RE: Skewed XY-Plane (Z Rods, Heatbed)
I would suggest to manually level the bed relative to the nozzle. That way you can be sure that any automatic bed leveling effects can be ruled out. Only then can you say for certain that any height differences in the corners are a direct result of the different spindle pitches.
But you don't really need to print. It's even easier to measure the spindles pitch over a 100mm distance with calipers. Two should measure the same and one should deviate.
RE: Skewed XY-Plane (Z Rods, Heatbed)
I took out all my motors/lead screws and measured 18 revolutions (see white thread, it's a four start screw), the maximum I could do with my calipers. I pressed the internal jaws into the root and out against the flanks carefully as deep and far as I could, while everything was aligned on a rudimentary "jig" I set up to keep everything straight. (I guess this is the most accurate way I can take the measurements.)
I was expecting two identical measurements plus an odd one out, as you guys write above.
- Spindle on the left side of my Core One (z left) 145.8mm
- Spindle on the back of my Core One (z right) 145.1mm
- Spindle on the right side of my Core One (z left) 144.7mm
I repeated the measurements, but got the same result for each spindle. Also, when I locked my calipers after measuring the spindle on the left (see picture), I couldn't fit them into the threads of the spindle in the middle (=back). When I locked them to the measurement of the spindle in the middle (=back), I could not fit them into the threads of the spindle on the right.
My Core One is a full kit btw, batch 6.
RE:
Wow, of all of the things that could be manufactured wrong, I'd have never thought to consider the pitch of the lead screws. I would expect them to be manufactured using a CNC process that results in reliably identical output. This is a really odd problem, and one Prusa should take seriously. Before doing my conversion, in addition to setting the x gantry brackets against a machinist square, I'll now check the lead screws on the Z motors as well.
I do wonder how (un)common this is. By now there are lots of Core-Ones out there, and presumably the vast majority are clipping along perfectly. Thankfully this issue is easy to check, and Prusa should be quick to replace any motors since the discrepancy is clear.
-J
RE:
I do wonder how (un)common this is. By now there are lots of Core-Ones out there, and presumably the vast majority are clipping along perfectly. Thankfully this issue is easy to check, and Prusa should be quick to replace any motors since the discrepancy is clear.
It might be more common than we think. Since the deviation over the whole Z travel range is still within the range which the software-compensated bed leveling can compensate for, the problem is not obvious. And it will only become notable (as a skewed print) in tall prints.
We had several early kit builders ask on the forum about a deviation of a couple of mm in their bed leveling, and have spent some time here chasing potential skewed frames etc. Maybe this lead-screw problem was behind those cases?
Personally, I can't figure out why I did not notice the issue earlier in my printer. I did the "Z calibration, drive the bed up, measure distances between bed and nozzle with a feeler gauge" routine back then, aiming to provide comparison data for what tolerances to expect. Looked great, with all distances within a couple of 1/10 mm. Maybe my CoreXY frame or X gantry is not level and just compensates for the height difference caused by the lead screw pitch?? (But it would do that only in the topmost position, of course, so this is not a solution to the problem.) I'll need to double-check!
As mentioned earlier -- maybe it's just the kit builds which are affected, because they combine Mk4s motor sets from older stock with a third Z motor from a new batch? For a factory build it seems plausible that all three Z motors materials would come straight from the same shelf. In which case the factory builds (and maybe the more recent kit builds) might have all three Z axes running too fast by 0.5%. At least that could be easily compensated in software.
RE: Skewed XY-Plane (Z Rods, Heatbed)
I don't think it's a kit issue; mine seems to be consistent - all the way down to testing with a feeler gauge across the bed (at each of the line crossings on the shipped plate) and no observable deviation. I am using a 1.4mm spacer front left and am also not observing z axis movement as the head moves in x/y. Hopefully it's a bad batch they can identify and address.
RE:
I don't think it's a kit issue; mine seems to be consistent - all the way down to testing with a feeler gauge across the bed (at each of the line crossings on the shipped plate) and no observable deviation. I am using a 1.4mm spacer front left and am also not observing z axis movement as the head moves in x/y. Hopefully it's a bad batch they can identify and address.
Have you measured and compared the distances between the left/right heatbed struts and either the CoreXY plane above, or the base plate below, for the two different heabed positions -- all the way up and all the way down? Just having things corrected at one specific Z height (in your case, all the way up) does not tell the full story. That's what I observed too after building my kit, and I totally overlooked the screw pitch issue.
Your 1.4 mm spacer on the front left may just be compensating for the fact that the front right lead screw has a faster pitch. You essentially give the slower (left) Z drive a head start, making sure that they both arrive at the top in sync -- but on their way to get there, at the lower heatbed positions, are they in sync as well??
Edit: Having said that, it may well be the case that your kit is fine. I did not mean to say they all have the problem, but that only kits might have the problem. When did you receive yours? Mine was early in batch 1.
RE:
Print a large triangle, with the corners placed where the motors are. Just a single-pass wall in vase mode is hopefully going to work, to limit the printing time and filament use. I don't think it is critical how the Z axes are referenced/homed/calibrated ahead of the print -- we should see different heights at the corners, in proportion to the different screw pitch, in any case.
I will probably only get to this later in the week. If you find time to prepare a test model, it would be great if you could share it here!
Attached, printing right now with 0.8 mm width for 1st layer + external perimeter ; 0.25 mm layer height (also 1st layer) ; vase mode ; 8 mm Brim inside+outside ; Chamber nominal temp set to 31°C ; vent open, chamber fans basically off (due to the nominal temp).
Had to set the "Slow down if layer time below: 15 s" to 8 s to get a decent speed. That's about 75 mm/s and 13.750 mm³/s.
Gonna report back in about 1 hour!
I would suggest to manually level the bed relative to the nozzle. That way you can be sure that any automatic bed leveling effects can be ruled out. Only then can you say for certain that any height differences in the corners are a direct result of the different spindle pitches.
But you don't really need to print. It's even easier to measure the spindles pitch over a 100mm distance with calipers. Two should measure the same and one should deviate.
I did level it manually before the print with a really flat card (0.72-0.73 mm, used the same spot for each level-point), leadscrews are moving a little bit on the straight lines, but are pretty much in the same position at the 3 corners.
I'll try to measure the pitch, when the print is done.
RE: Skewed XY-Plane (Z Rods, Heatbed)
Well, you made me second-guess myself and go rethink exactly what/how I was measuring. And you're right. That front left is moving the bed slightly faster such that perfect bed leveling at the top is off by the time it reaches the bottom.
Have you measured and compared the distances between the left/right heatbed struts and either the CoreXY plane above, or the base plate below, for the two different heabed positions -- all the way up and all the way down? Just having things corrected at one specific Z height (in your case, all the way up) does not tell the full story. That's what I observed too after building my kit, and I totally overlooked the screw pitch issue.
Your 1.4 mm spacer on the front left may just be compensating for the fact that the front right lead screw has a faster pitch. You essentially give the slower (left) Z drive a head start, making sure that they both arrive at the top in sync -- but on their way to get there, at the lower heatbed positions, are they in sync as well??
Edit: Having said that, it may well be the case that your kit is fine. I did not mean to say they all have the problem, but that only kits might have the problem. When did you receive yours? Mine was early in batch 1.
RE: Skewed XY-Plane (Z Rods, Heatbed)
Wow, so it's in fact very likely that the frame is pretty square on its own and the problems with bed leveling are due to this issue. I would say that since the difference is relatively small and the automatic bed leveling fixes the issue pretty well, there are probably a lot of people with this problem that are unaware of it.
RE:
I don't know why I thought printing a big triangle with rounded corners would work.. Somehow the brim didn't really clamp it down, the walls warped in all directions and I stopped the print after 30 mm in height.
I then printed 3x 100 mm cylinders, 40 mm diameter, single wall, "inside brim" with a chamfered edge right in the STL (attached, add 3 instances and place them accordingly).
Results, after clamping them between my little machinist square and steel ruler to get more accurate with my caliper:
Front Left Cyl. = 100,16 mm
Front Right Cyl. = 99,98 mm
Rear Cylinder = 100,07 mm
I then measured the pitch of the leadscrews as best as I could. For 5 convolutions, I roughly get:
Front Left = 11,3 mm
Front Right = 11,1 mm
Rear Leadscrew = 11,2 mm
The measurements make sense, since when I put two cylinders on my machinist square and put the steel ruler with its side edge on top:
- Left + Right = Right Cyl. is easy to move
- Left + Rear = Rear Cyl. can be moved, but has some friction
- Rear + Right = Right Cyl. can be moved, but has some friction
The Left Cyl. is always clamped down by the steel ruler.
"Sadly", it doesn't seem to be the case, that the two leadscrews in the set are equal and the third one is the outlier.
It rather looks like the leadscrews have quite big tolerances for 3D printing.
0.18/100 mm difference doesn't sound extremely bad, but if I would print something quite wide and with 0.07 mm layer height, that's a difference of 2.5 layers.
RE: Skewed XY-Plane (Z Rods, Heatbed)
Great find. I'm really surprised to see this much deviation in the lead screws. I've seen manufacturing tolerances on rolled lead screws as large as .25mm/300mm. Most seem to be measuring far more deviation than this on their printer.
I don't personally think even .25mm/300mm is good enough. and hopefully what Prusa are specifying is better than this. I'd like to see more like .1mm/300mm. This is mainly because of the fact that 3 lead screws are used and they need to move in sync.
Let me rephrase, the .25/300 deviation would be fine as long as all 3 lead screws were the same.
They do offer ground lead screws for ultimate precision, but those are usually used with ball screws.
RE: Skewed XY-Plane (Z Rods, Heatbed)
This is the last straw for me. I cancelled my conversion kit order today.
MK4S/MMU3
RE: Skewed XY-Plane (Z Rods, Heatbed)
This raises the question: calibration should be able to calculate the differences and compensate as long as you haven't gone and applied a spacer at the bottom (as I and so many others have done) to force for perfect alignment at the top. Does it work this way? If so it should lead to much better tolerance in prints than the screws have. Will have to test later myself.
RE: Skewed XY-Plane (Z Rods, Heatbed)
This raises the question: calibration should be able to calculate the differences and compensate as long as you haven't gone and applied a spacer at the bottom (as I and so many others have done) to force for perfect alignment at the top. Does it work this way? If so it should lead to much better tolerance in prints than the screws have. Will have to test later myself.
That won't fix anything. All 3 motors are run from the same driver. This means they all turn the exact same. Therefore your print may be level at the top, but any deviation in the lead screw pitch will be on your parts. The taller the part the greater the deviation will be.
If all 3 lead screws could be run independently then you could compensate with software.
RE: Skewed XY-Plane (Z Rods, Heatbed)
It only matters that the bed is in the right position at the print head - it could still compensate for uneven screw pitch I think. Individual drivers would mean it could keep the bed level at any height - while ideal, that's not strictly required to adjust for skew over z-axis travel.
RE: Skewed XY-Plane (Z Rods, Heatbed)
It only matters that the bed is in the right position at the print head - it could still compensate for uneven screw pitch I think. Individual drivers would mean it could keep the bed level at any height - while ideal, that's not strictly required to adjust for skew over z-axis travel.
That's not how it works.
The bed isn't skewed statically. You can't fix it by shimming the bottom endpoints as the different pitch will skew the bed again as it moves further away from the "flat" position. You also can't fix it by compensating with software. All it will do is make sure the first layer is right.
With the bed uneven at the top (no shims), the actual print will start slightly skewed and slowly "straighten" as the bed moves down.
With the bed endpoints shimmed to be even at the top (and uneven at the bottom), it won't compensate for the constantly changing skew as the bed moves down. The print will start straight and slowly skew as the bed moves down.
This is an issue when you print tall parts that need to be dimentionally accurate, especially for flush gluing. The only way to compensate for it software-wise is to drive each motor individually to adjust skew real-time for every Z height.
With static skew it would be easier, shim it and you're flat. But with dynamic it changes as the print goes on and the printer has no way of knowing nor adjusting for it.
RE:
It only matters that the bed is in the right position at the print head - it could still compensate for uneven screw pitch I think. Individual drivers would mean it could keep the bed level at any height - while ideal, that's not strictly required to adjust for skew over z-axis travel.
I'd be interested to see how this could be done.
In my mind if the screw pitch is not the same on all 3 screws, then that means that each corner lowers a different amount each layer, resulting in the angle of the bed changing throughout the print. This angle will show up on your parts. The test parts printed by others at each corner of the bed, each showing a different height proves this hypothesis I think.
There is nothing you can do to fix it without individual drivers, or lead screws with the same pitch.
*Edit* Ha ha, Shushuda beat me to it.
RE: Skewed XY-Plane (Z Rods, Heatbed)
If you watch a print it adjusts z as x/y changes (assuming no shims). My expectation is that a tall print would end up with no such adjustments as it approached z-max.
RE:
I think it could be fixed by software.
Here is an idea of the procedure:
- print a thin tower, with a solid top. for example 250mm of height and 20mm of diameter.
- measure the actual length of the tower print: T
- perform bed leveling at Z=0
- use the tower to perform bed leveling at Z=T. The nozzle will bump on the tower at different position (this needs to be assisted by a human who moves the tower)
- Set the value T for the correction software and compute interpolation between both matrices
This assumes that the pitch remains constant for a given lead screw across its length.
Should work, no ?
RE: Skewed XY-Plane (Z Rods, Heatbed)
If you watch a print it adjusts z as x/y changes (assuming no shims). My expectation is that a tall print would end up with no such adjustments as it approached z-max.
That's not how bed leveling works. It measures the bed at the top and keeps the same adjustment for the entirety of the print. It doesn't know the bed is skewing or getting straighter as it moves down. It doesn't probe in that lower position, the extruder doesn't move down.