Header


Build Table Squaring


James Smith - Posted on 17 September 2010

I've been thinking about how to better solve to issue of the build table not being perfectly level on the Fab@Home. I have tried shimming, but for 3D printing large parts we really need to hold plus/minus 0.10mm across the build base and It is very hard to do this with manual shimming. Yesterday I decided to take a caliper to my machine and measured a number of points across my plate to create a contour map.

You can see that my plate isn't very level at all and for printing large parts I see a +0.45mm to -0.45mm sweep across the plate. That is over 3 printed layers think for me. Not good.

The solution is to fix this with software. This is how the motion control industry fixes these exact problems. I think this would make an important project for the CS team. Imagine a GUI similar to the above image image where we have a complete grid of points (lots of points, maybe it is an option how many points you have) with each point having a blank number input field. The user would measure all these points and input them into the GUI. FabStudio can interpolate between these points and automatically compensate its Z height as it moves around the build table.

Think about it, I don't think we are really going to ever get every Fab@Home's z table to be square. There is just too much variation during production, etc. All a user would need to do is move the build table up until it contacts the tool and note the position from FabInterpreter. Repeat this for all the points. Thus, every Fab@Home will have a perfectly squared build table no matter where or how it was made. I think this would be the most elegant solution to this issue and warrants some attention.

The brief discussion among the team suggested that we should implement some sort of quick calibration procedure that would prompt the user to contact the tip to the table at some fixed reference points to do some basic correction.

I agree that trying to level the table with washers and the like is basically pointless.

At Maker Faire, some passer by suggested projecting a laser grid onto the table and using machine vision to correct for tilt. This may be a viable solution, if we can get around practical issues of filtering out the grid.