Yeah, I just need to find the time to get to it. So many projects, so little time.
1 Like
You might want to take a look at this. Working robotics with high school students, we’re always on the lookout for inexpensive projects for the off-season. The author used a laser cutter, but I don’t see any reason this couldn’t be adapted to the Maslow.
3 Likes
Here’s the prototype for the camera mount. I have to sand/cut off the bottom some to clean it up, but it seems to fit the camera well. I learned that I needed to compensate for PLA shrinkage. My first “real” one (I had a couple goofs earlier) came out a little small. 0.467 inches vs. 0.5 inches so I scaled my model up by 7% and its come out right on the money. I googled PLA shrinkage and someone mentioned it shrank about 7% so I can verify… 7% is about right.
Now, I need to recalibrate my Maslow after disassembling it and turning it around in the shed. I’m not sure why I built it on the wall where I had all my electrical outlets for power tools and workbenches… wasn’t thinking clearly.
5 Likes
Well done!
Thank you
The mount was causing issues with the cable, so I’ve redesigned it. It’s now two pieces. I’m going to use a piece of 1/2-inch steel as the insert into the router and then place it into the “hole” in the new mount.
4 Likes
Hi in case it helps, I just saw this on another thread: Acrylic on Maslow
A camera to attach to a CNC to help locate it: https://ezletter.en.made-in-china.com/product/xsqmWgaHvpku/China-Ezletter-CNC-Engraving-Machine-Spare-Parts-Eye-Cut-Camera-in-Customized-Plastic-Cutting.html or here: http://www.ezletter.in/eye-cut.html
Which led me to this site on camera based zero’ing: http://beatty-robotics.com/zeroing-a-cnc/
1 Like
The problem with those cameras with our setup is that our sleds rotate. So if you offset a camera from the center of the router bit, you have to also know how much your sled is tilted to a very high accuracy to account for it. I didn’t find a cheap and simple solution for that. So I opted to put the camera precisely (as precisely as I can) dead center where the router bit goes. I have my 1/2-inch steel tube and just need the free time to get out to the shed to cut it into 1 or 1&1/2-inch chunks…It fits really nice and secure into the opening I printed.
4 Likes
As an update, the camera modules I got just couldn’t focus close enough. I was worried this might be the case. So, I resorted to using my USB endoscope with a turned block that @marm sent me and moved forward with the software. I’ll post a video of it in action once I figure out one graphical bug (some reason, an image is updated to all black rather than the actual image). My machine may not be well calibrated but I did a test to see what I could measure.
I have a grid pattern of five 0.5-inch squares, spaced 2-inches apart in a dice pattern (picture a 5 on a die) with the center square positioned fairly close to the center (0,0). I did a series of measurements to calculate the distance of the center of each square from the center of the camera image (the middle square is used to determine the offset of the pattern since it wasn’t dead center at 0,0). After measuring the four corner squares, I remeasure the center square to see how precise the machine/measurements are. This is a chart (in mm) of the measurements. I ran it four times. The black diamond is the target locations (+/- 50.8 mm) and the colored dots are the data points. The right side seems more “precise” than the left side.
I’m going to reprint my pattern tomorrow (the inkjet I used has a few registration issues that causes some noise on a couple edges) and then I’m going to recal the machine and see if I get any improvement.
5 Likes
Glad that the block is working out for you.
The results that you get will be interesting to see what kind of accuracy and precision you are able to come up with.
2 Likes
The block is a little tight in the housing (maybe I need to grease it up). I wrapped tape around the endoscope to build up the thickness. The tape’s not perfect, but It’s good enough at the moment.
1 Like
I think that I put paste wax on my block and it seemed to slide better, if not you can always sand it down some more.
1 Like
I managed to recalibrate the machine and run a few more tests. Accuracy has significantly improved. This is just two runs. Now I need to make a bigger pattern…
5 Likes
I recalibrated (yet again) and used a larger target printed on a 11x17. This time, its a 13-square pattern (but it’s black, not red):
The spacing is still 2-inches from each other so the outer squares are four inches horizontal/vertically separated from Paul Lynde (let’s see who can figure out that reference).
The results are interesting in that the measurements along a line from top right (TR) to bottom left (BL) seem shifted a little clockwise but the measurements along the line from top left (TL) to bottom right (BR) are really close to on the mark:
The maximum error I measured was the bottom left at ~2mm. Of course, the question is whether or not the target is wrong, the machine is wrong, or both are wrong. I wouldn’t have been surprised to see all points shifted CW or CCW as the target might not have been perfectly level/plumb, but to see the shift only from TR-BL and not TL-BR is a little confusing.
4 Likes
Paul Lynde
Hollywood Squares. He was almost always surrounded by a bunch of other actors. There. I think that I left that vague enough to not give it away completely. 
2 Likes
I think another small thing that might play into the fine tuning is the gyroscopic effect of have a spinning motor trying to be moved fairly quickly by something that’s flexible.
I think more cuts all over is a great idea to help calibrate.
The gyroscopic effect actually helps i think, it should keep the router in the same plane, moving it around shouldn’t be a problem, but flipping it over should get some resistance.
1 Like
A movement should be induced 90 degrees further into the direction of rotation when moving based on what I know (the chains and ring will do their best but they are flexible to some extent).
How much that movement is I don’t know. Maybe not enough to worry about with the sled as big and heavy as it is and the motor being relatively small and everything free rotate within the ring. But for fast small jerky movements carving letters and fine details?
If I get off my duff I’ll try to hold the sled in free air running without a bit and see if it really does much. I’m probably over/under thinking this =)