Notes from a cardboard factory in Taiwan

In October, I visited a Taiwanese cardboard factory in Taiwan with Brilliant Bicycles. A few notes:

• Paper comes in, finished cardboard boxes go out.
• The paper rolls come in various shapes, sizes, and presumably thicknesses, but they're generally *huge* and brown. 
• The whole place, but especially the sheet assembly line, was really loud - one of the loudest places I've ever been. 
• There are three big portions of the factory: the sheet assembly area, the box printing & slotting area, and *lot* of cardboard sheets stacked up around the factory.
  • The sheet assembly area takes brown paper rolls, each of which weigh upwards of a ton, and transform them into cardboard sheets. The machine that does this is huge - probably 20 feet wide and maybe 200 feet long. The cardboard we saw getting made was 5-ply, which has three flat sheets (the two faces and one in the middle; think the slices of bread in a club sandwich) and two corrugated sheets (think the bacon, turkey, etc in the same sandwich). Presumably the exact specifications of the corrugation (e.g. its thickness) affect the structural properties of the cardboard; I think the sheets we saw getting made had thicker corrugations (maybe for the inside of the box) on one side and thinner on the other. Anyway, the sheet assembly machine takes five big rolls of paper and spools them out together, corrugating them as necessary and applying adhesive between the sheets. Then it applies pressure and feeds the whole continuous sheet through a pair of heated plates, which help cure the adhesive. Then it trims the sides off, slots the sheet in half lengthwise, and then chops each half into sheets of the appropriate size for what they're making.
  • Then the sheets are stored in stacks around the shop. There were a *lot* of stacks of cardboard sheets - a large portion of the factory was just storage.
  • The box printing and slotting section was a bit more disperse. There were a bunch of machines to perform these steps; some did one or the other, and some (including the one that Brilliant's boxes were made on) do both. The printing is done with big, colorful, silicone stamps. Most of these were presumably made out-of-house, but there were some (like the "no knives" and "fragile" stamps) which seemed to be standard and were being assembled onto clear plastic sheets (mylar?) by people in one area of the shop. We didn't see steel rule slotting dies being made (perhaps they were outsourced), but they were stored all over the shop. There were a bunch of varieties of those - some were used in manual die presses, and others were loaded into slotting machines and then operated automatically.
• A bunch of work was put into aligning the stamp and dies in the machine. If I were to guess, I'd say 5-10 sample parts were made before the alignment came out right.
• Once the alignment was right, the boxes came out really quickly. We were only having a few hundred boxes made; I'd guess they'd be done within a few hours. 

Kane + Jacob + a lot of paper.

Each one of these rolls weighs between 1000 and 2000 lbs.

Jacob + the time clock.

April and our tour guide (presumably the shop manager).

The main cardboard assembly machine. Raw paper goes in a bunch of big slots on the left, and is spooled to the right, being corrugated and adhered to each other in the process.

Brilliant's rubber stamp. Almost the entire box is inked, with only the logo staying white.

The box stamping & slotting machine.

Jacob & Allan check the stamp out.

The small inner box's proof - a clear overlay that shows what the box will look like.

Adam & Jacob with the box machine.

Adam.

Some rubber stamps around the shop.

The place was huge - one of the single biggest unobstructed spaces I've ever been in:

View fullsize

The cardboard printing & slotting machine was really cool. There's nothing like a black box that takes raw material in and spits out a useable object on the other side:

Seeing general purpose manufacturing is always a lot of fun. Cardboard boxes are used *everywhere.* Having seen how they're made gives valuable perspective on something that's on basically any BOM.

Notes from a tire factory in Taiwan

In October, I visited a Taiwanese tire factory in Taiwan with Brilliant Bicycles. I didn't take many photos, but do have a few notes:

• The factory made a bunch of solid tires for forklifts and other equipment in the front.
• In the back were two sides: One for bicycles, the other for scooters and other larger pneumatic applications.
• The shop did *not* make their own nylon cord for tire casings; they purchased that from a vendor. The cord comes in a wide sheet, is cut on a bias and put together into a long strip (video below).
• They also did not make their own kevlar (foldable) tire beads. Apparently all of the Taiwanese tire shops (there are a few) purchase these from one single supplier.
• They do make their own steel beads. Wire is straightened, doubled up, and wrapped into a loop with a bunch of unvulcanized rubber and bound together into a single unit.
• The machines they use to assemble the tires - casing, bead, and tread - are *really* cool. There's a decent video of the process here.
• Tread molds seemed eminently reasonable - they only cost a few thousand dollars. The uncured and relatively shapeless tires are put into molds, where pressure and heat form the shape and cure the rubber.

Here's a photo of the front of the shop, where solid tires were being stored:

View fullsize

And here's a video of nylon cord being cut on a bias and spliced together:

And, two random photos: One of a magazine in the factory's conference room, and one from the restaurant that the (extremely eager) factory manager took us to for lunch:

A few notes from Boeing

Last week I visited Boeing's Everett, WA factory, and saw the 747, 777, and 787 lines. I couldn't take pictures and had no time for notes, but I did have a few observations/factoids that I wanted to record:

• The tour was *not* free - it cost $16 if you reserved ahead of time, which I did. They do tours every hour or two, and my group filled two buses - probably 50 or 60 people in total.
• The Everett facility is the largest building in the world by volume. Our tour guide said that last year someone in China opened a mall that's bigger by floor space, but that Boeing is expanding to get that title back too. I assume that that is a) hyperbole, and b) true nonetheless.
• The facility employs something like 41,000 people on a daily basis. Fuck.
• The 787 is pretty cool. It's made mostly of composites - presumably carbon fiber + epoxy + some mix of aluminum, titanium, ceramics, and other metals (there's a good explainer on its construction here). Its wings can flex something like 21 feet vertically during flight, giving it a lot more vertical compliance and presumably making the ride a lot smoother.
• The 787 is significantly smaller than the 747. Boeing is proud of this, and scoffs at the Airbus A380's enormous size. They're trying to push away from the hub-and-spoke model, and encourage airlines to purchase more planes, which will fly a wide variety of routes at relatively high usage rates. Since it's really hard for airlines to make money when flights aren't full, this strategy seems to make sense - it's *really* hard to fill an A380 (they fit 853 passengers!), so why not just buy two or three 787s (242 to 335 passengers) and fly them fully booked?
• Planes are purchased without engines or seats; those need to be bought separately, and Boeing will install them at the end of their production line.
• I asked what engines Boeing uses, and was told that while the makers vary by model, Boeing typically uses a mix of GE, Rolls Royce, Pratt & Whitney, and CFM (a collaboration between GE and Snecma).
• Interestingly, Pratt & Whitney (now a subsidiary of United Technologies), Boeing, and United Airlines all once shared a parent company: the United Aircraft and Transport Corporation. That company was disbanded in 1934.

I would *love* to know more about the 787 supply chain. The Japanese company that makes the prepreg, Toray, looks interesting. And obviously both Mitsubishi and Kawasaki (both big suppliers for Boeing) are fascinating as well. 

What's with Hoffman Estates, IL?

I've been searching through machine tool manufacturers today, and stumbled across this town of 51,000 residents twice. What's going on here?

View fullsize

DMG Mori, which is headquartered in Hoffman, doesn't even rank big enough to show up. Ditto Renishaw USA.

It could be that this is just a manufacturing-heavy suburb of Chicago; I guess the name is just throwing me. Anyway, I'd be interested in visiting.

You're buying services

A thought occurred to me this morning:

As you move up the value stream - away from on-the-shelf packaged goods and towards raw materials - the fundamental nature of what you're buying changes dramatically.

At the retail level, it's relatively easy to infer the correlation between price and brand language (which may or may not be true product features). 

The next step up - buying commodity manufactured parts - is in many ways the most transparent. Here the brand language has been removed, and price correlates with product features themselves. Parts that require more, more complex features are more expensive to purchase.

But as you move up higher still - to the point where custom designed parts are being converted from raw material - pricing moves away from features and towards services. Whether or not you're cognizant of it, the price that you pay has moved to a time-and-materials model. Even though you may ask for a part count: You're buying services.

Boxes

Cardboard boxes being made outside Taichung, Taiwan. A serious hat-tip to Kane and Adam at Brilliant Bicycles for taking me here. More reports soon :)

Public Radio changes

The Public Radio has moved a little slowly over the past few months. Zach has been busy, and I've got too much on my plate, and scheduling has been tricky. But we *have* made progress, and I'm here to tell you about it.

After the last hardware round, we spent an hour and developed a feature wishlist. It included:

• No vias under battery clips
• Larger thru hole pads & holes
• Reverse polarity protection on the battery line
• Schematic file is legible
• Values on all parts are accurate
• Redraw FM IC package to reflect accurate dimensions
• Smaller lid ground wire hole
• Speaker wires don't cross
• Orient all SMT parts in the same direction
• Reorient ESD
• Add serial number & tuning frequency markings
• BOM shows manufacturer part number only

All of these (plus a few more technical ones - see our GitHub repository for the current designs) have been completed. A few stragglers have yet to be implemented:

• Boards have white solder mask & black legend
• Schematic is 100% consistent with BOM

I like the way the new design looks. Eagle is *not* my favorite piece of design software, but board layout is fun - and I'm really proud of our current design.

View fullsize

On Sunday evening, we ordered a batch of 10 of these from Advanced Circuits. Advanced is expensive, but they turn the parts around quickly; ours should ship today. Meanwhile, our custom antennas departed (two or three weeks late, but whatever) Hong Kong yesterday. With any luck, the antennas will show up within a day or two of the boards - giving us time to assemble a few PCBs and then quickly put the radios into service. 

In the meantime, Zach hacked an awesome way to tune the radios using an Arduino. It should save us a lot of time, and will be *key* when we're shipping these things across the country.

So. Movement. Happening. 

Public Radio Progress

The last few days have been a bit of a whirlwind. 

First, we visited Todd to discuss some circuit layout concerns. On the way over we nabbed a cheap radio at RadioShack, and took it apart with Todd.. 

What we found was really interesting. The device uses a different chip than we had been prototyping with, and as a result their circuit is *much* simpler. Our chip (Silicon Labs' Si4703) requires digital tuning, which we were accomplishing with a microcontroller. But RadioShack was using the Si4822, which is mechanically tuned. The result is drastic from a circuit standpoint - they no longer need a voltage regulator, or a microcontroller, or a bunch of additional passive components on the board. They're also in a better position re: noise on the board: an MCU (and a voltage regulator too) will produce a bunch of noise that might get on the antenna, affecting reception. RadioShack doesn't need to worry about this at all.

Over the past few weeks, we've been looking for just a breakthrough like this - a way of rethinking the problems we were dealing with. Our issue was that we were stuck conceiving this as an Arduino-descended product, which it doesn't need to be. 

We also made a bit of progress reengineering the hardware layout of the board. Our new design (this is just the board layout, not the circuit schematic) will be significantly easier to assemble, program & service.

We're also pretty sure we can shorten the antenna *significantly.* The mockup here shows a custom solid stainless steel antenna, 50mm long. I rather like it, and it's designed such that a longer telescopic antenna can be swapped in easily. 

Our PCB will now be circular, and we're soldering the potentiometer and speaker directly to the board. There will be a bunch of SMT stuff on the top (same side as the speaker & pot) and a battery back and trimpot (not shown) on the bottom.

View fullsize

View fullsize

I *really* like where this is headed. More updates soon. 

Public Radio progress

The latest rev (v1.1) of The Public Radio:

The current state is: 

• Arduino Pro Mini
• Si4703 breakout
• Mono audio amp breakout
• VREG
• A bunch of screw terminals

The next rev will be a little PCB that the Pro Mini can mount to, and on the back will be all discrete components comprising our own FM tuner & amplifier. This will reduce the cost of the assembly by a *lot* (these Sparkfun boards are easy to work with but silly expensive), and will allow us to play with the circuits a bit. 

The incredible thing is really how cheap the Pro Mini is. These things cost under $9 (cheaper on eBay) and have a lot more features than we'd ever need. Eventually we'll strip that off too and run the whole thing off of an ATTiny (or similar) but for now there's no point in getting ahead of ourselves - the Pro Mini is cheap enough and requires very little work on our part to get our MVP out quickly.

We should have the updated circuit done tomorrow and will be working on board layouts over the weekend :)

Symbols

 :)

via everythingradio.com.

View fullsize

View fullsize

Batteries

From Wired's (great) recent piece, "How Apple’s Lightning-Plug Guru Reinvented Square’s Card Reader."

But ditching the battery meant more than saving space. It was also a huge step towards that simplicity Dorogusker set out to achieve from the start. “With a battery, you have to somehow connect the battery to your circuit board,” he says. “You do that with two little wires. They have to be cut to length, stripped on both ends, tinned on both ends and hand soldered to the battery and then to the board. It’s a huge amount of labor.”

In my robot door life, we were lucky enough to have someone aboard to warn us of the physical danger of lithium ion batteries; we ended up finding some very powerful lead acid batteries to do the trick instead. But we spent a *lot* of time configuring our battery pack, which was a PITA to assemble and mount. Wires suck, and batteries often require them. Avoid if possible.

IRL Crowdfunding

Every few days someone asks me about one of the projects I'm working on. Most often they've seen something I've put up on Instagram or Facebook, but didn't have enough context to really understand what it is I'm working on or even whether it's for sale.

Most of my projects will eventually be crowdfunded, and I expect to be able to convert some of the folks who I've talked to during development. But I had a thought the other day: Why not start taking orders immediately, whenever someone asks? 

The thought is this: If you ask me about The Public Radio, I'll give you my little pitch and then (assuming you've acted enthusiastic) ask for $20, cash, now. I'll then whip out my phone and email you an informal receipt, and will deliver you a v1.0 when it ships (probably the pre-Kickstarter version). 

This would help me in a few ways. First, it locks in a customer. Second, it lets me know whether I'm actually onto something - if everyone says "no," then maybe I should pivot. Third, it gives me a little cash to help keep the project moving forward. And I can be pretty sure that you'll ask me about the project status in the future, which is the most thrilling parts of building a product like this.

I think this is a decent idea. If you're reading this, ask me about what I've been working on the next time you see me - we'll see how it works :)