 100mm for example.

If you didn't already know (unlikely...), there's this problem with the gantry : https://github.com/VoronDesign/VoronUsers/tree/master/printer_mods/whoppingpochard/extrusion_backers

Far from being negligible. 0.2 is in the same order of magnitude. The 2.4 tends to print taller in the center after heat soaking. Did you install backers ?

[Beez1965]

I don't have backers, and thanks for throwing open that rabbit hole for me to peer into! 

Let me print some cylinders.

[Beez1965]

While that's cooking, I came up with a possible theory why some of the early test cubes are more accurate in Z.  No enclosure installed yet.  

Edited December 21, 2023 by Beez1965

[VDZ]

Paper test is just to get you close, so you don't hit the bed. After paper test and the printer is running; print a single layer part @ 0.2mm and measure the hight of the print. Make baby step adjustments until your printed 0.2 mm part is .... well 0.2 mm. Then save z offset to your config. I will often print a three loop single layer skirt and measure that with the callipers to monitor my first layer hight accuracy over time. 

[Beez1965]

So the cylinders all have a fixed offset, oversize in Z, about 150microns in this case.  Maybe I need to print objects in the corners and center and check if it's positional.  That would really support the gantry bowing theory.

[Beez1965]

38 minutes ago, VDZ said:

Paper test is just to get you close, so you don't hit the bed. After paper test and the printer is running; print a single layer part @ 0.2mm and measure the hight of the print. Make baby step adjustments until your printed 0.2 mm part is .... well 0.2 mm. Then save z offset to your config. I will often print a three loop single layer skirt and measure that with the callipers to monitor my first layer hight accuracy over time. 

That's genius.  I print a skirt anyway to get the nozzle primed...

So my skirt is running around .33 so that's nearly the whole error right there.  Sorry for my ignorance but where do I input this offset?  And how would this not hit the print surface?  I used approx .1mm paper, so theoretically it's that high off the bed.  If it's trying to print .2 and getting .33 I need to go down .13 which would be .03 below the surface, right?

Edited December 21, 2023 by Beez1965

[dshess]

If you want to pursue the paper-style approach, get some feeler gauges.  Printer paper isn't calibrated and it is squishy, so while it is fine for roughing out the tramming of the bed, it's not really great for actually setting z offset.

That said ... getting your first layer right is a combination of tramming, measuring, and adhering.  You can't always get it perfect, and sometimes getting it perfect in one area makes your printer less functional.  For the most part, very few things require a 30mm tall model to be PRECISELY 30mm tall, and for the most part these printers aren't geared towards that level of precision.  So don't put too much time into fixing this "problem" up front, instead stay aware of what you're printing, and worry about it if and when you print something that really needs this level of precision.  For most people, good repeatability is much more useful than high precision.