Working from afar

Hey everyone. I hope you’re doing well. TL;DR, if you’re *not* doing well as a result of COVID-19, and if I (or The Prepared) could help, please be in touch.

I’ve been working from home for the past week and a half, and it has been a real emotional rollercoaster. I’ve spent a decent amount of my time trying to help out with medical supply sourcing for NYC - Tyvek suits, respirators, ventilators. The shitty thing to say about it is that it’s totally fun. I genuinely enjoy figuring out how you’d set up a new manufacturing line for a weird product, and right now pretty much everyone I’ve called (side note: using a telephone is great) has been more than willing to help me learn.

The other important factor, of course, is that I’m just *so* lucky that I can seamlessly move the large majority of my work to a makeshift home office, which then gives me time to bake bread every morning and also the freedom to take half hour breaks to hunt through the kitchen with Nora to figure out what dissolves in water and what doesn’t, or to step into the backyard with Tess to let her poke around in the dirt. Note also that by any measure, Ada has taken on a much larger burden than I have, providing us all much needed stability and joy.

But the rollercoaster part mostly has to do with the place I’m at in my career, and the degree to which I’m both in an ideal position to help the response effort and also feel woefully unqualified to take it on. I am presumably not alone on this - at least the latter part.

I think there’s another weird factor here, and it’s the combination of the feeling that this is a generation-defining event and also the sense that it’s all being hashed out on Twitter and Slack and Reddit. All of my reference points (wartime production during WWII, The New Deal, the Apollo program) for existential collective struggles involve a lot of top-down coordination, and it’s hard for me to imagine a path forward on this disaster that is fundamentally reliant on individual people 3D printing plastic bits and hand delivering them to intensive care units. It’s possible that my impression of wartime and wartime-adjacent efforts is naive, and that under the surface they’ve all been chaotic and ad hoc. It’s also distinctly possible that the hierarchical structures present in 20th century industrial America are actually counterproductive. But the current crisis still feels confusing.

Anyway. If you’re in need of help related to COVID-19 and I or The Prepared can provide it, please let me know. And wash your hands.

To be able to walk barefoot through the fields of the mind

From an old recorded celebration of MIT Building 20:

One of the problems you see in modern research is whether you can persuade the authorities (at NIH, NSF, and so forth) to let you take a chance. Sometimes it's very difficult to defend chances. That freedom to be able to walk barefoot through the fields of the mind... that was something that was so central to Building 20 at the time... it was astonishing. There is a curious aspect about buildings where there has been dedicated work. You might almost say that it's far better than a monastery; it's more like a temple, where there's been dedicated work and where the scars and the marks on the walls and so forth all have meaning. All of the artifacts that are left aside somehow or another have meaning. There is, as it were, a passing on, not of the tradition, but of a vague memory that somehow or another is very rich.

- Jerry Lettvin, Professor of Electrical Engineering and Bioengineering

Recent lessons, pt. 3

I’m recently off of my second paternity leave, and again have spent some time reflecting. So, like parts 1 and 2, a few (more or less random) thoughts:

• As one-off projects - discrete things that require planning, management, and execution - nothing beats cooking a good meal.

• As one-off engineered objects - things that require design, fabrication, and systems integration - nothing beats a bicycle.

• Writing software brings with it both a strong sense of opportunity and also a creeping feeling that the whole thing very well might be fake.

• Empires can be built through persistence - through late nights, doggedly following every lead, and waiting around long enough to catch a little luck. But finding your crew - without which your empire might not be worth it - requires patience, empathy, and *incredibly* good timing, and in some ways these are things that empire-building seems to breed out of people.

Wanting people to see it

Recently a reader wrote in asking for advice about gaining an audience for their work. This is something that I’ve had some modest success at, but also something that I feel conflicted about; my reply, below, felt worth sharing.

If you want to get attention, tackle a subject that other people want to learn about. This is hard to do, hard to judge, and kind of obvious to point out - but it's critical if you want to reach a large audience. Bin of broken dreams (the piece that, despite being outdated and outwritten by many subsequent years of work, is still this site’s most-trafficked blog post month after month) was successful largely because I was debunking an idea that a lot of people were either invested in or curious about. “Just press print” was a popular concept, and the story I presented in that piece directly countered it. That was lucky, but there are a LOT of other ideas floating around that would warrant a similar approach.

I should caution you, however: If you prove yourself a good communicator - and worse, if you build yourself an audience - then you may find yourself increasingly boxed into marketing either your own or other people’s stuff.

That may be fine; I value communication, and enjoy promoting both my own work and good shit around me. But it’s also limiting. On almost a daily basis, someone writes me to ask for either an introduction or help promoting their thing. But my (apparent) skills in writing, and in building a community around myself, have largely overshadowed any of my more technical or managerial traits.

I started blogging in early 2013, and I started The Prepared late the same year. I would do it all again, but I recommend that you consider what you want your day-to-day work to look like - and remember that fame (however modest) mostly breeds a continued pressure towards fame.

Swag

To my surprise as much as anyone else’s, the little manufacturing newsletter that I started five years ago has swag now:

Thanks to everyone who’s read this site - and The Prepared - as it has grown into an actual thing. Now head on over to The Prepared’s Subscriptions page to get on the distribution list :)

Why it's sixty dollars

Note: If you’re reading this in the 2018 holiday season, use code “WHYITS60” to get The Public Radio for its mid-2017 price of $39. Happy holidays :)

---

Recently The Public Radio, the FM radio that I co-created, rather unceremoniously instituted a 33% retail price hike. It was a tough decision to make, and one that felt both like a capitulation and an experiment. It came about four and a half years after we first launched the product, and about a year after we relaunched it with the intention of building a sustainable, US-based supply chain. In other words, it was pretty late in the product’s life cycle to be making such a drastic change - something I thought a lot about.

For those who haven’t fully thought through what it would mean to manufacture consumer electronics in the US - and for those who’ve idly contended that we live in an age of mass customization - I offer my experiences.

How our supply chain works

Like all consumer electronics, many of the components (microprocessors, passives, electro-mechanical components) in The Public Radio are either only made in China or simply not practical to purchase from the US. For instance, the FM receiver module we use is marketed by a US company (Silicon Labs) but made in China, and even the worlds largest electronics brands would have little power to change that. Similarly our speaker could theoretically be made in the US, but to do so would effectively kill our cost structure, probably moving the radio’s retail price well above $100 from an already expensive $60.

As a result, there are only a handful of physical components which we source from the US: The mason jar that the radio is housed in (purchased from Newell Brands, which has a license on the iconic Ball jar product line) and our custom cardboard box, which is only cost effective when purchased within ~100 miles of where the radio’s final assembly occurs. We’ve seriously considered changing jars to a Chinese made version, which would probably result in a 50% price cut on that component, but the scale of our production didn’t warrant it. And besides, being able to order a few thousand jars at a time is quite appealing in contrast to the 10-20,000 unit orders that an overseas supplier would require.

Our assembly labor, however, is in the US and accounts for roughly half of our cost of goods sold. Our excellent assembly partner (Worthington Assembly) produces printed circuit board assemblies in batches of 324 (36 panels of 9 PCBs each; this number ends up working nicely with the trays we store the radios in). They then use those PCBAs, plus three mechanical components which we supply them with, to create mechanical assemblies which sit in inventory until a customer places an order. At that point a mechanical assembly is taken off of a shelf and programmed to the FM frequency the customer requested. It undergoes a full functional test and is put in the mail - typically less than a day after an order is placed. See this blog post for a more detailed description.

In other words: We’re making a piece of consumer electronics just-in-time in the US.

I can’t stress how unusual this is; it’s simply not how things are done. Most consumer electronics rely intensely on the supply chains of the Pearl River Delta - the same places where our speaker, battery clips, knob, and potentiometer (and likely a number of other tertiary components) are made. They’re made in batches between 10,000 and a million, and are then containerized and shipped to large fulfillment centers in places like Pennsylvania’s Lehigh Valley and Chino, CA. There they sit, sealed in their retail packaging until someone places an order. And when that happens, a relatively low-skilled worker picks it off a shelf and puts it in a box.

As appealing as this may sound, it isn’t a realistic option for The Public Radio. TPR is customized to one of about 300 possible FM frequencies, and unless we were willing to purchase a huge stock of pre-programmed radios (with more of the popular frequencies, just like T-shirt makers buy more medium and large sizes than XXXL) it just wasn’t practical to assemble it in China. A few proposals that we considered:

1. Assemble un-programmed radios in China, put them in boxes, and ship them to the US; then use a contactless programming method to program the FM frequency at a US fulfillment center. This isn’t crazy, and Josh created a pretty impressive technical demonstration. But finding a fulfillment partner was problematic, and switching to this process would effectively prohibit us from doing the laser marking that our radio station customers love so much.

2. Do partial assembly in China (mechanical assemblies only - no jars or boxes) and do the rest in the US. Similar to #1, finding a fulfillment house was an issue - and the laser marking would have been tricky as well. Additionally, we’d be stuck using expensive US-made mason jars, keeping our COGS high while requiring a bunch of additional work on our part.

3. Just add a tuning knob, turning the product from “a single channel radio in a mason jar” to “a mason jar radio.” This is a fair suggestion, but one that (for philosophical reasons) didn’t make sense for us.

In summary, The Public Radio is a relatively rare example of a consumer electronics product whose supply chain relies heavily on the greater Shenzhen ecosystem but then is assembled and fulfilled completely from the US. We’re proud of this fact, but have been pragmatic (as opposed to idealistic) about how our supply chain operates. So with that in mind: How exactly does The Public Radio end up as a $60 product?

Where the money goes

So, here are the big takeaways:

• Our cost of goods sold has varied a little over the past year, but it’s currently around $18.54.

• 75% of our cost of goods sold goes to US vendors, and the majority of that stays in the US.

• About 50% of our cost of goods sold goes towards assembly labor. Part of this is automated (pick and place PCBA) but more than half is hand assembly.

• The two most expensive components in our BOM are:

  • Our speaker, at $2.35/ea, accounts for 12.68% of our COGS. This is manufactured in Dongguan (see here for photos of the facility) and is custom for us.

  • Our FM receiver module, at $1.53/ea, accounts for 8.26% of our COGS. This is a standard component from Silicon Labs, and I’m told that its claim to fame is that it was part of the original iPod Nano. This is probably the first component on our list to replace, as there are certainly less expensive chips options with all the options we’d need.

The broader story is that at our scale (5-10,000 units per year), consumer electronics is expensive. Purchases at that level simply don’t bring much leverage, and our just-in-time production model ends up being expensive on a per-minute basis (partly due to switching costs).

Here are the raw(er) numbers:

COGS by vendor location

How you sell a product with an $18 COGS

Back in 2014, we sold our first few units for $45 direct to consumers. Our first Kickstarter sold radios for $48/ea with shipping (about $5 our cost) included. In our second Kickstarter, we dropped the price to $39/ea and then charged shipping separately.

In mid 2017 we got our first retail account and listed The Public Radio for $45/ea. We set up drop shipping terms with a wholesale price of $33.75, a price that was intended to give equal margins to us and our retailers.

Sadly for us, this arrangement failed to gain traction. We got about as many 2017 holiday orders as we had hoped for (and about as many as we could have handled; see here for more history, including the ~1000 radios that I personally assembled and shipped from my basement on nights & weekends), but per the section above it became evident that we would need to increase our volumes significantly to make the numbers work out at a $45 price point. And while we maintain 4.9 star reviews on Uncommon Goods, the process of signing up large drop-ship accounts is simply a new skill, and one that hasn’t so far aligned well with our operating model.

So, we increased our prices. On the one hand this sounds crazy - we had a relatively successful product that we wanted to sell more of; the typical answer to that scenario is probably not to charge more for it. But the price increase allowed us to offer keystone pricing to brick-and-mortar retailers, and it gave drop-ship retailers significantly more margin to play with - making us a better partner for them as well.

In the meantime, we doubled down on direct-to-consumer sales. This also is a new skill, but I’m happy to say that since our price increase, we’ve managed to more than double both our direct-to-consumer revenue and our direct-to-consumer sales volume; in other words, we’re selling more units and making more money off of each sale.

Looking ahead

Every time I look back at The Public Radio, I marvel at just how much work has gone into it. It now runs as a well oiled machine; we’ve invested heavily in order processing & inventory management automation, and have structured our vendor & customer relationships in ways that allow us to produce a piece of consumer electronics as a side job. And in spite of (or perhaps because of) a fair dose of uncertainty early in 2018, it’s now clear that this will be a profitable year for us.

But there’s still a lot of work to be done, and I continue to wonder whether there are fundamental changes we could make to allow our US based supply chain to thrive even more. I’m lucky to have been able to explore this kind of manufacturing as much as I have thus far; I look forward to more of it in the next year.

The Prepared is hiring a part time operations manager

The Prepared, my (excellent!) weekly newsletter, is growing. And while I never intended it to be anything other than something I did to keep myself busy in the evenings, it has become evident The Prepared needs part-time help.

So I’m excited to announce that The Prepared is hiring a part time operations manager. The role will cover a variety of research, archiving, and growth operations, and is remote-friendly. If you or anyone you know is curious, energetic, and interested in helping build a community of like-minded people (most of whom work in or adjacent to manufacturing), please apply!

As always, thanks to The Prepared’s sponsors, patrons, and subscribers for making this possible.

3MF on GitHub

Back in 2013, I wrote a short blog post titled “ISO should be an open source project.” In it, I (somewhat offhandedly) argued for a new paradigm for cross-industry collaboration - one in which both individuals and enterprises work together in the open.

Well this week, the 3MF Consortium completed its transition to hosting all of our specifications - as well as sample files, implementations, and much of our internal consortium documentation - on GitHub. Where they can be viewed, downloaded, forked, and pull requested by anyone in the world.

This transition took a considerable amount of work, and maintaining a culture of openness will require maintained effort. But the benefits are significant: Not only will we encourage newcomers to use and improve our work, but we have also streamlined our own workflow (emailing Word docs back and forth is pretty inefficient) and aligned it with the very stakeholders (product managers and software engineers) who implement 3MF specs.

Five years ago, I wrote: “I would *hope* that in ten, fifteen years max I'm seeing these standards in a web browser for free.” I’m proud and thankful to have that hope fulfilled :)

Strategy, sacrifices, and the margins your business needs

In an interview last week on The Amp Hour podcast, I put forth a provocative opinion: That it's okay to take a margin hit for strategic reasons.

Note: The full interview, which is too wide-ranging and generally entertaining to properly recap, is here.

Since I said the words below, I've been thinking more and more about the interchange and what my thoughts on margins are exactly. Here's a partial transcript (edited for legibility) of the relevant section, which begins at around 48:00:

SW: "I think there are a bunch of things that have to happen to make [stateside production of consumer electronics] feasible, and honestly one of them is that you have to change your expectation of your own standard of living, and you have to change your expectation for what your margin is going to be. Which is really, really tough. It's something that I [personally] feel deeply ambivalent about - like on the one hand I know I'm motivated by money, but also: Would my life actually be better taking a 40% margin, and then I'm able to see the person who's making my thing every couple months? Is that tradeoff worth it? I think in a lot of cases it actually is...

"I think that engineers and product companies jump to outsourcing stuff *way* too early, and there's a lot of units [of your product] that you could [personally] touch that would improve your ability to ship product and it would improve your customer experience in a lot of ways and it would just make your [social/professional] ecosystem a lot more healthy."

CG: "So when you say give up margins, do you mean give up margin at the beginning or give it up generally?"

SW: "I mean, both."

CG: "But it sounds like you're saying to give some of it up, and work out some of the wrinkles, and then eventually start to cost optimize and move to other facilities."

SW: "If that makes sense then sure. But I can say that the labor rates that we pay are way, way higher than we would pay going to China, and maybe that's okay. You know, you don't *need* your 60% [margins]. You can do things differently than the rest of the industry does... It's easy to read someone's blog post from some venture capital backed company talking about how they did it and think 'Okay, I'm gonna do it that way.' And thinking through those things critically is a good idea in general."

To start with, I should have added that sustainable (in the "financially sustainable" sense) businesses are *cool*, and I'm not recommending that anyone should put themselves in a position that'll eventually be untenable. I also recognize that when you've got a lot on your plate (as any entrepreneur does), there are a lot of things that end up being decided based on whatever the default is (i.e. whatever someone else has said publicly that they did).

That said: As David Ogilvy famously proclaimed, “The essence of strategy is sacrifice.” And just like you might give up a little extra money for your product to be higher quality, or for your customer to be a little happier, or for more efficient capital allocation, I think it's eminently reasonable to give up money for a more responsive supply chain, or one that isn't subject to big swings in international trade costs, or one that's easier for a resource-constrained team to manage.

As I said on the podcast (and as I've written here and on The Prepared extensively), I *like* doing business with people from other countries and cultures - and am particularly enchanted with China's culture and ability to execute all manner of undertakings in the physical world. But I'm a bit put off by the hardware scene's predisposition towards manufacturing *all* of our consumer electronics in the Pearl River Delta, and I would encourage anyone out there reading this to really consider the potential costs (both direct and indirect) of doing so.

Mini-review: Amazon Part Finder on iOS

A week or two ago, Jordan gave me a heads up about a new feature on Amazon's iOS app: A "Part Finder" feature that supposedly will identify fasteners using your camera's phone. The tech press is, predictably, excited about this: It's arguably augmented reality, and AR is hot right now, and most of the people writing about AR and Amazon probably don't own a pair of calipers, metric or inch thread gages, or a thread checker set. 

As one might expect, I was a bit skeptical to learn that it might take a journalist ten tries to even get the app to produce a result. So this morning I gave it a shot.

More or less at random, I opened one of my parts drawers and took out a selection of parts. I tried the following McMaster-Carr part numbers:

• 92196A581, 18-8 Stainless Steel Socket Head Screw 5/16"-18 Thread Size, 3/4" Long
• 91771A542, 18-8 Stainless Steel Phillips Flat Head Screws 1/4"-20 Thread Size, 1" Long
• 97517A025, Aluminum Blind Rivet with Steel Mandrel Domed Head, 1/8" Diameter, for 0.1880"-0.25" Material Thickness
• 97395A451, Dowel Pin 316 Stainless Steel, 1/8" Diameter, 3/4" Long
• 92196A194, 18-8 Stainless Steel Socket Head Screw 8-32 Thread Size, 1/2" Long
• 90895A029, 18-8 Stainless Steel Belleville Spring Lock Washer for 1/4" and M6 Socket Head Screws, 0.264" ID, 0.374" OD

TL;DR: I do not find this feature even a little bit useful. A few notes:

Lighting

My shop is lit mostly with task lights: Clamp-on fixtures with LED bulbs. This tends to cast shadows, which the Part Finder had a *really* hard time with; it took maybe 25 tries to get a single result. Worse yet, the app shows you the instructions every single time, making the photo process itself rather painful.

I next went outside and set up in full sun. This was even worse - the high contrast shadows seemed to throw the Part Finder off even more. Lastly I went inside in an area with a decent amount of indirect light, which worked much better - but still far from perfectly. 

In a few cases, the Part Finder couldn't even find the penny I was using, and once it claimed that it couldn't find the part. Some of these failures can fairly be blamed on my lighting, and some can probably be blamed on Amazon's image processing software. But the user experience sucked, and that falls squarely on Amazon's product team.

"Additional part details"

When I finally did get a match, the following message appeared:

Great, we have analyzed the photo and gathered specs. Please select the additional part details below:

The Part Finder seems to operate *only* on fastener diameter, length, and basic classification. It doesn't deal with head type or thread spec, and when I scanned the blind rivet the app prompted me to ask whether it was a "Drive Anchor, Post, Hex Bolt, or Blind Rivet." 

Some of this filtering may be useful to leave in the user's hands, but the range here calls into question exactly what "specs" the Part Finder has gathered. Which leads me to:

Thread spec

The Part Finder cannot distinguish thread pitches, nor can it tell the difference between metric and inch threads.

To me, this is a need-to-have feature for any fastener identifier. I almost *never* lack a method of determining a fastener's diameter and length (I bring a caliper with me if I know I'm doing much work at all), but my thread gages mostly stay in the shop. If Amazon wants to get serious about this, they *need* to include thread identification in the Part Finder.

Conclusion

As I've written before (see this and browse here), Amazon's total lack of product hierarchy puts them at a severe disadvantage when it comes to this kind of feature. The obvious counterpoint is McMaster-Carr, whose catalog structure is perfect for establishing the appropriate scope in a part identifier feature search. 

In other words: Partly, Amazon's Part Finder just sucks at identifying parts. But there are also institutional barriers in the way of Amazon ever being good at something like this, and (if AR gets better) I do still hold out hope for another player to tackle this problem.

Flying probe testing

This is an old video, but I only recently learned of it: The Public Radio's printed circuit board undergoing flying probe testing at the board house in 2015. 

Our design has changed quite a bit since then - the circular PCB was for some reason hard for us to get away from, but its current rounded-octagon shape is much more space efficient. Thanks to Chris @ Worthington Assembly for sending this over!

Looking for a freelance Ruby developer

Centerline Labs, the company I cofounded with Zach Dunham, is looking for a freelance Ruby developer to help improve our automated order & inventory management system.

I have written extensively (1, 2, 3) about our system ("The Coordinator") on this blog and on The Prepared's podcast (1, 2). It is the core of our ability to manufacture customized consumer electronics in a just-in-time manner. This is no small task, and we're proud to be doing something that enterprises and startups around the world struggle with.

We are currently looking for a freelance Ruby developer to help us improve and build additional integrations for tpr-coordinator, the backbone of our manufacturing operations. Applicants should have experience developing and maintaining web applications and should be comfortable deploying them on Heroku. Experience with Amazon's and Squarespace's commerce APIs, Shippo, and manufacturing systems in general are all pluses.

If this sounds like you, poke around tpr-coordinator on GitHub and then give me a holler here.

An interview with me!

Over the past few years, I've had the growing sensation that my public life is primarily focused on being a commentator or analyst rather than a practitioner. This is largely due to my role on The Prepared, but I'm sure is also influenced by my professional priorities and the fact that ultimately, I'm interested not only in any specific trade but in the broader saga of product development and business maturation.

And so it was just a delight to be asked to appear on 100 Product Managers, and a total pleasure to be interviewed by Suzanne Abate. Our conversation spanned not only The Prepared and The Public Radio, but also my time building high end custom bikes and robot doors. Throughout, we focused not only on the products being sold but also the products being used to make them - manufacturing tools, fixtures, and software. 

For the whole interview, head over to the 100 Product Managers site. Thanks to Suzanne for having me on!

The 3MF Beamlattice extension: A better way to describe lattice geometries

Today, the 3MF Consortium officially announced that its Beam Lattice extension has been ratified and released. And I'm proud to share that nTopology Element is the first piece of software to support it and that I (via nTopology) was a part of its development and adoption.

Brad and I first approached the 3MF Consortium about developing a lattice-specific file format almost two years ago. At the time, we had recently released nTopology's own LTCX file format, which we created and open-sourced in order to share files internally and with nTopology's early customers & research partners. LTCX is lightweight, portable, and high fidelity, and we wanted other companies in industrial 3D printing to have access to similar functionality. 

The 3MF consortium was an excellent fit for this, and late in 2016 I represented nTopology at our first 3MF meeting to discuss how LTCX could be adapted to meet the broader industry's needs.

For context: STL files have numerous well-documented issues, but they're particularly bad at representing lattice structures. STLs describe the surface of a part as a tessellated surface, which is (with large caveats) a fairly good general-purpose method of describing arbitrary geometry. But when you use STLs to describe lattices, you either end up with bad surface resolution or manifold/watertight issues, or both - and regardless, the file size is *enormous.*

The 3MF Beam Lattice extension solves this by describing lattices via their node and beam properties. This results in multiple orders of magnitude in file size reduction, while *improving* surface resolution. 3MF Beam Lattices live right alongside all of the other data that 3MF can handle: Not only meshes but materials, slices, etc. As a result, 3MF Beam Lattices are incredibly rich, and can be used for parts that comprise both lattices and solids with extremely high fidelity and portability.

Personally, the process of working on the 3MF Beam Lattice extension has been a pleasure, and I want to thank the entire 3MF Consortium for their dedication to our shared goals. In particular, Alex Oster, Scott White, Kris Iverson, Jordi Gonzalez, Mark Forsyth, Mike Facello, and Kurt Renap brought both enthusiasm and experience that was essential to the specification. I also want to commend the team at Netfabb for introducing 3MF Beam Lattice support with incredible speed, and for driving the development of the 3MF open source implementation. 

3MF has some great other projects in the works, and I can't wait to share them as well. For more information, visit the 3MF Consortium's website.

Looking for a laser marking vendor

An open RFQ:

I am looking for a vendor who can perform laser marking services on stainless steel parts. This is a customization that we offer for The Public Radio as seen below; we purchase the laser marking in MOQs of 100+ and have an estimated total usage of 2500 this year. 

If you do laser marking or know someone who does, please get in touch here. Note, my preference is to find a supplier that's within 1 day of ground shipping from New York City. 

The Public Radio's assembly & fulfillment processes

In preparation for our full rollout at our contract manufacturer, Worthington Assembly, I spent a bunch of time honing The Public Radio's mechanical assembly & fulfillment processes - organizing the workspace, testing each error check, scanning random shampoo bottles to make sure a weird barcode doesn't screw things up. The processes and tools we've developed will inevitably evolve yet, but at the current setup is solid.

So, video documentation!

Mechanical assembly

We begin by taking in printed circuit board assemblies and making mechanical assemblies. We typically do this in batches of 18, and once a batch is done they're scanned into inventory and put in a tray on a shelf:

Obviously, the mechanical assembly step is *not* done just in time; a mechanical assembly might sit on the shelf for a day or a week before its time comes. Then comes:

Order fulfillment

This is where the real engineering comes in. Here we have multiple systems - Tulip, our order management database, and a bunch of scripts & custom hardware to round things off.

So, two videos!

I gotta say - it's a pretty cool process :)

The Public Radio's inventory dashboard

Every few weeks since The Public Radio went into production last summer, I've been struck with a pressing - and often urgent - need for some new and/or improved business or operations tool. One recent occurrence was particularly pressing, as we've had a few (ultimately manageable) inventory issues and are facing continually more complex logistics: We needed to know more about what parts and assemblies we had on hand at any given moment.

The Public Radio is assembled, programmed, and fulfilled from two locations: Worthington Assembly (our contract manufacturer and PCBA house in Massachusetts) and our corporate headquarters (a.k.a. my basement). The materials we use come from China (our speaker, potentiometer, and a few other electronic and mechanical components), Taiwan (our lid), a few stock component distributors in the US (our jar and a few SMT components), and a few custom US manufacturers (our box, and for radio station orders some laser marking on the lids). Some of these parts are used by Worthington to make our PCBAs; some are used to make our mechanical assembly; the rest are used to fulfill orders.

What we need to track is that conversion chain: Parts to PCB assembly, parts (and PCBA) to mechanical assembly, parts (and mechanical assembly) out the door.

For my current purposes, there are three events in this chain: Receive inventory, Create mechanical assembly, Fulfill order. (Note that I'm ignoring the components in the PCBA at this stage, and instead simply tracking the PCBA at the assembly level.) The first of these happens every few months and can be done manually, but the latter two will happen something like ten thousand times this year; it simply needs to be automated.

Luckily, our entire manufacturing process runs online. Our order database runs on Heroku, and each manufacturing cell needs internet connectivity in order to process orders. As a result, I had a perfect opportunity to build a quick and dirty cloud based inventory dashboard.

As these things go, I ended up only having a few days to build the system, and I was pretty sure that the early data would need a good amount of ad hoc massaging (i.e. I would need to use a database that could be manually edited whenever inventory needed to be reconciled). So, Google sheets it was :) 

With manual edits being sufficient for receiving (and reconciling) inventory, what I still needed was a way to record inventory events and then send them to Google. The result was two scripts:

• The first records inventory events in a CSV that's stored locally to the event (on the manufacturing cell's Raspberry Pi). Each record contains a timestamp, the event type, and some username (either a string that's passed in as an argument or the Pi's hostname, which is unique).
• The second script parses the CSV line by line and updates our Google sheet for each inventory event. 

Our Google spreadsheet contains many sheets: One for each inventory part/assembly, and a dashboard which sums up the current stock on hand for all parts/assemblies. As a result, our second script needs to make multiple updates for each line in the CSV. This took a bit of fiddling to get right (note: appending rows one by one is slow), but now the whole process only takes a few seconds, even with 1-200 events (which is about what one manufacturing cell can do in a day) to update.

The result is simple, but powerful: A real time (or nearly real time - the second script currently runs once a day on a cron task, as there's really no need for more granular data) dashboard showing on hand inventory of 8 different items:

Critical levels all over the place. But also, just in time!

This will eventually get a few updates. The most important one is to track inventory of the two manufacturing locations (Massachusetts and Brooklyn) independently - a relatively easy task. I'll probably also create an interface for reconciling inventory, which is currently done by manually adding reconcile rows (instead it could probably be done by removing all old inventory events and adding a single "starting inventory" row). Lastly, there's a high likelihood that the work being done in the inventory scripts ends up being integrated directly into Tulip, which is already handling the vast majority of our operations.

But even without those improvements, these two simple scripts form an incredible tool. 

A year of commits

For years - through at least my three most recent jobs - I've wanted to write more software. The desire has been vague at best, driven partly by curiosity, partly by FOMO, and partly by wanting to be more personally effective - to be able to do more stuff. My jobs have supported me in broad strokes, but in my experience real progress occurs in times of real need, and ultimately my (nonexistent) skills as a software engineer simply haven't been on anyone's critical path.

Until 2017.

2017 saw a totally new pattern of work. At nTopology, I spent much of the year building an integration between Element and Abaqus, using the workflow to optimize material use on metal 3D printed lattice structures. This laid a good baseline of commits, mostly during weekdays. 

Later in the year, as The Public Radio's operations spun up, my commit frequency jumped significantly. Some days see dozens of commits, as I pushed in-development code from my laptop onto one of the Raspberry Pis we use to fulfill orders. This is a somewhat weird process, and I'll admit that it has the effect of juking my stats. But I'll also note that a lot of these commits occurred on weekends, which feels like it should balance my credibility out a bit.

It's yet to be seen if 2018 continues this trend. I'm still no developer, and the level of complexity that I'm able to take on myself will probably remain pretty low. But it's nice to see some progress :)

Logistics

Over the past few years I've become fond of defining manufacturing as follows:

Manufacturing is just people
doing things
again and again
in a predictable way.

One of the main reasons The Public Radio exists today is that it's an effective instrument for honing our skills at manufacturing engineering & management. We've seen our volumes jump an order of magnitude a few times now, and each time our manufacturing skills have needed to evolve. But to a large extent those evolutions have been incremental, whereas upgrades to our logistics have followed a step function.

As suggested in the definition above, manufacturing is (perhaps more than most people believe) mostly about people; logistics is similar. But logistics, being transient, requires trust among parties that will never fully get to know one another - and that, by their very nature, are separated at great distances.

Two TPR shipments make their way into the US.

Logistics also requires tools that aren't common in a shop (maps, week-of-year calendars) and is subject to delays that are often incomprehensible in their complexity (port labor disputes, weather, harmonized tariff schedules). And worse yet, logistics companies often impose strict firewalls between customer service and operations - making one's own self-eduction all the more frustrating.

That said, the magic of it - planning out a three month long production cycle and seeing it come together at the end - is remarkable. 

(He wrote, as he waited for packages to show up from around the world...)

A good, old, tool

In early 2009 I purchased a used (and nonfunctional, at the time) Abene VHF3. It was a rite of passage: The machine was over 40 years old when I bought it, and at the time my knowledge of machine tools was tenuous at best. But I wanted the challenge, and I had decent intel that the VHF3 was a good machine, and (perhaps contrary to popular belief) the internet provides thorough documentation for even obscure machine tools. 

In retrospect, the experience was transformational. In addition to the mill itself (and in addition to the way wipers, bearings, oil, tooling, and other sundries needed to restore a machine that's probably been sitting for a decade), I purchased gallons upon gallons of kerosene. Stripping the machine down to its core components - and stripping every bit of caked up lubricant off with kerosene - took about a month, during which time I learned not only how the machine worked but also how to navigate a good portion of industry as a whole. Simply purchasing the right lubricant would require a half dozen phone calls, a purchase order, and a drive up the island to some sleepy distributor. Replacement parts aren't just unavailable; they're totally undocumented, lost in some Swedish file cabinet. 

On the other hand, this was a tool; a machine. Its complexity only went a few layers deep, and given a week or two almost any problem I came up against was tractable. And the more I worked on the mill, the more I realized that it would eventually make the things needed to improve itself.

It was a beautiful machine. It was strong, versatile, and capable of precision work. In the right hands it would have put in another century of service. But ultimately, mine were not the right hands. I sold it a few years later, and since then have not had a shop space that would support a tool of its caliber. 

I'm not sure when I'll own another tool like it. It was a treat: Empowering, challenging, terrifying at times. I recommend the experience thoroughly.