slightly leaning frame
So far the printer has been absolutely amazing - couldn't be more pleased. I have noticed though, when printing tall objects with straight sides, I'm seeing a lean towards the back.
Sure enough, using a square I can see the front of the bed (when all the way forward) is below the bottom of the square by about 4mm (square against vertical smooth rods). Everything else on the printer is sweet - I am not skewed in any axis the printer can measure (I am "perpendicular"). And, of course, there is no way for the printer to measure the angle between z axis and the x/y plane. I also checked distances between vertical smooth rods and frame both sides, top and bottom - difference is almost impossible to measure.
So, the solution is fairly clear, i just need to adjust the angle between them. I've checked the PSU isn't pulling it out of alignment by loosening that, so i can discount that.
I'm guessing others have had the same thing - wondering what the common solutions are?
Idea 1 - shim under the smooth rods in the front y corners
Idea 2 - print something like this to pull the frame very slightly forward http://www.thingiverse.com/thing:1656945
Idea 3 - some kind of shim/modified washer/s where the M10 rods bolt to the frame
Am I missing something more obvious?
Thanks in advance 🙂
Re: slightly leaning frame (Z and Y not perpendicular)
Andrew's post describes my observation with my brand-new i3 MK2S (assembled yesterday).
In my case, the power supply actually was aggravating the problem, so I shimmed the gap between the plastic parts to be force-free.
I disagree with Andrew on one point - the printer DOES have a way to measure this skew with bed levelling and I can see it compensate for the skew when printing (Z-axis movement when printing in Y direction) but my inner German engineer keeps yelling at me to fix it.
I have measured all distances and the aluminium plate for planarity with a straightedge and have come to the conclusion that the skew must be somehow induced by the clamping of the plate with the M10 nuts on the long M10 Y-members.
Trying to correct by shimming the power supply bottom bracket edge doesn't work, as the required force is considerable and leaves the aluminium plate twisted due to the force only being applied on one side. It also deforms the power supply bracket.
So - I have not found an answer to the question on what to do in 20 minutes search - maybe someone can help? Or point me to the solution I missed in my search?
Greetings from Germany, Stephan
Re: slightly leaning frame
Very quick update:
I have found the culprit - as I had already suspected... The M10 nuts. They are tapped at an angle. From what I know about industrial production of screws and hardware, this is probably quite easy to happen for thin nuts such as these.
I have just wasted 4 hours over this and it is off to the hardware store now. I managed to get the Z-frame upright and in return it now pulls the y-stage into a trapezoid.
If you have similar issues, take off one or two nuts and look at both sides. In my case, there is one side of the nut with markings, which has a centric taphole (on all of the nuts, I dismantled my Y-stage). If you flip the nut over, you can easily see the taphole on the other side being off-center. When you thread the nut onto the rod, you can see it make a nice excentric move. The faces of the nut seem to be ground parallel and perpendicular to the wanted axis, but when you tighten them, they are forced into a skew following the axis of the tap.
For PRUSA (if you read this): this is a parts QC issue and probably also warrants a simple design change. If you want reference angles, nuts are not really the thing of choice (unless you specified these to be good for flanges). Consider another printed part or a precision cut section of aluminium tube to put on the threaded bar... That may make the whole Y-stage section construction a lot faster, too.
Stephan
P.S.: Of course Andres was right - the printer CANNOT detect this, so you should use good care to get the Z-Frame perpendicular to the build-plate.