This is a list of weekly developments for the drawing-thing updated every week.
Big updates from last week!
I decided to make a single control board to make it easier to send people supplies to complete due diligence on what the cheapest version of the machine might be. A contributor to SVG PCB recently submitted some PRs adding excellent Gerber support so I took great satisfaction designing the boards in my own CAD tool and ordering them from JLCPCB. You can find the design here. You can also play with the de# the SVG-PCB editor here.
Here is a picture of the board.
We've made heaps of pen holders but none were stiff enough. Not anymore! By 3D printing a rail and using a similar v-wheel carriage design to the rest of the machine we were able to make this slick little servo-driven z-axis which works very well. I'm extremely pleased with this design!
Thanks to finally having a good pen holder we were able to put together a well functioning complete machine. Conveniently this happened on the day Ella came to visit cambridge so we got to run it together.
We also wrote a small Turtle drawing library which can drive the machine. You can first visualize what you will draw in the editor and then run the machine.
Here are some shots of the complete device below.
This week I released a toy re-implementation of the networking library we are using for modular things. You can find this library here modular-things/nosap. The point of this is better understand what OSAP (current JavaScript implementation here) does and what it needs to do. Primarily it's a library for packing and routing messages across devices.
It differs somewhat from what Jake (the original OSAP author) describes in his approach for the sake of making things simpler and easier to comprehend. Hopefully these difference will diminish in the near future. In a series of long conversations with Jake I've come to better understand OSAP itself and I believe to persuade him to remove notions of hierarchy from the graph design.
I'd like to make a visualizer for networks in the toy implementation which can be used to investigate actual OSAP networks in the future. Which brings us to another thing we worked on last week.
We released a TODO list to help people find ways to participate in the project. This will evolve into proper GitHub issues with good project management but we started just by dumping thoughts and tasks into this Markdown document.
This week we tested some resin parts, built two pen holders, and published the BOM (bill of materials) for the most up to date machine design.
We ordered the parts from JLCPCB's 3D printing service which is amazingly cheap and quick. The concern using resin prints over FDM ones is that resin tends to be more brittle. The print quality is very nice and consistent though so the verdict is still up in the air.
A friend Rob Hart helped design a sliding mechanism for a pen holder with very few parts. It uses 3 little posts on compliant levers which act as little springs holding the moving portion of the de# place. The other pen design is an adapted version of the lever which B. Smith used all the way back in our v0 prototype. You can see both of these designs in the photos below.
Below is the full slider machine with resin parts:
Here is the front of the slider:
Here is the inner portion:
Here is the same machine with a lever pen holder:
I also organized and published the materials and STL files for this current machine design. The BOM can be found here.
https://hackclub.slack.com/archives/C04GCH8A91D/p1678497440093879
Time for this week's update! We've got yet another rework of some key parts of the design.
First, those who have been following know I've been battling for weeks with making the motor brackets stiffer so they don't bend under loads from the belts. They were starting to get a bit chunky when what I really wanted to do was make the side supports taller. You can see in the second picture I was able to thin out the material and make the stiffest bracket yet. I also made the lever the belt pulls on a bit shorter by moving the motors closer to each other. The two main wins in decreasing the distance were removing some extra material where the aluminum extrusion attaches to the rest of the bracket and moving the foot to the other side (so it's no longer between the motor and the plastic).
The biggest change yet is we're testing a 3D printed carriage design. It won't be as stiff as the sandwiched metal plates but it should be stiff enough and it's much easier to assemble. I was able to knock more than 10 parts off the bill of materials (BOM) with this.
I also spent some time with our collaborators at MIT looking at the networking system used among the boards in the machine. I started to implement it from scratch to test my own understanding and to see if there are parts we can simplify. One of the most interesting parts was learning about consistent overhead byte stuffing.
Looking forward we still need to sort out our pen holder, design a custom connector for board-to-board communication, figure out power delivery (right now everything is powered by 5V), and put the whole system through its paces.
https://hackclub.slack.com/archives/C04GCH8A91D/p1677875892149909
This week, I got the latest version of the 2d-cut parts fabricated using metal for a better test to see how the machine flexes/bends/moves when tensioned. I sent the DXF files to a super cool local manufacturing company to get them fabricated. The pieces were cut from 1/4" thick aluminum on a waterjet, then sandblasted to remove burr.
https://hackclub.slack.com/archives/C04GCH8A91D/p1677285634597069
Time for this week's update. The hardware design is starting to anneal. The key thing I've been paying attention to is how belt tension bends/twists parts of the device. This is primarily an issue on the side motor brackets. You can see how it's been beefed up in the first image.
In addition to Ella's pen holder we experimented with a minimal part sliding design which you can see in the second image. Thanks to Rob Hart for assisting with this one.
Ultimately the flat pieces (carriage plates and feet) will be made out of metal. Consequently they'll be much stiffer than the current plates which bend (causing the wheels to wiggle) if you stick out floating axis arm and press on it. In order to get a more representative test I cut some plate from Delrin on a ShopBot (seen in image 3). The best test though would just be to have them fabricated from aluminum as the de#tends. I'll try an order for that next week.
https://hackclub.slack.com/archives/C04GCH8A91D/p1676690988312429
This week’s update in development of the drawing thing includes the fabrication yet another prototype and improvements correcting some issues. The images below are the CAD models of latest prototype. Notable revisions include:
- Using flanged bearings instead of normal bearings inside the 3D printed idlers. I found the normal 625 bearings a bit difficult to fit inside the idlers and cracked some plastic idlers in the process of getting them in.
- The carriage is now a sandwich, which helps with making it more stable. In previous designs, it was a flexing a bit. Adding the extra v-wheel was also helpful.
I also thought about a new design for the pen holder. Specifically using a compliant mechanism, to reduce the number of parts needed, and because I think they are cool. I haven’t fabricated this part yet, but it will be one of my objectives for the week ahead.
https://hackclub.slack.com/archives/C0M8PUPU6/p1676083722576069
Here's another update on the drawing machine @Ella and I are making. You can follow along in #development-of-things-and-stuff.
This week we made another prototype design (1st photo), fabricated it, then yet another prototype design correcting more issues we found. The improvements involved 3D printing idlers with 625 bearings inside in order to reduce part count and hone in on the dimensions we need to align the belt. The belt sits in a cross formation which allows two motors to control the x and y motion through coupled kinematics. The idlers sit on the carriage which moves and redirect the belt along each side of the perpendicular axis. You can see this in the 2nd photo below.
The main issues with the design we fabricated this week were the whole machine tipping because the feet had too narrow a base, the belts being misaligned, and the carriage flexing. In the third photo you can see we made the machine more stable by adding wider feet, we repositioned the idlers to align the belts, and we also swapped the double beam for a single one (plus added an extra v-wheel) to help with the carriage flexing issue. In the final machine we'll probably make the carriage plate out of aluminum with a service like SendCutSend which will make it much stiffer than our 3D printed prototypes.
Ella also made a very nice pen holder (4th photo) which cleverly uses nails as rails.
I took a pass at creating an SVG importing interface within the modular things editor. You can import svgs which get separated based on color, then rearrange them and apply some basic affine transformations to them. This is the start to an interface for dropping in vector drawings which the machine will then make.
Very excited to start making art with this soon!
ohyeah it's a drawing-thing