Hopefully, you’ve gotten a rough idea of my workflow if you’ve read any of my writeups. That’s, like, the entire purpose of this portfolio; the issue with presenting them in terms of project specifics is that I don’t get to talk as much about my general process without the risk of boring you. Despite not having any formal-training, I like to think that I follow a logical progression of steps that mirror the engineering process by using automations along with my experience and skills.
With simpler projects, sometimes I’ll jump right into computer modeling but ideally I’ll start off conceptualizing with a rough sketch and/or a list of requirements. My medium of choice for this stage is paper and pen, as it’s often the quickest and easiest option for visualizing the idea in my mind. These half-baked ideas emerge anytime between waking and lying awake at night ruminating, so I make sure they are labeled and clear enough to decipher whenever I get time to expand on them. Between paper thinking and engineering a solution, sometimes I’ll draw out something similar in CAD programs that I can expand on and alter quickly, banging my head against a wall until I’ve got a good idea of how to progress. Sometimes I’ll even throw together a rough 3D model with simple geometry if I need a visual for myself of others, or some other medium to convey ideas; I try to be flexible to new strategies, since I can try new ones quickly with computer aid. As such, the digital and editable nature of CAD makes it easy to hammer out a to-scale concept without being good at sketching, which has let me focus on the ideation and critical thinking bits rather than artistic skill.
After the rough sketches, I get to designing mainly with automated tools. Here CAD has become my main medium for solving problems, which I can incorporate chain calculators, variables and auto-updating formulas, and other tools into to limit the time spent not on design. I generally start with a backbone CAD-sketch or two, which will model key dimensions and interactions accurately and in a more usable fashion than my concepts. I used to just brute-force ad-hoc 3D models like I do now for some concepts, but that limited the effectiveness of my designs in the past. For example, after I landed on designing a four bar linkage for my 2023 robotics season I could make an admittedly shitty arm with guess-and-check but only made a versatile, effective one with more intentional planning.
After making a drawing, I typically address smaller details or components in order of importance (ie. power transmission, mounting, lightening) using separate sketches/drawings to keep each organized. Being able to use a computer program in place of meticulous hand-drafting and physical organization has let me self-teach effective compartmentalization and approaches to design, which I wouldn’t have been as enthused about had I gone into engineering a century ago.
These drawings are a good chunk of CAD modeling, but on their own they aren’t too usable outside of being handy visuals. Different modeling tools/scripts that build off of these blueprints let me thicken them into a 3D object, make curves, etc. Besides a couple key details, they aren’t too interesting or relevant. The main ones that automate my process are variable and formula tables I use to keep track of and call upon key dimensions, importing phantom-models to constrain to and picture aids for reference, and a category that repeat former operations. In the end, I’ll get a 3D-part that took more effort to problem solve and design than to model due to the tools I use and my experience with them.
After I model a part, I like to insert it into an assembly containing models that it’ll interact with. Sometimes this is an actual model of another mechanism or situation, sometimes a simplified one, and sometimes even just a computer drawing. Generally, it’s pretty clear if something will or won’t work if I do have a model to verify with, so after checking I can pretty quickly solve that issue with some edits or move on.
Next up, if a model passes an ad-hoc inspection, I’ll 3d-print a prototype that answers some questions or a final version if I’m done prototyping. I’m fortunate to have access to printers and CNC routers, and that I can use student software to fully utilize them. Generally, my go-to prototype is a low infill, thin wall 3D-print that I can prepare in software with some clicks before the machine builds it autonomously. These are generally quick, low effort, and cheap since it only uses a bit of plastic and machine time. Final versions take more of those resources, but less than I’d use if I were to make whatever they are myself. The easy iteration that a printer provides has let me knock out some small projects in a day, which is pretty sick.
If a part can’t be 3D-printed or isn’t suited to it, I’ll mill it on a computer controlled milling machine. Doing so is almost always more involved than printing, since there is only so much automation able to be done to this process for a reasonable price. With some input a CNC router machine can autonomously spin up a cutting edge to tens of thousands of RPM and control movement in three axes, but it really can only trace out preplanned paths that you provide. Even the fancy 5-axis mills that can change their cutter and some other insanely automated stuff require a good amount of prep and hands-on work, which the machines I use have in excess. Most of the preparation can be handled in software that converts your data and model and decisions into machine-readable code, which has allowed me to get the foothold in manufacturing that I do; I’d love to go into all of the specifics, but that’d involve too many equations, geometry considerations, and bored readers for this writeup. The cliff-notes of the process are that I’ll arrange some parts in CAD and export them into Fusion, which is my CAM program; from there, I’ll use semi-automated presets that I’ve honed for specific cutters and path templates to do individual material-removing operations for boring holes or cutting a profile. Most of the time I’ll also use a simulation feature to make sure I don’t destroy something that I don’t intend to, then tweak settings based on what I learn.
After making programs in Fusion, I’ve still got work left to do. As a precursor to tiny aluminum chunks cut from the material being shot at me, I have to set up the machine by securing the stock down by any of a few various methods that vary in the degree of risk-for-metal-flying-across-the-room-at-low-speeds. I also have to attach the specific cutting tool for the next operation before running its associated files and set the program start point. After all that, I’ll tell the machine to run then watch it autonomously follow instructions. Sometimes I’ll have to change the tool or readjust the stock between operations, but then if no crashes or power outages occur I can just run the next one. Yes, I have had the power go out mid-operation and yes, it took some ingenuity to continue milling.
If it’s not clear by now, I love the machining stage. What comes next is often just as interesting: testing and iterating. How a part is tested varies wildly, but for part of the process I’ll use a digital caliper to check that it’s dimensionally accurate. I like dial calipers, which have a spinny bit that points at the correct dimension, but my digital set automates reading it and removes a lot of human error. More human error than I’d like to admit. If a prototype passes or doesn’t pass whatever test I throw it at, I’ll make a version that answers further questions or that is more finalized. That involves alterations in CAD, using a computer brain for the processing and mine for problem-solving; using manual tools to iterate is considerably slower, so I’m thankful for this division of labor. I couldn’t be at the level of understanding of engineering concepts without these automated tools, nor could I have learned machining without formal training. These tools have let me focus on the cognitive skills of the disciplines I’m passionate for, quickening the learning process and creating possibilities I’d have squandered without them. Sometimes it feels a bit like cheating to go almost straight into CAD, but in the end it fulfills the purpose of all automated tools: making a person’s efforts faster and more fruitful.