17. Applications and Implications¶
My Fab Academy Masterpiece - Coordinate Measuring Machine (CMM)¶
My final project will be a coordinate measuring machine. Or at least the hardware components that will measure the real world.
Who’s done what beforehand?¶
There are many types of CMMs like robotic arms with special probes at the end but the most common ones are probably total stations that most people have seen topographers use all around the world.
You can read more about the research I’ve done in my week 2 progress update here. The Inspiration for the project was a product made by Pro
Who is it for¶
Anyone trying to digitally replicate or reverse-engineer a physical object/feature.
3d scanning can have a lot of precision but can take time and the signal to noise ratio might not be acceptable for some uses. Robotic arms have limited range and large CNC probes can be extremelly precise but would take ages to measure everything.
The kind of coordinate measuring machine I’m building has a good balance between usable range, precision and measuring time.
What it does¶
It will have a wired probe connected to a few axis/sensor combos and at the press of a button it will give an audible cue that the sensor data was “sent” (to serial) to be interpreted as spacial datum as well as the relationships between them. The digitalization of these points can be used to reconstruct real world features for CAD.
Originally the plan was to work on the triangulation of the sensor data as well and maybe even storage in an available open-source format but it wasn’t possible within the dead-line.
My machine will function similarly to total stations; 2 sensors are used to infer the direction the probe is being pulled from and the another sensor will be used to determine how far the probe was pulled. Assuming a stationary machine the sensor data can be used to calculate each point relation to the machine’s reference location and between each other. Raw data will be relative so calibration must be done a posteriori to link digital sensor data and physical world measuring units.
What I designed¶
All the PCB’s, main motherboard and 3x daughter boards. The whole structure and just about everything that goes on/in it. Currently the only thing’s I haven’t designed myself are the bearings: The main Slew Bearing up top that was 3d printed from an available model, for the bottom I’ve printed a Tappered Roller Trust Bearing also from an available model but it’s a bit tall. I might change to a needle bearing if I can find a decent model. Don’t have the time to test designing and printing one now. On each for each of the 3 horizontal axis I also got 3x pairs of regular metalic roller bearings to make sure they move as smoothly as possible.
I couldn’t find flat coil springs and didn’t want to break perfectly working measuring tapes so I ended up designing and have been test printing some. Tolerances are iffy but I’ve managed to print at least 1 good spring which means it’s possible.
The successefull test happens to have 3 layer thick walls and given it’s strenght I think 2 layer thick walls will be the sweet spot between strenght and density as less thickness means I can add more coils to get more range out of the same volumed spring.
What materials, processes and components are used¶
The raw steel was included in the laser cutting service. The main structure will be metalic because I tried welding for wildcard week and kinda needed a few other parts I nested into the waste from laser cutting. One was actually to try and fix the lever that controls the attachment power clutch on my fathers tractor. (I think the technical term is Power Take-off clutch, PTO clutch.) If I can I’ll try to document that because why not.
That said the final machine can be modified to use cheaper materials for the main structure, like laser cut or cnc machined plywood/MDF. It would probably benefit from an overall rework looking at making everything more compact once all parts used are finalized as I had to give myself extra room in case I couldn’t source something. Like the flat coil spring, a metalic one would be an order of magnitude more compact.
PLA is used for all 3d printing and it’s just genetic PLA from which ever vendor happened to supply the FCT FabLab last. My instructor did mention we could try and print a spring in PETG to see what changes but currently I don’t think it will be needed for the final project.
FR1 cnc milled PCBs with an Atmel ATMega 328p TQFP32 micro processor and 3x AMS AS5600L SOIC hall angle sensors in individual daughter boards connected with I2c. The ATMega328p was available at the Fablab, the AS5600L I had to order from Mouser.com as most similar sensors weren’t available. All electronic components besides the AS5600L were stocked by the FabLab. Although the piezo buzzer I salvaged from an old broken 3d printer.
The NS25 2,54mm cable connectors were ordered from Mauser.pt. PCB header Cable header
I ordered the ball bearings and the diametric magnets from Digikey.
Cost¶
I’ll add a detailed Bill of Materials to my final project page, eventually.
Questions left to answer¶
Can I build my CMM in 4 months? How much precision can I get out of my machine? Can I 3d print the flat coil springs? What will the probe look like?
Personal Evaluation¶
As someone who knew virtually nothing about basic electronics I want to know If I could learn everything and build my CMM within the time-frame given. If I can do it then I’m sure I can design other machines in the future in much less time when I can focus on them.
Regarding my CMM after scaling back my plans at this point all I want my machine to do is that whenever the button is pressed the buzzer buzzes, an LED flashes, the register data is read from all 3x individually addressed sensors over I2C and the values for the status and angle registers are printed to serial.
If I can point the probe and get decently consistent readings on serial I’ll be quite happy. It may not be the most complex machine in the world. It may need more work before it’s a finished product but given where I started a few months ago I’ll be pretty proud of myself.
An effort was made to keep the pcb single sided, linning up the LEDs and Buttons to be easily covered and leave as many left-over pins accessible in case I or someone else in the future wants to add more features. I put in a lot of time to get the best possible sensors for the job. Hopefully I can get decent precision out of them.