Question Places to look when troubleshooting skew

[m00dawg]

Printer Model

LDO Trident 300

Extruder Type

Galileo

Cooling Type

Stealthburner

I have a strange situation it seems. I've got my 300 Trident printing quite well but have been wrestling with Skew. Using Vector3D's Califlower, my pre-correction skew is 0.23 degrees. I've printed multiple post correction califlowers and I get a skew of .01 to 0.02 degrees. Gantry seems properly racked. Belts seems to be good, they ring at the same pitch. I've checked to make sure screws are snug, including on the frame. I've checked full_steps_per_rotation for all the motors.

Trouble is, even though the califlowers show low skew, printing actual parts shows more and I can't figure out why. The center of the items (this is close to where I install screws) seems good, it's getting further out where there seems to be issues. I've done both where I've printed multiple parts at the same time and one by one, but still seems to have significant skew.

Starting to wonder if somehow my calipers have become inaccurate. I've had excellent results with the Califlower for my 2.4 and V0 for both skew and part shrinkage. To me it almost seems like my linear movements might not be linear. Or perhaps the shrinkage isn't being calculated right?

Thinking of trying a different skew model just to see what I end up with but given I've nailed skew with just 1-2 go 'rounds using Cali on my other printers, it feels like something with the printer I'm just not seeing (even though my post skew cali test prints seem quite accurate).

Any thoughts or suggestions?

 

[m00dawg]

A bit of an update as my Prusa XL arrived yesterday and I'm about halfway through it. I noticed three things:

• They suggest far more torque than I have tended to use for my Voron builds (even providing a crude torque gague)
• The belts on the XY axis seem much tighter than I tend to have mine
• They only use 2 leadscrews for the base (not really related to the issue at hand, but I thought that was interesting)

I know the frequency method of tightening belts is contentious but I tend to tighten mine to those and haven't had any issues on the 2.4 and V0. I did print the Klipper recommended skew tool on my Trident and it shows the same as the califlower post tuned (0.01 degree skew). Having printed parts on my 2.4 and Trident I actually have similar levels of skew. So I think with the Trident I just got unlucky in what I was printing as accentuating the skew.

So my plan is now to check the belt tension and torque on the frame screws on both my 2.4 and Trident. I ordered a 300mm calipers and will be making a 300mm skew calibration test model. The idea being I can try to reduce measurement error that way. The goal is to get closer to what my Prusa MK3's can do. I don't expect to be perfect since the MK3s have auto skew calibration but if I can get a little closer I'll be happy.

 

[NoGuru]

Oh man, this just across my feed

I am going to print one off now and see if my Skew is off.

 

[m00dawg]

Indeed I've printed that exact model. I can also scale it by say 2.75x - the problem there (and this is true of the Califlower too) is then the width of the frame gets to be large and ends up eating plastic. That's why I was thinking of making one sized at around 275mm or so where I can keep the width of the frame thin. I actually quite like the Califlower as it also assesses part shrink and chamfers the bottom a bit to help avoid measurement error due to elephant's foot. I'm gonna try to combine both of these models into one simple one that just looks at skew. I think using the normal sized Califlower is probably sufficient for part shrinkage (at least when I'm printing PLA since the shrinkage seems rather small).

Right now my skew seems to be ~0.5mm for the parts I'm tried to optimize around. On my Prusa's they print perfect but I have to do multiple print runs and the quality of the Voron parts is substantial (not that the MK3s are bad, just Voron's are better). .5mm isn't awful but for customer prints that's more than I'd like. Crossing fingers the larger calipers do the trick.

 

[m00dawg]

Here's what I've come up with to try on my Trident and 2.4. The hypotenuse is 296.99mm and the square is 210mm. Those aren't beautiful numbers but I think that'll suffice. I basically combined the model referenced in the Klipper docs (same as the above vid) with the Califlower in that the bottom is chamfered to avoid elephant's foot. I also added a fillet along the top just to help with any plastic stress which might mess with bed adhesion.

I'm gonna give this a try and report back. If it seems to actually work and I didn't mess something up I'll happily share it with folks (probably toss it up on Printables).

 

[m00dawg]

Update there, I spent part of the weekend printing the above model and re-skew-correcting both my Trident and 2.4. I got my skew down to 0.01 on both. As noted above I recall getting to that level with the Califlower on my Trident too - but still somehow getting some visible skew with some of the larger parts. After using the larger model, I _finally_ got basically a perfect print from it. I've only cranked out one so far so still need more testing but it rivaled the tolerance of my Prusa's. So at least for me I think I found a workable solution.

 

[NoGuru]

Glad you got dialed in. You should share your model. I printed the one from the video I linked but it was pretty small so I discarded it.

 

[m00dawg]

I put the model up on Printables though it's not the fanciest as it's not parametric. I included the FreeCAD for folks that wanna mess around with it though.