Personally, I'm debating whether to go the Arduino + webcam route as BQ has done, or the Raspberry Pi + cam route as Atlas/FreeLSS. But since the FreeLSS mechanical design isn't yet released, I'm going to start by building up the Ciclop platform, and then decide. . .
The first one looks good to me. However the parts for this thing are vague. I wonder if he can document these a bit better. I would be willing to buy the electronics and bearings and build the rest myself. However I don't see much in there. I saved the link and will be taking a longer look when I get home.
I've started printing, just a couple of the smaller parts. My bearing is going to take a little while to get here so I won't be assembling right away, but I can at least get some threaded rod and try fitting the major parts together. I think the exploded diagrams will make it fairly easy, assuming everything prints like it is supposed to. . .
After a further look onto the design. I think I will borrow some of the electronics and build the rest of it out of imagination and recycled parts. I have an old VCR head kicking about that would be ideal for the project.
I've printed the parts for the turntable and assembled it, and I'm impressed; the unit is solid and stable. With the big bearing it really feels like a machine. My only feedback so far is that the hand-drawn supports that are included in the STLs don't really work, at least with my printer. The large arches and central hollow in the turntable base were very rough and needed a good bit of cleanup. The motor mount holes didn't print at all, I had to estimate where they should be based on the top fill pattern and drill them out. It will always be a difficult print but I think that allowing the slicer to generate support is going to be a better bet.
I love what you did there. Certainly looks like something I can build. I just bought that bearing. About 12.26USD shipped. God knows its going to take like 3 weeks to get to me though. Its ok, as I am planning on heading down your road. Once I get the bearing I am planning on redesigning the whole contraption and using a pi2 with camera hacked for macro. I am planning on putting this ontop of my printer integrated into it. Hopefully it works out
I like all in one machines. I have to figure out the motor driver circuit, I am tempted to get an adafruit simple stepper driver and wiring it in. Although I am a bit fearful of doing so as I plan on using a nema 17 to drive the table and it may require more power than the driver can supply. I wonder if 200 steps per rotation are enough for resolution.
Edited 1 time(s). Last edit at 02/06/2015 02:19PM by jaguarking11. My Personal Blog. Build blog.
I can't say for sure about the steps/rotation, but the FreeLSS configuration defaults to 6400, which is what you get with a 0.9 degree/step motor and 1/16 microstepping. I only have 1.8 deg/step motors handy, so that's what I'm going to start with. The scan defaults to 800 steps in the "detail" setting, which should work fine with either kind of motor, but not without some kind of microstepping. Given that the firmware already knows how to talk to a standard Pololu-style A4988 driver, why not use one of those?
Looks like I am going to wait until my bearing gets here. I like the bearing construction. I am planning on designing the mechanical parts from scratch and testing them out as I build them. The only thing borrowed will be the stl. As for lasers, I found harbor freight has some 5v lasers they sell that are powered by button cells. I bought 1 and will be buying another one soon. Very cheap as I recall they were like 3USD and include 3 small neodinum magnets that I needed for my printer anyway.
As for the stepper driver, I took a second look at the circuit board. I guess I have a bunch of parts to buy. I plan on enclosing the PI and camera at the back of the printer with an adjustable heigh on guide rails. Hopefully it works out. But until I get my bearing I wont do any more buying as the PI 2 I plan on purchasing is on back order. I may even try to do a 3d printed breadboard and just solder the wires properly to each component.
The latest is that the Piclop hybrid scanner is working pretty well, and I'm getting used to the software and figuring out the quirks of calibration and setup. I've added a couple more update posts at [bill-owens.blogspot.com], and uploaded my versions of the laser holders, camera tower and camera mount to Thingiverse (links in the blog posts). The author of FreeLSS, Uriah Liggett, is planning to release a new version soon with features to make calibration easier, and that will be very helpful. There's at least one other person doing his own version of the parts as well, and I suspect his will be much more elegant than mine once he gets them done. I may be too lazy to reprint though
Not much to see, but here's a picture. The DIP is the ULN2003A Darlington array that drives the lasers; the little power supply hanging off the edge is producing 3.3 volts to feed it. There are separate 5 and 12 volt buses on either side of the board, and I'm using a surplus Cisco power brick that can do 3 amps for each, more than enough. There's a Pololu-compatible A4988 module in the middle of the tangle of wires, and an Adafruit T-Cobbler adapter bringing the Pi GPIO lines to the breadboard. It's a Pi 2, though I had previously run the software on an original B and it seemed fine. The camera ribbon cable has been replaced with a longer one (200 mm/8 inches) from Adafruit.
My plan is to make a permanent electronics board using an Adafruit HAT protoboard, but I haven't gotten around to that yet, I've been too busy playing with settings and calibration. For now, the breadboard works fine, it just takes up too much space on my already-crowded desk!
Edited 1 time(s). Last edit at 02/20/2015 07:21PM by owens.
You only need the regular one; the NOIR is for night vision applications and specialized stuff like plant monitoring. I recommend going through the step of removing the glue spots from the edge of the lens to make it focusable; you can find the instructions on the Raspberry Pi foundation website (basically just scrape the glue away with a fine tool like a dental pick until the lens will turn with a gentle twist).
I picked up the Pi2. I absolutly love it to bits. I picked up the camera as well. I have imaged the pi, updated it and installed FREELSS on there. It launching fine. I created a rc.d script that will allow me to start it from any directory, and eventually it will start it with the pi on boot up.
I just ordered a pack of caps and the ULN2003A for now. I just ordered the PSU for this as well. 12v 2A and 5V 2A. Its meant for external hard drives. Although I expect the PSU's to take 2-3 weeks for delivery. I am not in a hurry. I have other means to power this thing.
One thing to note. The original STL files are not prime time ready for printing. The motor holder would not print right or slice right. I used your source files, removed your support and exported out the STL from freecad, then I ran it through netfabb for clean up and it printed out beautifully with slicer generated support. I will be printing out the rest of the parts this weekend.
Thank you for the design. I love the aesthetics of it.
I really like this and considering making one very similar. I have a comment and a question, though:
In the sample scan the lasers are not illuminating the top of the head and under the chin very well. This is demonstrated through the scan being sparse at the crown, for example.
I was wondering.... What would happen if you ran the rotation and scan process twice - once with the lasers and camera lower than where it is now to reach under the chin, then a second full rotation with the camera and lasers raised to illuminate the top of the head? Could the resulting scans be stitched together to give a more complete scan? I guess this could be augmented by doing 3 scans or more as well if necessary.
Mounting the camera & laser assembly on a hinged bracket that is raised through a stepper motor would be one way to achieve this, I would think, but there isn't much point in trying it if the scans can't be stitched together properly.
The scans can be stiched together quite nicely using meshlab or similar. You don't really need a hinged camera, having the camera at a set location works nicely with this scanner. You can rotate the object and do multiple scans of the object to achieve similar results. From what I can gather just about every scanner in the market including the ones that have the camera mounted on a stepper taking vertical slices need multiple scans to get a good model. The thing that is impressing me most is that once tweaking was done this scanner works admirably well and captures allot of info.... especially once you get the 5mp still mode working nicely. The learning curve of the software for post processing of scans is allot higher for me than building and tweaking this design.