Skew problem
Hello Forum!
I need to print some technical parts with quit tight tolerances. I know in the tolerances in the xy plane are good enough, but I never printed parts as high as I need now. So I purchased the Calilantern MK2 from Vector 3D to see how well my printer is calibrated. To my surprise, I found out that I have an enormous skew error in the ZY direction.
To answer some questions right of the bat:
- The belt tension is exactly as recommended by Prusa. I checked the belt tension prior printing the calibration part
- I took great care of the perpendicularity of the frame during assembly, using a large machinist square I borrowed from our QC-Department.
- The calibration part was printed as suggested, and I printed 3 of them, just to be sure. The measurements were taken with a calibrated Mitutoyo caliper, which I also borrowed from our QC-Department.
The results:
So, how do I solve this problem?
The Vector 3D-website proposes a piece of firmware code which compensates for the skew error, but I have no idea where to put this code. In Prusaslicer, under the tab "Printer" there is a field where additional code can be added. Should I put it there?
many thanks,
Jan
First the bad news: If you calibrate those measurements back into the original CAD you may get a near perfect calibration part - but that does NOT mean your production parts will be accurate. At high precision each part will have to be calibrated individually.
The main causes of dimensional error from a perfectly dialled in printer are:
- Thermal contraction, worst along long runs of extrusion, differential where different, longer or shorter, runs of extrusion follow different paths.
- Warping due to uneven cooling or differing amounts of (thermally contracting) extrusion in separate areas of the part. An accurate printer will place higher layers correctly but warping lower down may distort them.
- Rational approximation of the lines of extrusion when the original CAD exports to shape files which *can* (according to settings) be re-approximated in the slicer. Eg., a curve has to be broken into short straight-line segments of gcode; there are ways around this but all the defaults will be set this way.
- limitations of 'ooze' accuracy, the extruder follows a central gcoded path but the extrusion may be 0.45mm wide so oozing 0.225mm on either side.
- The tendency for extrusions to be dragged towards the inside of a curve as all the tensions including thermal contraction exert forces along the chord.
... and there are a bunch of other minor effects that might contribute.
For high accuracy you will have to print, measure and calibrate each part for each printer/filament combination. You should be able to get within one extrusion width in X&Y and one layer height in Z. The best place to put these calibrations is into calibration gain parameters in the original CAD, be careful to annotate them with the particular printer/filament combination to which they apply.
The figures you measured look quite good, I suspect warping influenced the outlier. You might find that protecting the printer from drafts will reduce the error.
There is a tendency for users to assume that just because parts can be designed/specified to high precision the printer will necessarily reproduce that precision. With careful calibration it can get close but it will never be perfect.
It's best to regard prints as blanks ready for finishing just like a casting or forging is regarded as a blank ready to be machined to working dimensions. Print slightly oversize and machine to fit.
Cheerio,
RE: Skew problem
I do not agree with what you're saying. The problem I'm having is completely independent from all the effects you summed up. To summarise it real short, is that the Printer (the mechanical structure) is misaligned and there is no way to compansate this in the slicer or firmware. I'm fully aware that every printer is not 100% square and has some degree of misalignment, but the real issue is not being able to do something about it other than compensate on CAD level, which I ultmately did.
I was able to get near perfect results with pre-skewed CAD-files (only changed ZX and ZY, so I'm sure I can improve even further; see score below), but these parts are of no use for anybody else. What frustrates me most, is that there is a super easy solution! You only need to add a shearing modifier in the slicer and you don't even need to change anything in the G-Code.
For me this reduces the printer to an expensive toy, and I cannot recommend Prusa printers anymore for users who want to make technical parts. The same goes for Bambu Labs, and probably other brands with proprietary firmware. I personally would have been better off with an Elegoo Neptune 4 by example.
I'm sorry to say, but your reply sounded more like an excuse than a cause. If Prusa wants to claim that they make to most capable Printers (they even advertise their consumer printers for professional use), they better live up to that.
I'm pissed!
I
compensate on CAD level, which I ultmately did.
That's the correct approach. Calibration into CAD.
these parts are of no use for anybody else.
Which is why you need to use calibration gain parameters, other users can generate their own set..
You only need to add a shearing modifier in the slicer and you don't even need to change anything in the G-Code.
...but remember, every part is different, there is no one-size-fits-all-parts solution.
You have to factor-in these considerations whether the machine is hobby priced or more expensive than your house: the place for this is in the CAD export.
Cheerio,