Eric Weinhoffer

The Power of Universal Clipboard

Sat, 08 Aug 2026 00:00:00 GMT

I've been building more of my own software lately, usually starting with a small problem I want to solve for myself. One of the first was Dual N-Back, a memory-training app I made for Mac and later brought to iOS.

As soon as there were two versions, I ran into a very normal problem: how do I keep statistics in sync between devices? I considered QR codes and a more traditional cloud-sync system, but I quickly ruled out QR codes. I didn't love the idea of taking a photo of my Mac (or holding up my phone to my Mac camera), and QR codes are pretty limiting in the data they can contain, only a few thousand characters with a very dense image. The cloud sync system makes the most sense, but when I started dabbling in making apps for myself I decided to start out without paying the $99/year to be an Apple developer, which is what unlocks iCloud syncing. What I really needed was already built into my devices: copy and paste.

Apple calls the copy-paste feature between devices Universal Clipboard, one part of its broader Continuity system. Copy something on an iPhone or Mac and, if the other device is nearby, it can be pasted wirelessly. It's simple, usually happens within a few seconds, and can support many thousands of characters.

Sync without a server

In Dual N-Back, I added two buttons: Copy Stats and Paste Stats. The app turns my session history into a small package of text, puts it on the clipboard, and merges it into the other device. Each session has its own ID, so copying the same history twice doesn't create duplicates.

It isn't invisible, always-on syncing like we're used to in many apps, but for a personal tool that I'm still testing and building, I actually like it. I know exactly when data is moving, and I don't need to run a server or create a paid account.

I took the same idea further in Eric Health, my personal health tracking app that began as a way to log and display workouts but has grown into much more (maybe I'll write about it in more detail someday). Its clipboard package can include workouts, meal logs, medications, ailments, and other health data. I can make changes on either my Mac or iPhone, then copy everything to the other. Before anything is applied, the receiving app shows what will be added, updated, or deleted.

That preview is important. Copy and paste may feel casual, but replacing a clipboard full of text can still change a lot of data.

A better feedback loop

More recently, I realized the clipboard could help make the apps themselves better.

I used to notice a problem, leave the app, write it down in an Obsidian project note, and later copy that note into Claude or Codex. It worked, but it was enough friction that I didn't always do it.

Now Eric Health and EricWiki (yet another personal app I've been developing) have a small feedback area built right into the app. I can write a note right there in the app, while the problem is fresh. Later on, when I'm back at my computer, I tap Copy, move to my Mac, and paste it directly to the coding agent working on that app. The note starts closer to the thing it describes and it reaches the right place with fewer steps so the loop between usage, noticing something, and improving it is shorter.

A few notes for building this in Swift

Here are the lessons I would give another developer, or another coding agent, if they are trying to implement the same thing:

Use plain text as the common format. Encode the data as UTF-8 JSON. On iOS, UIPasteboard.general.string = json has been the most reliable write path. On macOS, use an NSPasteboardItem, include .string, and call writeObjects. A custom pasteboard type can be useful on Mac, but plain text is the important fallback for Universal Clipboard.
Wrap the data in a versioned envelope. Include an app identifier, a format version, and the records themselves. Reject data meant for another app, unknown versions, oversized payloads, malformed JSON, and implausible values before touching local data.
```
struct TransferPayload<Record: Codable>: Codable {
    let appIdentifier: String
    let version: Int
    let records: [Record]
}
```
Give every record a stable ID. Merge by UUID, not by a name or timestamp. That makes repeated imports safe and prevents duplicate records. Copying the same payload twice should produce the same result as copying it once.
Keep copy and paste explicit. Let the user press Copy and Paste instead of reading the clipboard automatically. Preview additions, updates, and especially deletions before applying them. Only show “Copied” after the pasteboard write actually succeeds.
Treat the clipboard as transport, not storage. Keep the real data in the app's normal database or files. Clipboard contents may be visible to other apps, so avoid secrets, warn users when personal data is included, and only paste data from a trusted source.

To be clear, this isn't the right sync system for every app. Large files, collaboration, and automatic background updates need something more robust. But for small apps used by one person on nearby Apple devices, Universal Clipboard is an unusually useful piece of infrastructure hiding behind an everyday convenience feature.

Welcome Back

Thu, 16 Apr 2026 00:00:00 GMT

Wow, it's been awhile since I've written a blog post, over four years! In my defense, a lot has happened since then, but now there are no excuses not to be sharing more often. Like many others, I feel more invigorated than ever to be working on personal projects, thanks to the advent of LLMs and coding agents.

This site is now hosted on Netlify and written (by Claude and Codex) in Astro. In the past I've found the whole domain and web hosting situation a bit intimidating, but no longer! Now I have all this help and a clever Obsidian > GitHub > Netlify pipeline that makes posting here very easy.

I'm excited to be back. Expect more from me soon.

Making Art Every Week in 2021

Sat, 01 Jan 2022 01:30:53 GMT

Like last year, I set a goal for myself to create and share artwork online every week of 2021. Again, it wasn’t easy, but I made it, had some fun and learned a whole lot. In 2020 I didn’t set a theme for myself so I wanted a bit more structure this year, and I did that with the following rules:

Digital Art Only. I followed this rule strictly and I’m glad I did. I learned a lot about vector artwork and how to use Affinity Designer on my iPad effectively.
Limit of 2.5 hours per week. I totally failed at sticking to this, but I’m not mad about it. I just didn’t keep track. I probably spent a little less than 2.5 hours on some weeks and far, far more on others.
Share on Twitter every Sunday. I stuck to this again, and I’m glad I did—one of the ways I motivate myself to do this every week is by sharing it on social media.

Week 10: Sunset City

Going into 2022, I’m changing it up with a new theme and two simple rules:

Physical Objects Only. The final work must be three dimensional. No digital work, and no drawings or paintings.
Share on Twitter every Sunday. Same as the last two years.

I suspect 2022 is going to be the toughest yet, since physical objects take more time for me to create, and I can’t work on them from anywhere. For that reason, I’m also going to allow myself to work ahead when necessary. In 2021 I did this a few times, to account for life events that would prevent me from working on art for 1+ weeks at a time, such as the birth of my first child. In 2022 I’ll continue working ahead when I want to, so although I’ll post work every week, that doesn’t necessarily mean I finished it that week. I can understand how some would say this doesn’t count as “creating art every week”, but that’s fine: it’s what works for me and I consider sharing new work every week an accomplishment.

Week 40: Color Connections

Follow along with my progress this year on Twitter and on my website page here, where I’ll post every work, or on Instagram, where I’ll post occasionally.

Making Art Every Week in 2020

Fri, 01 Jan 2021 00:02:31 GMT

At the beginning of this year, I set out to create and publish some sort of artwork online each week. It was not always easy or enjoyable, but somehow I persisted. In general, I don’t think very highly of social media, but it’s clear that sharing a goal online and having a desire to improve is good motivation for me.

I didn’t start the year with any real art goals but loved the fact that this process forced me to try new things. On a few occasions, I tried a process, found that I enjoyed it, and then embarked on a weeks-long journey of exploration.

This happened in a dramatic way twice, as you can see in the below breakdown: once with the Milling and Painting of little PCB pins (inspired by this Adafruit guide) and another with the discovery of Affinity Designer, a fantastic subscription-free graphic art program for iPad.

Lessons Learned

I didn’t have any strong intentions going into this regarding mediums to explore or time to dedicate to it each week. I learned a lot about myself, what I enjoy, and how to be more successful at this in the future. Two main lessons I learned:

Making Art I’m Proud of Takes Time

A lot of work that inspired me was done by real artists who have been doing this for years. I often felt like I was nearing the end of a work, but looking back at something that inspired me made me feel like I had a long way to go. I’d then spend hours refining the piece or working on details until it felt worthy of posting. I’ll be setting limits to prevent this in the future, as I’d like time to explore other things while still improving my art skills.

Graphic Art is Fun and Flexible

One of the challenges I faced with creating some sort of physical art was the time and parts required: I couldn’t really work on projects from anywhere other than home, and I often didn’t start making the physical piece until the weekend. It was also tough to motivate myself to work on something like a CNC milling project since I already felt fairly proficient at it and wasn’t learning as much.

Graphic art, on the other hand, was totally new to me and had a few noticeable advantages:

I could work on it from anywhere, as long as I had my iPad with me.
I follow (and continue to find) incredible artists on Twitter who provide a near-endless supply of work that inspires me.
I was starting from zero skill, so it was easier for me to come to terms with posting something lackluster.

Some Favorites

Here’s an assortment of projects that I find notable from the year, either because they were especially fun to create or because I’m proud of the end result:

Week 15: Skull Trio

Week 19: Small Wire Sculptures

Week 27: Colorful Shapes

Week 34: Missing Blue Skies

Week 37: Bubbly Wood

Week 48: Orbits

What’s Next?

I’m excited to continue creating and sharing art every week in 2021, with some modifications. Rules I’ll be attempting to follow:

Digital Art Only. I’m excited to continue improving my graphic art skills but hope to also share some digital sketches or paintings. I’ll continue taking advantage of the fact that I can do this work from anywhere.
Limit of 2.5 hours per week. I often spent too long finessing and obsessing over details this year, and I’d like to have time for other projects! This is more of a wish than a true rule. I have no idea if I’ll follow this all the time.
Share on Twitter every Sunday. Same as this year.

Thanks for reading! Follow me on Twitter to see what I post each week.

Critical items for organizing a shared shop space

Tue, 15 Oct 2019 21:08:27 GMT

Keeping all the users in a shared prototyping space safe, while still allowing them to work efficiently, is hard. That’s why keeping things organized, marked and easy to find is so important.

I recently published a post on the Bolt blog about this called Organize to Optimize and I thought it would be helpful to see a simple list of the items I recommended here. I’ve broken them into the categories used in the blog post: Marking, Organization, and PPE.

Marking

LW-400 Label Maker: you don’t need this one, per say, but do yourself a favor and buy a label maker.
Adhesive-lined heatshrink: heatshrink is great to have around for all sorts of electronic work and for marking tools.
Branded heatshrink tubing: I was pleasantly surprised to find that custom-cut, branded heatshrink is not very expensive.
Label Maker with heatshrink support: If you’re really serious about marking wiring harnesses, print on heatshrink with this fancy label maker.
Painters Tape: great all-around tape.
Twin-tip sharpies: the twin-tip style is the only way to go.
Dry-erase tape: great for marking things with dates and other notes that will change over time.
Cricut Explore Air 2 Vinyl Cutter: a decent low-cost vinyl cutter.
Roland Vinyl Cutter: a fantastic, more expensive, vinyl cutter.

Organization

Foldable cardboard bins: cheap bins that pack flat and are great for storing spare parts.
Stackable plastic bins: more expensive, stacking plastic bins. Especially great for small parts that would slip through the cracks of fold-up bins.
Bin-mounting rail: attach this rail to the wall for mounting stackable plastic bins.
Big storage bins: super robust, nesting bins for deep storage.
Boltless shelves: strong shelving that can be put together in a variety of configurations with only a hammer.
Kitchen utensil trays: surprisingly great for separating tools in Ikea drawers.
Tripp Lite power strip: the best power strip for wall or under-desk mounting. Comes in a variety of lengths.
Schaller bins: cheap molded plastic bins for storing endmills, cutting inserts, driver bits and other small things.
Wall-mounted broom hanger: ideal for keeping your mops and brooms off the floor.

PPE

Ear plug dispenser: wall-mountable dispenser of disposable ear plugs.
Safety goggles for glasses-wearers: the only good safety goggle option for glasses-wearers.
Face shield: protect your face!
Stainless steel hemostats: the length and material makes this ideal for grabbing parts out of a 3d printer support bath.
Spider strainer: pairs nicely with the hemostats for fishing parts out of the 3d printing support bath.
Sentry Air System fume extractor: a powerful fume extractor made for welding. Works well for extracting soldering fumes from multiple workstations.
Locline hose: the well-known vendor of rigid hosing for coolant in machining also makes large-diameter parts that work well for other plumbing jobs.
Dust extractor: this dust extractor is pretty affordable, considering its power.
Glove rack: keep your gloves organized!

If you’d like more commentary about why I recommend these items, read the Bolt blog post.

How do you keep your shop safe and organized?

Note: ericweinhoffer.com is a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to amazon.com.

Making the CNC Mill Enclosure, Part 1

Wed, 17 Jan 2018 06:25:25 GMT

My Othermill Pro is great for rapid prototyping of PCBs, plastic or metal parts. The speed of the spindle makes low-load applications like PCB cutting extremely quick, which is great. Unfortunately, the 26,000 RPM the spindle rotates at results in a pretty loud noise when a tool is cutting.

Due to this, I’ve been thinking for awhile about ways to dampen the sound of the machine while it’s operating while also improving the infrastructure that surrounds the machine by taking advantage of vertical space. Enter the CNC Mill Enclosure, a big plywood box that houses the Othermill, some space for plastic Sterilite bins (which I use for stock and project storage, among other things), and a big space underneath for some drawers.

My work volunteering at Jacobs Hall offers me access to a 4x8 ShopBot router, which is fantastic for cutting plywood. I knew I’d be purchasing the plywood through Jacobs, but hadn’t picked out a sheet prior and therefore didn’t know the final thickness. Since I intended on using some close-fit joints from the start, the sheet thickness was important, so I included it as a parameter in Fusion360 as I worked on the model, along with some others that I could quickly adjust on the fly.

For the mill to access all corners of the inside holes, and so the pieces could fit together at all, I added dogbones everywhere with diameter 0.005" larger than the diameter of the cutting tool (this is why I also included "ToolDiameter" as a parameter). The male plugs that fit the dogbone-shaped holes received a bit of a chamfer to aid in assembly, as well. The 0.01" gap between hole size and plywood thickness worked out well: it wasn’t a press fit, but a close fit that located the parts well and didn't require a ton of hammering on my end.

Instead of modeling all the panel pieces individually and assembling them later, I extruded each panel as a separate body in one large part file. This ended up working pretty well for the modeling portion. Unfortunately, when it came time to do the CAM, I had to rotate, align and move all the individual panels to fit them into the stock area.

The milling went smoothly and took about an hour overall with a single ¼” flat endmill. In order to prevent the pieces from flying out after they were cut out, I used the (fantastic) tabbing feature that’s built into Fusion360. It was super easy to use in software and removing the pieces only took a few minutes with a chisel once the cutting was done.

Quick side note: I’d never used plastic brad nails as a fixturing method on a sheet router before but WOW, what an awesome method. You don’t have to worry about the tool milling through any metal nails or screws and they’re fast to remove from leftover stock with a quick snip.

After a bit of sanding, assembly went pretty smoothly, despite my lack of clamps at home. Next up: sound dampening. For this I bought some foam from McMaster-Carr (PN 5692T15). In hindsight, I probably should've gone with less expensive foam from Amazon or elsewhere. Since I plan on replacing these in the future, I needed a way to fasten them to the wood in a removable way, and decided on 3D printed umbrella-shaped plugs. Each orange plug gets an M4 heat-set threaded insert (McMaster PN 94180A353) so it can be tightened against the wood. I love using heat-set threaded inserts for adding threads to 3D printed parts, as they're easy to install with any soldering iron:

I cut all the foam pieces down with a knife and used four or five of my printed umbrella things to hold them in place. Early tests of running the mill inside the box were not inspiring, but the door really made all the difference. I couldn’t fit it on the original 4x8 sheet but cut it out later on, using similar techniques. I left a window area, which I later filled with some tablesaw-cut polycarbonate. With some hot glue to hold the window in place and two hinges to allow the door to swing, I was ready for another test!

I milled a portion of the aluminum base part for my Spinlock mechanism once the door was installed and was impressed by how much the sound was dampened. I later add two simple sheet metal parts and a band to keep the door closed, and am fairly happy with the results so far!

Video: enclosure sound test

You may notice that the mill seems to be sitting very low in the box, or one of two already-installed drawer slides, in the above photos. The eventual goal is to create a bent sheet metal platform, on which the mill will rest and slide in and out of the box on the drawer slides.

So, that brings us to the list of improvements for Part 2:

Add the aforementioned sheet metal platform for the mill to sit on.
Once the platform is in place, I’ll need to drill a hole in the right side and create some sort of plunger so I can still hit the E-stop from the outside of the box.
The bottom two compartments are filled with junk now, so they’ll need some organizing. I’m thinking of trying some base-mounted drawer slides, which I’ve never used before.

Stay tuned for more!

Looking back at the first Maker Faire in Atlanta

Fri, 20 Oct 2017 23:43:30 GMT

I started the Atlanta Mini Maker Faire back in 2011 and am so proud to see that it’s still going strong. The annual event is going into its seventh year, is still free to attend, and is happening this coming weekend. Head over to their website to reserve tickets and see who's going to be there.

I haven’t been involved in the event since that first year, thanks mainly to some amazing folks I was able to work with initially and hand it off to. The main inspiration for me to start up the event was the fact that I could never make it out to New York or the Bay Area to attend one of the “main” Maker Faire events, as I was always busy at school. By utilizing some resources (and a parking lot) that belonged to the ME department at Georgia Tech, I was somehow able to create and arrange everything from California, where I was working at an internship during the summer of 2011. It wasn't anything super special: there were only a handful of exhibitors and the whole event only spanned the space of a small parking lot, but clearly it was the start of something grand. And for that I'm glad I started it.

I thought I’d share some photos from that first year to provide some perspective on how far the event has come. For comparison, here is Make's summary of the 2015 event. If you’re in the Atlanta area, check it out this weekend!

The Whitworth Three Plates Method

Mon, 31 Jul 2017 05:10:58 GMT

A key contribution to precision machine tools was the understanding of the importance of flatness, and the development of processes that are used to make surfaces very flat. Henry Maudslay invented the first machine that could cut standard screw threads and contributed to the invention of the now-common surface plate, which serves an important role in metrology today. A classic book called Foundations of Mechanical Accuracy described this and other concepts in detail. It’s available as a PDF here and in print here.

Typically made of granite, surface plates act as a datum, or the basis upon which precise measurements and movements can be made. They can be finished to a variety of grades of flatness, based on their intended use, and can even be used to build up precise structures (that whole video is a must-watch, btw). When dealing with such flat surfaces, tiny imperfections or gradual wear can have drastic effects: using a gauge over the same spot on a surface plate, or leaving it in a space where the temperature varies by more than a few degrees, can negatively affect their flatness. Due to this, calibrating and conditioning your plate is important. The process is really neat, partially because it requires an autocollimator and an incredibly precise repeat-o-meter. Watch a great video from Tom Lipton about the process here.

The neat thing about surface plates is that they do not require precision tools to create. By using the “three plate method”, developed by Joseph Whitworth, flat surfaces can be created by using gravity and a simple hand-scraping tool, or by lapping the plates against each other. By starting with three plates of relative flatness, rubbing the plates against each other in alternating pairs to remove the high spots can yield fantastic results.

The process can be completed in six steps, and then repeated until the desired level of flatness is achieved. In this visual explanation, the surface finishes of the three plates are exaggerated. Before beginning this process, the three plates (Red, Green and Blue) should be machined or ground to as flat a surface as possible, to remove all unnecessary lapping work. In addition, a fine abrasive compound is often used between plates to assist in material removal. Tom also has a great video series about this, which includes some compound and technique recommendations.

Step 1: Lap Red & Green

To begin, the Red and Green plates are lapped against each other in an alternating manner. That is, one plate remains stationary while the other is lapped against it, and then the opposite is performed:

At the completion of this step, the Red and Green plates “agree with” each other, but that is all.

Step 2: Lap Blue on Red

Next, the Red plate acts as the control (it remains stationary) while the Blue plate is lapped against it:

At the end of this step, both the Green and Blue plates have picked up the error from the Red plate. They do not agree with each other, however.

Step 3: Lap Green & Blue

Next, the Green and Blue plates are lapped against each other in an alternating manner:

Since both of these plates picked up the error of the Red plate from the first two steps, lapping them together removes some of the error from the Red plate, bringing them closer to flat. At this point, the Green and Blue plates are more flat than the Red plate:

Step 4: Lap Red on Green

Next, the Green plate acts as the control and the Red plate is lapped against it:

At the completion of this step, all three plates are of roughly equal flatness, but one (Green) is convex and the other two are concave:

Step 5: Lap Concave Plates

Next, move back to an alternating pattern and lap the two concave plates against each other:

This results in two fairly flat plates. At this point, only the Green plates needs to be brought to the same level of flatness as the other two.

Step 6: Lap Convex on Blue

Finally, the single convex plate is lapped against the Blue control plate:

And the result is three plates that are in agreement with each other!

It’s also worth noting that, although this yields great precision “from nothing”, Joseph Whitworth later improved upon the technique by utilizing engineer’s blue and hand scraping, as mentioned previously. Engineer’s blue alone would be an instrumental improvement over using no indicator - with the blue, it’s easy to see which areas have, and haven’t been, scraped.

Let me know if you’ve tried this yourself or have any comments about how I can improve the instruction. Most of what I know about the original Three Plates Method came from this PDF.

Making the Ranger Tab

Fri, 10 Mar 2017 22:56:19 GMT

I made a replica Ranger Tab as a birthday present for my Dad, who was a US Army Ranger. I found the below photo from an online archive of graduation photos. Thankfully, they’ve improved the large version of the tab a bit since 1983.

I began by tracing out the letters and outline from a downloaded image in Inkscape, then moved into Rhino to prep two DXF files for laser cutting, as my Othermill is not large enough to fit anything but the letters. Once I sent the files in for cutting, I began work on the letters. They’re each a little under a 1/4” thick and were milled out of hardwood on my Othermill with a 1/16" flat endmill. The milling went pretty quickly and the 1/16” endmill was ideal for getting into the tighter corners.

Once the lasercut plywood pieces came in I did a dry layout, seen in the photo below, with the milled letters, back and outline. At this point the plan was to mask everything before painting and then glue it all together.

I masked the backs of all the letters first, to make sure the glue had a bare wooden surface to bind to. After thinking about how I was going to assemble this thing, I decided against using only glue to hold the letters in place, and opted to drill small holes to use screws instead. I was concerned that, while clamping the letters to the back as the glue dried, it would be hard to maintain their position.

Before painting, I used an outline I printed to position the letters, mark their position, and then drill holes in the back piece. Once the through-holes were in the back piece I could temporarily clamp the letters in place and drill small pilot holes for the screws.

I started the painting process with two coats of primer on everything. Next, the letters and outline were painted yellow and the back was painted black. Notice the holes in the back piece in the photo above, for capturing the letters.

I'm really happy with how this came out, and enjoyed the process. I hope to make more signage like this in the future.

1 Week of MIG Welding

Sun, 22 Jan 2017 21:02:35 GMT

Recently, I had the opportunity to spend almost a whole week MIG welding. Before this, I'd never spent more than an hour or so with a MIG machine, so it was fun to see my progress throughout the week. Previously, the only welding experience I had was through a short intro class at The Crucible in Oakland. It was cool to get "under the hood" for the first time and get an idea of what MIG welding is like. I left with a desire to learn more.

So, here we go - my week of welding in three progress photos.

Day 1: pretty hideous

On the first day I really didn't understand the importance of things that make a good weld: how fast you move the gun, how much stick out to start with and maintain, the cleanliness of the tip, and most critical, the feed and voltage settings on the machine. I was just sort of arbitrarily moving the gun along the joint and wasn't getting good penetrating welds.

That night, I came home and watched the first of Jody's MIG Tips video series from (the fantastic) Welding Tips and Tricks. Going into Day 2, I had some idea of the sort of movement to make with the gun (lowercase cursive e's), how much stick out to have before starting a weld, and what sort of sound to listen for (bacon sizzling!). I still didn't really mess with any of the machine settings, though, so progress was minimal.

Day 3: still hideous but better

By Day 3, I was starting to feel a little more confident, and was getting addicted to making progress. My lowercase cursive e's were getting a little tighter, but I still had a lot of work to do. I think one of my biggest issues in the first few days was speed - I was moving too quickly and wasn't giving the puddle enough time to penetrate the metal. Also notice the little balls of metal around the weld - at this point I wasn't clipping off the end of the wire before each weld, so I was getting some splattering.

Finally, on the last day, I was getting somewhere.

Day 5: a decent looking weld!

I was getting some decent penetration, especially while going slightly downhill on joints like this one, and was starting to understand how to adjust the machine.

Now that I've been bitten by the welding bug, I'm excited to do more of it. It's clear that MIG welding is a useful skill to have and super handy for creating structures quickly. I definitely still have a long way to go and look forward to trying other types of welding in the future.

BOINC: Software for Volunteer Computing

Wed, 30 Nov 2016 17:00:00 GMT

Do you have a computer that sits idle for some portion of the day, unused and unloved? Well, I did, and recently started putting it to work while I'm away or asleep.

BOINC is an open-source software project developed at UC Berkeley, intended to allow anyone in the world to utilize the idle time on their own computer to contribute to scientific research.

Many scientific discoveries require intense computing power to comb through lots and lots of data. Instead of spending their budgets to rent computing time from servers somewhere, universities and other research organizations can utilize the untapped power of idle computers around the world.

Thankfully, you can choose how your computer spends its idle time, as there are many projects to choose from. I'm only contributing to the Einstein@Home project right now, which searches for pulsars with data from the LIGO gravitational wave detector. It was through the great work happening there that scientists were able to confirm the existence of gravitational waves in February.

In fact, just a few days ago, the discovery of the youngest-known radio-quiet gamma ray pulsar was announced. Oh, and Einstein@Home comes with an optional, but badass, screensaver:

Video: Einstein@Home screensaver

The application that supports all this is easy to use and provides a lot of information on how your computer is being used. It's super easy to specify times during which the software can and cannot run, which is great.

It's neat to think about other areas in which volunteer computing could be useful. I imagine it would be great for low-budget movie producers, 3d artists, and engineers that need to harness lots of computing power in order to render scenes, create animations based on physics, and perform complex simulations.

This is the sort of software that should be a no-brainer for most, as it's free and definitely a "set it and forget it" service. What's not to like? Consider installing BOINC yourself and begin volunteering your computer - for science!

Making a Quick-Change Bed for the Othermill

Mon, 22 Aug 2016 16:29:43 GMT

Fixturing is one of the most important parts of machining. Without good fixturing, your material is subject to vibration, which will not yield the best surface finish or the dimensions you want, and may even lead to tool breakage. I've recently found myself lazily leaving the Othermill's spoilboard in place and using tape for milling thick material, instead of switching to the more-rigid Precision Fixturing Set. So I set out to make the transition easier.

Video: Quick-Change Bed overview

What is it?

The Quick-Change Bed is a modified Othermill Spoilboard that is not only functional for milling thin material under low load, but also very quick to remove, to give fast access to the mill's mini T-slot bed.

Why?

The typical time required to swap between the Spoilboard and Tall Alignment Bracket is at least a few minutes. This long setup time results in lazily using Nitto tape where a Toe Clamp should actually be used, yielding often-poor results. Reducing the time to swap fixturing will, in theory, lead to better parts.

Even though you lose some workable Z height with the Quick-Change Bed installed, it's not really a big deal - the bed encourages you to use the Spoilboard surface only for thin material, and the T-slot bed for anything else.

Here's how it's made:

There are two main components to the Quick-Change Bed:

A modified Othermill Spoilboard, with slots for screw access and two pockets to give the Toe Clamp some room:
A spacer block that's milled to a precise size. It's attached to the modified Spoilboard with (4) M4 screws. This is the part that's actually held by the Toe Clamps:

Here's a video of the Quick-Change Bed in action:

Video: Quick-Change Bed in action

It still needs work! Here's what I'd like to improve:

Make the spacer block out of aluminum, for strength and conductivity. With a Delrin spacer block, tool changes can only be accomplished by manually completing the conductivity loop between the Quick-Change Bed and the T-Slot bed. This is not ideal. It will be nice to tap M4 directly into the block as well, instead of using press-in threaded inserts.
I'd like to modify the Toe Clamps and/or the spacer block to improve rigidity in the fixture. I'd like to try modifying Toe Clamps anyway, but with the caveat that they still have to be totally usable in every other situation, as I wouldn't want to have to remove them every time I needed to clamp something other than the modified Spoilboard.
The slots in the modified Spoilboard need to be a bit wider to accept the bolts that currently ship with the Precision Fixturing & Toe Clamp Set. I'd also like to shorten the length of the pockets on bottom of the Spoilboard - there's no reason for them to extend across the entire width of the part when clamping only occurs on one end.

Resin Casting: Part 1

Mon, 08 Aug 2016 02:25:57 GMT

I've wanted to try resin casting for a long time, and have finally done so a few times in the last three days. It's never been easier to do before, thanks to my easy access to CNC mills. Once I become proficient at this, I'm mainly interested in making two-sided molds for producing complex parts that would be hard to make otherwise, and casting with materials that would be difficult to machine (like soft rubbers).

I've made three silicone molds and casted two parts in the resin from one of those molds (see above). One side of the parts (the side that was in the bottom of the mold) look great, but the top is full of air bubbles. There are a few things I hope to improve in the process soon.

I used a prototype Othermill Pro to mill out the patterns for each silicone mold in machineable foam. This made quite a mess, but the mill didn't have a problem. The material itself isn't ideal - I hope to try something different soon. I think I can find something with a better surface finish.

Here's what the setup looks like. Not pictured: gloves, mold release spray, popsicle sticks and a scale. For the silicone molds, I'm using Quantum Silicones QM 262, based on suggestions in the fantastic Resin Casting guide by Michal Zalewski. It has a long pot life, so it's easy to work with - this gives you plenty of time to adequately stir the mixture (important!) and properly degass it in the vacuum chamber.

I'm still working on releasing silicone molds from the machineable foam in a less destructive manner. This was my second attempt at a mold, now in a more complex pattern - these are Bit Fans, which are used for chip clearing on 1/8"-shank endmills. Note that two of the female Bit Fan molds are missing the main column in the center, due to air bubbles.

Closeup on the Bit Fan mold. I was pleasantly surprised by the amount of detail from the machining that was picked up by the silicone.

My first few days went better than expected, but there are definitely improvements to be made. The resin I'm using is the very well-rounded Innovative Polymers IE-3075, which has a fairly short pot life of 8 +/- 2 minutes. With the low volume of resin I've been mixing at a time, I try to wrap up in faster than 6.5 minutes.

Things I hope to change, in order of priority:

Acquire a better vacuum pump. I think the one I'm using was broken and (maybe?) repaired in the past, but it isn't doing well. And with the cheap chamber I'm using, I simply can't degass resin fast enough.
I need a more accurate scale for weighing things.
Find some dedicated sheets of plastic, that have been faced flat, and smaller weights to help set up the resin parts after casting. The material I've been using thus far hasn't been flat enough, which can cause air bubbles.
Start blanketing the bottles with canned air to better preserve them over a long period.

I feel the need to reiterate that 99% of what I've learned so far has come from the fantastic guide by Michal Zalewski. Highly recommend.

More soon, I hope!

BMW i3 Factory Tour Highlights

Fri, 15 Jul 2016 18:50:47 GMT

The number of factory tour videos available online is incredible. I recently stumbled upon this one, which might be a new favorite. It's a tour of the BMW i3 factory in Germany, and it's fantastic. I've included some highlights below, for your convenience:

2:51 - use of a 5-axis Waterjet for cutting holes in carbon fiber. I never thought I'd see a Waterjet used in a factory setting like this.

9:38 - spot welding of bolts, which are fed through tubing, onto the chassis.
14:47 - tidying up the motor stator wires with some automated cable lacing, which is typically used instead of cable ties and other methods that leave obstructions along the cable.
17:25 - using a hammering tool for engraving part numbers.
18:00 - some fast vacuum forming.
18:48 - neat tooling for pressing parts together.
19:24 - absolutely incredible shot of tooling spreading apart to release a part.

19:48 - a nice milling shot with vacuum fixturing. All those air lines hold onto the parts as they’re milled, and then release once an operator can remove them.
21:58 - the operator uses a bone folder to help seat leather sections into some tooling. I highly recommend bone folders - A+ useful tool.
24:32 - this shot really shows the amount of tooling involved in making these things. The back arm is holding that giant tool inside the chassis to cure adhesive - it carefully retracts at 25:03.

27:44 - the application of a protective coating of some sort, followed by heating with open flames to cure it. All this is done by robotic arms, of course.
28:34 - painting begins!
32:43 - notice the complexity of jigs required to install part of the doors, and the individual e-stop buttons on them. Important!
36:36 - elsewhere, the chassis is being assembled. Look at the size of those connectors!
39:05 - an interface for the line, which gives the operator a glance at the status of vehicles as they move.

Share your favorite factory tour in the comments!

Microscope Zoom Mechanism

Mon, 06 Jun 2016 18:00:00 GMT

At work, we use this microscope for inspection. Overall, it's fantastic, although photographing what you see through the eyepiece is extremely challenging without an attached camera. Removing the main viewing assembly was the best way to decide if we could build an adapter ourselves, but it also revealed these glorious barrel cams:

This mechanism adjusts zoom in the microscope - as one of two knobs is twisted, a spur gear on the inside of the assembly is turned. This gear drives the two barrel cams via two other spur gears, driving the optics inside each eyepiece up and down. You can see the gears a little bit in the video below.

Video: microscope zoom mechanism

So, why use complicated cams instead of a simpler mechanism? Take a close look at the rate at which the four cam followers move - the rate of change isn't linear, so a barrel cam is by far the easiest analog way to control the position of the optics. Something similar could be done digitally, but this is far cheaper.

I've broken down what I think is happening on the inside of these eyepieces in SolidWorks, seen in the GIF below. Each optic is held in an assembly that can slide up and down inside, maybe on some small rails. There's also a threaded hole in each of the optical assemblies so the cam follower can be installed after the barrel cam itself is slid over the eyepiece assembly.

This is the sort of stuff that really excites me. Cams are such neat, simple mechanisms that can accomplish complicated tasks, like moving things at a non-linear rate. They're used all over the place, and for good reason. One of my favorite cam applications is described in detail here. Very cool stuff.

As usual, if you're more familiar with this sort of thing than I and can provide more insight into how this specific mechanism works, please share! I'm particularly interested in learning more about how the lenses slide up and down inside the eyepiece.

Microsoft Arc Mouse Teardown

Thu, 26 May 2016 18:00:00 GMT

The Arc Mouse from Microsoft is an affordable, battery-powered mouse that folds up for portability. Unfortunately, I recently lost the little USB dongle that came with mine, rendering it useless. I've seen the inside of a mouse before, but was intrigued by the fact that you can turn this one off just by folding it, so I took it apart to find out how it works. I initially thought the on-off feature was accomplished by a mechanical switch or contact inside, but I was wrong.

Unlike my last teardown, it was not immediately clear where to begin, as I couldn't see any screws during the initial inspection. So I started by gently prying up the thin plastic piece where your fingers rest, and surprisingly, it popped off with a satisfying click.

A look at the underside of the cover revealed two passive clips for retaining the part on the body of the mouse and some ejector pin marks - a sign of injection molding. Also note the two thin fingers protruding near the wheel cutout - these contact the momentary buttons for left and right mouse clicks.

At this point, I couldn't get any further without some serious destruction, so I went on a search for some screws to remove. Where could they be?

And this is where I learned something that I'll remember for future teardowns: screws are often hidden beneath rubber feet, stickers, and other adhesive elements. First, I removed the two rubber feet, revealing four phillips screws.

Hey look, more screws are hidden under the battery bay sticker! Admittedly, these took me far longer to find - I began cutting plastic away and only tore the sticker back after peeking inside and spotting some threads. Weirdly, the middle screw in the battery bay is the only non-Phillips one in the whole assembly - a T6 head.

The smaller folding portion of the mouse came off easily after removing two more screws. It's a pretty beefy hinge mechanism, made out of bent steel.

Now we're getting closer to see some PCB assemblies. But first, look at the size of this light pipe! It's responsible for making the only LED in the mouse, the on-off indicator, visible to the user.

After removing this piece, we finally get a nice look at the two PCBAs in the mouse. The bottom board is held in place with two small phillips screws (one of which is visible in the bottom left corner of the below photo), and the top board is prevented from moving side to side by a few plastic pins. Power is fed to the top board and the two are connected to each other via a single ribbon cable.

The mouse wheel itself is a single injection-molded piece. One side of the wheel sits on top of a momentary switch (for clicking the mouse wheel) and the other side, which has a hex profile, fits inside a little rotary encoder for recognizing scrolling.

The bottom PCBA, seen above, includes the optical elements that point downwards and detect when the mouse is slid across a surface. This is the bottom side of the board - the opposite side includes some other elements, including a connector for the ribbon cable.

That giant cutout in the center of the top PCBA is for the mouse wheel, which sits between the rotary encoder (on the left) and on top of the mouse wheel click button on the right. In the above annotated photo, the two red buttons are for left and right clicking, while the orange button on the left is for the little button on the side of the mouse. The blue connector retains the ribbon cable that hooks up to the bottom PCBA.

Now that I was far enough into the disassembly to see the PCBAs, I realized how the mouse recognizes when it's closed - a tiny reed switch found on the backside of the top board. The magnet inside the folding section accomplishes two things: it holds the USB dongle in its pocket during transit and activates the little reed switch when the mouse is in the folded configuration.

With the folding section removed, I was able to replicate the on-off action with a few magnets. Note the green LED under the scroll wheel shut off when the magnet is near and slowly fade back in when it's removed.

I shouldn't be surprised - reed switches are extremely cheap and very reliable within their detection range, so if you can afford to use a magnet for actuating them, they're a great option. We use four of them in the Othermill to detect when the windows are removed.

Ok, another teardown in the books. I hope you enjoyed it. I learned a lot with this one and look forward to the next! Please comment below if you have anything that you'd like to share about this teardown that I didn't mention.

Dial Test Indicator Teardown

Thu, 12 May 2016 18:00:00 GMT

We use Test Indicators all the time at Other Machine Co; whether it's during incoming part inspection or during final QA, they get a lot of use. Unfortunately, we recently had a casualty, and I was able to bring it home and tear it down. This is an inexpensive 5 tenths indicator from Amazon, by Türlen (available here). I'm not exactly sure how it was broken, but I was happy to take it apart.

I relied heavily on my iFixit Driver Kit and Magnetic Project Mat for this teardown. I highly recommend the former to all hardware folks - the array of drivers included in the set is expansive, considering the price, and the flexible driver extension is particularly handy. The magnetic project mat isn't quite as useful as often, but I'd definitely recommend it to folks that deal with tiny fasteners regularly.

As seen above, the teardown started simply, with the removal of a few phillips screws. Although the lever action between the pivot assembly and the rest of the components was clearly broken, it was neat to see how these precision indicators really function. Once inside, I first removed the sector gear arm, and then proceeded to remove the rest of the parts, which came out easily.

After removing as many of the internal parts as I could, I reassembled the sector gear arm and main gear between the two plates. Both the main gear and sector gear arm pivot on ruby bearings, which are pressed into the plates. You can see the tiny amount of pressure from my finger required to rotate the main gear here:

Next, I disassembled and examined the pivot assembly, which includes a long lever arm and the stylus subassembly.

A retaining clip is built into the lever arm, as seen above, which captures the pivot assembly and allows the user to rotate the stylus freely during use.

I was surprised to find a few bearing balls inside the main pivot component - this part holds the stylus and rides on two pointed screws, which gently secure it from either side.

I was able to use our microscope to get some up-close shots of a few components. I took these photos with my phone, freehand, so I apologize for the quality.

This final GIF explains the mess of wire that's shown in the first photo of this post - it acts as a radial spring inside the indicator subassembly. Unfortunately, I was not able to remove the bezel to get a good look at the internals of this subassembly, but I was pretty satisfied with what I'd learned at this point.

I was impressed by the quality of the components in this indicator, considering the price. It's definitely a good bargain! I always learn so much from doing teardowns, and this was no exception. The precision work in the gears was particularly surprising, but I suppose that without said precision, this would not be a five tenths indicator.

I hope you enjoyed this inaugural teardown on my blog. I'm hoping this is one of many.

Handlebar Plugs Build

Mon, 18 Jan 2016 16:00:00 GMT

I wanted to push the limits of the Othermill by making custom handlebar plugs for my commuter bike. I set out to make the style of plug with fingers that are forced open by a nut, gripping the inside of the handlebar.

Typically, a CNC lathe with live tooling is ideal for making a part like this, but I was excited to try using a complex fixture to hold the part during multiple milling operations. I couldn't think of a great way to make this in only two milling operations without taking the outer diameter of the stock down myself on the lathe, so it required three fixturing areas - one for each Op.

Op 1

To make one of two clamping areas in the fixture plate, I first milled out the clamping member and clamping area. Next, I drilled two holes in both the clamping member and fixture plate with the drill press. The holes in the plate were drilled smaller so I could press in two threaded inserts (McMaster PN 92398A121). With two M3 bolts, I was able to clamp the stock in place.

The original stock in place. I did cut it off on the lathe and do a face cut on the bottom side before milling, so it sat flat on the Othermill's t-slot bed. Both the fixturing plate and stock are Delrin.

The first machining operation was a ramp toolpath that brought the OD of the stock down to slightly smaller than the ID of the handlebar, so it could fit inside.

After the first machining operation.

Op 2

The clamping area for Op 2 was created in the same way I did the Op 1 side, only smaller.

Ready for milling.

The first toolpath in Op 2 is a counterbore:

Video: Op 2 counterbore

(I'm using my Air Coolant setup for chip clearing in these two videos)

Next up is a "scallop" toolpath to cut in the main 3d surface. I've swapped over to a 1/8" ball endmill to get a nice surface finish here.

Video: Op 2 scallop toolpath

Op 3

The fixture in Op 3 is the most complex, requiring a 3d mating surface in the fixturing plate and an aluminum cap to hold the workpiece down. First, I milled the cap, which will hold the workpiece down via four M3 thumbscrews.

Note the four press-in threaded inserts used here again. In total, there are 8 in this fixturing plate.

Once clamped in place and deemed reasonably level, the first cutting toolpath for this Op bored out a lot of the remaining material on the inside of the plug. Doing this and bringing down the wall thickness will allow the fingers to bend out and grip the inside of the handlebars, once a nut is forced into the bore.

No chip-clearing here, unfortunately. And those Delrin chips really stick to everything.

The next photo illustrates how this fixture could be improved - lining things up. Since I already had a counterbore on the bottom side of the piece, I really should've left a little plug there to better align the workpiece. I could've also drilled a hole and later pressed a pin into it. Anyway, something I learned.

First toolpath done with a bit of misalignment, now onto the fingers.

Video: Op 3 inner bore cut

The fingers cut without much trouble. This part is done!

The Nut

The nut is a small, tapped (M4), tapered HDPE piece with two protrusions that grab the fingers of the plug, preventing it from rotating. It's cut to fit the plug snugly, and is forced down into the plug as the user tightens the M4 bolt, expanding the fingers to grip the inside walls of the handlebar.

The next two photos illustrate how this works:

Loose.

Tight.

Once the nuts were complete, my new handlebar plugs could be installed:

Welcome

Sun, 17 Jan 2016 05:33:43 GMT

Hello! Here on the blog I’ll post periodic updates to share a thing that I’m working on. I will also post more in-depth coverage of a project, if it’s exciting enough. Possible topics include machining, CNC, 3d design, bicycles, juggling, and outdoor activities.