I have been thinking about auto-focus for my cheap Chinese laser (CCL), I will need to address the following if it’s going to work:
Batch or real-time processing,
Separate processing system or an adaptation of existing system
Scan entire surface and hold a “topographical” map that directs bed height by:
a, independently modifying the Z axis directly as the laser moves
b, processing the “g code” to add Z axis commands
A computational intensive solution that, I think, will add significant time to any job.
I do like the LIDAR as mentioned by @takitus but it would need to be modified, its specs state: Accuracy: +/- 2.5cm at distances greater than 1m
My favoured solution is a completely independent system that can be switched in as needed. “Point scan” the surface adjacent to the beam/material interface and modifying the Z axis directly as the laser moves. Take three measurements of the height of the material under the laser head, take the average and adjust the Z axis appropriately.
There are still a number of issues I need to get think about but I’ll get there.
Yes, I agree. A simpler approach might be to make a 5x5, 10x10, ?x?, array of depth measurements over the entire surface in order to approximate its shape, and use that model in the wide angle image correction algorithm. I’m getting out of my depth here, though.
I think you might be looking for something like this – it’s a time-of-flight-based IR distance sensor that’s supposedly good to 1mm at distances from 5-200mm. FWIW, my vote would be for b, producing a foo_adjusted.gcode file based on an interpolated topo map, because that leverages the existing firmware that already knows how to do smooth paths from one place to another. (And in most cases you won’t need a very detailed map, because if the material is that screwy you’re already in the weeds.)
Thank you @paulw three of those will do the trick just nicely.
I get what you’re saying about b. and if I were looking for a system to put into production I’d probably investigate that solution more. As it is, I’m going for the QDOS approach, easy to implement, puts no additional processing strain on existing systems and, once completed, just about anybody will be able to modify their CCLs to auto-focus.
I read this one a couple days after, but now feels like a decent time to say.
While I think this is some great high end thinking here, is it possible we are thinking about this too hard?
Shouldn’t we have the expectation that proof grade would not have warping unless stored improperly, but by the very measure of it being proof grade, means it would be free of defect and optimal for lasering?
I don’t expect to worry about any proof grade materials… Or is this way off base?
Warping post-production isn’t a defect as much as it is a result of the storage conditions. Once it’s left GF’s shipping department it’s at the mercy of the world.
Plastics & leather will be relatively unaffected. Thicker woods somewhat affected by environmental conditions and storage. Thin wood potentially affected a lot.
What you should expect is that a 1/2" thick piece of maple is spot on 1/2" over its entire length & width, no loose or maybe not any, knots, smoothly finished and ready to use without additional prep.