Twenty-four hours before our launch, Kickstarter approves our project, and Shawn and I let out a collective sigh of relief.  We’ve been working steadily towards this launch for months now, but even by kickstarter standards, it’s not exactly traditional.  Most Kickstarter projects dealing with products and technology are very straightforward–”we designed and prototyped this thing, if enough people order them, we can make lots of them and ship them to our backers.”  Our project doesn’t quite fit that mold–we’re raising money to build a new type of machine that makes solar panels.  There aren’t a lot of people that want to buy machines that make solar panels, and we spent a lot of time thinking and revising rewards that we could offer for our backers that would be closely tied to our project, but would also be a more compelling fit into people’s lives than a large piece of machinery.

After a lot of discussion, we settled on kits for backers to make their own solar panels, using the same techniques we’d developed as we built the pocket factory.  It felt right–the kits were connected to our projects; a DIY version of the same process that we were working to automate and streamline.  Still, it wasn’t quite the same story as most other projects on Kickstarter, and we spent a nervous weekend, heart in our throats, waiting for Kickstarter to get back to us.  And now it’s T-1 day, and all we have to do is hit that ‘Launch’ button and tell everyone we know.

I reckon this next bit is gonna be interesting.

(photo by shawn)

“Where is your umbrella?” he asked, friendly eyes fixed on me from beneath a giant canopy emblazoned with the 7/11 logo.  He’s shirtless and soaked, and he has a point.  A ruthless series of cyclones has dumped rain on Manila for three weeks straight.  Half of the city is underwater.  Rainwater is trickling down the insides of my jeans. “I don’t have an umbrella,” I tell him.  ”I don’t like them.”  He squints at me, and I try to imagine what he sees.  A barefoot guy dripping rain in the middle of the street, heavy sports duffel slung over my shoulder.  Yiddish mothers take a look at me and bring me a bowl of soup.  Fashionable girls cross to the other side of the street, avoiding eye contact.  ”You don’t like umbrellas,” he repeats, eyeing my rainsoaked clothes.  ”That’s too bad.”

But honestly, I don’t care about my clothes.  I’ll dry out soon enough.  I care about the forty pounds of aluminum framing, linear actuators, electronics and solar cells inside the shoulder bag.  It’s the latest iteration of the pocket factory.  The guys in Hong Kong designed and built it, and using a special soldering technique we worked out in the mantis lab, I just got the first soldered, working panels coming off it thirty minutes ago.  In an hour, I’ll be boarding a plane to Hong Kong to bring the factory back to the Haddock team (plane tickets between manila and Hong Kong are cheaper and faster than sending a fedex, and we’ve quickly grown accustomed to bouncing between cities and crashing on each other’s couches).  But right now, it’s 4:00 in the morning and I’m very wet.  Everything in the factory is waterproof, I tell myself, or it can dry out. I turn towards the main road to catch a taxi, and a fountain of water pours down my collar from a neaby gutter.  I say that bit about how waterproof everything is, again, and whisper a quick prayer to the invention gods.

The trick to getting this version of the pocket factory to work was a technique that we call Twice-Baked Potatos.  In Hong Kong, Shawn, Samtim and Angus built a new, simplified mechanism to place and build up solar cells.  Our first mechanism is a traditional placement machine, using a vacuum to pick and place solettes onto a backing and pressurized syringes to add little dabs of solderpaste between the solettes.  This new design holds the backing upside-down, covers it with a spray-on glue, and sticks the solettes to the board upside-down.  The design is great–it’s much simpler and smaller than the pocket factory we’re working on at the Mantis lab, but until an hour ago, it wasn’t complete.  The problem with this upside-down placement is that it’s tricky to get solderpaste onto the solettes.  Solderpaste is…well, what it sounds like.  It’s a thick mixture of solder particles and flux, and when you heat it up, it melts and holds the pads of different electronic components together.  We use a special lead-free, low-temperature solderpaste to electrically connect the top of one solette to the bottom of the next.

In our main machine, we dispense the solderpaste with a pneumatic syringe, pushing solderpaste out a needle with a timed burst of high-pressure air.  To do that with this machine, we’d have to build a special mechanism that can move the needle over the solettes to squirt on the solderpaste, and then get back out of the way so we can push the next solette onto the backing.  It’s not a simple design–in fact, it would be much more complicated than the rest of the design.So we tried something different.  A few weeks ago, we worked out the Twice Baked Potato method.  Solderpaste has three states:  paste, cracker and reflowed.  When it’s a paste, it’s smushy and sticky.  If you put paste on solettes and then stack them up, they’ll stick to one another in an unfortunate, lead-free menage-a-toi.  If you heat the solderpaste up to 150 degrees C, it reflows–the flux evaporates and the solder melts and flows together, forming a small spherical lump of solder.  The thing is, once you’ve reflowed a solette, all its flux is gone, and if you try to reflow the solder again, it doesn’t make electrical connections as reliably.  The trick is to hit it halfway.  At ~100C, the solderpaste hardens to the touch, but there’s still some flux trapped inside.  We can stencil on a thin layer of solderpaste onto a solette, harden it by baking it briefly in a toaster oven, and we’re left with a stack of thin solettes than, when you heat them up, will form a solder joint to anything that’s touching them.  Bley Joel made fifty of these once-baked solettes today, and tonight I worked through a bunch of them, loading them into the hopper, placing panels and reflowing the panels in the toaster to solder everything together.  I’m starting to get complete, working panels where twelve solettes solder together nicely, but I’m stuck on the connection between the solettes and the fiberglass circuit board I use as a backing.  I’m just not getting a 100% reliable connection–in fact, I’m lucky to get one good connection between a the solettes and circuit board.

I ruminate on this during the flight, flipping through reflow profiles and thinking about how to get the perfect reflow.  Waiting for my baggage at the Hong Kong airport, there’s no doubt which bag is mine.  It’s the one sitting on the conveyor belt in a growing puddle of rainwater.  The one holding an itty bitty factory that makes solar panels.

In Manila, the Mantis team is similarly engaged.  Bley has modded a toaster to act as a temperature-controlled reflow oven, and we spent hours finding the perfect heating profile to solder solettes together.  Panels are cranking out, and we’re reflowing and gooping them, testing different settings and going over the panels with a multimeter, seeing which connections are solid and which ones aren’t reliable.

Shawn and I decided that we need a benchmark to decide which of the ideas we’ll build out into our pilot machine, so we’re racing to build a prototype that can automatically make one hundred perfect solar panels.  That’s a tall order–each solar panel has twelve cells, and it means that we have to feed, pick up, apply solder and place 1,200 cells, each to within a quarter of a millimeter of accuracy.  So we get to work.

First off, in my lab in Manila, Eilrem slows the machine way down.  For our production machine, we need to place a solar cell every second, but right now, accuracy and reliability is more important than speed.  So we slow it down and Eilrem whips up a clever procedure to align each cell perfectly, taking out any error from the backlash in our gears.  We’re using suction to pull a solar cell off a stack and hold it place while we move it over to the panel we’re assembling, line it up, and then drop it in the right spot.  While we’re carrying the cell over, we need to put a little dab of solderpaste onto the top of the cell.  Once we’ve placed all the cells, we’ll melt the solderpaste in an oven, and that will make an electrical and mechanical connection between cells.  Samtim got us some special, low-temperature, lead-free solderpaste that will be easier and more forgiving to heat up and melt.

So a couple days later, we’re cranking along in Manila, rigging up our factory to add a high-powered halogen light over a conveyor to automatically heat and melt the solder on our solar panels once we’ve finished placing them, and we get a video from the team in Hong Kong.  Angus and Samtim have spent the last week designing a cell placing mechanism that puts cells upside-down onto a circuit board, using a spray-on glue to hold the cells in place.  It’s simple, small and smooth.  Our factory weighs about 100 pounds, but this little mechanism can fit into a backpack.  And it’s beautiful to watch it place a panel–it does it smoothly, popping out a panel after thirty seconds.  It’s not done yet–there’s nothing to electronically connect the cells together yet, but over in Manila, it pushes us along.

Life in Manila, though, has turned wet.  Typhoon Gener plowed through, followed by Typhoon Saolo, bringing torrential rains every day and sending the rivers burbling over their banks, and washing out my favorite karaoke joint for a couple days.  When it rains like this, life is unpredictable.  Streets fill with water three feet deep.  Violent cloudbursts trap us inside for an extra hour here, an hour there.  Traffic backs up for hours on the freeways.  Eilrem was planning to grab a couple components saturday morning and get in early.  At nine PM, I’m exhausted from two consecutive all-nighters and I’m ready to lock up and go home, and in a flash of lightning, Eilrem is in the doorway, dripping from the storm.  Undeterred, he pulls a chair up to the factory and cranks away for a marathon fourteen hours while the world goes to hell outside.

In the morning, we’re both staggering.  But we’ve got a temperature-controlled oven hooked up to the factory, and to the sound of roosters, Eilrem presses the button that tells the machine to make its first fully-functional, fully-automatic panel.  Five minutes later, it comes off, soldered together.  It’s not perfect–one joint is weak, giving an intermittent connection.  But it’s going to work.  That’s what we’re here for.

I rented the Mantis’ first lab space in Barangay Barangka, Marikina.  Barangka is a lovely village very near two of the major universities in Manila, Ateneo and University of the Philippines.  I’d stayed in this neighborhood on a previous trip to Manila and liked it, so I was delighted when I found a ~600 square foot space that would be perfect for a first lab space.

I rented the space just six days after I landed in Manila, and I was feeling very pleased with myself as I brought in the first set of electronic test equipment and started setting up the lab.  It was a lovely Monday afternoon, and I busied myself puttering around the lab, plugging in oscilloscopes, putting up tool racks and making the lab functional.  I sat down on one of the couches and started making a list of the various things I needed to buy for the space.  Soldering iron, broom, toilet paper, whiteboard…everything I’d need to work here.  I was interrupted by an earsplitting bang.  A tropical thunderstorm had poured in over the mountains, and the sky suddenly turned black with rain.  Brilliant flashes of lightning lit up the sky over Manila, the thunder rattling the windows.  A wild wind picked up, and I ran through the lab, shutting the windows to keep the rain out.  The storm intensified and turned into a deluge, the rain pounding a staccato against the door.  Anyone setting a foot outside would be soaked instantly.  Unable to do anything else, I sat down to continue my list, and it was then that I noticed the water coming in under the door.  The lab is up on the third floor, so it’s fairly floodproof, but there was a small gap in the door’s weatherstripping, and a rivulet of water, driven by the wind, started streaming through the door, pooling and then draining out a hole someone had thoughtfully installed in the corner.  Marooned on my sofa, I looked down at the notebook in my lap and added ‘mop’ to my list.

The space is really quite nice.  Here’s the view from the balcony, with another thunderstorm sweeping in.

It’s a converted boardinghouse, so there’s a large room with four smaller rooms off to the side.  The large room is the common workspace, and I’m turning each of the small rooms into a specialized work area.  One will house the pocket factory once it arrives next week, one is a mechanical engineering/construction area, one is a quiet work area, and I haven’t figured out what to do with the last one, yet.

The beginnings of the tool area.  Everything you need to build great stuff!

The electronics bench after a debugging session.

And, of course, the Mantis’ first crop.  Sprouting melons, herbs and okra for the balcony.

Those are the new digs.  Mantis Shrimp Invention has its first home!

What I love about this tool is the minimalism.  For the uninitiated, a hotwire cutter runs electrical current through a thin piece of wire, heating it up so that it can melt through plastic foams (like styrofoam or polyurethane insulation foams).  A good foamcutter can cut through foam like a knife through butter, giving you a lot of power to quickly sculpt shapes in a cheap, lightweight material.  Designers use this all the time to mock up new ideas that they can pick up, hold in their hands, and generally get a feel for a new project they’re working on.

I found this cutter at a corner store in Hong Kong, and I was attracted by its extreme simplicity.  There are no moving parts–it’s just molded plastic with a couple metal inserts, but it’s surprisingly sturdy.  The cutting wire itself is crimped music wire, and it’s something I can easily replace myself with cheap materials, if it ever breaks.  The whole thing is powered by two D-cell batteries.

There are no moving parts–not even a switch or button.  It’s extremely simple, I can see what everything does at a glance, and it just works.  My kind of tool.

The room is small, with scuffed walls, torn carpet, with many of the styrofoam tiles missing from the ceiling, exposing the wires and pipes above. The center of the room is dominated by a large card table stacked high with taped cardboard packets, plastic chairs crammed in along the walls. In the back room, the man’s gloved hands never stop moving. The rest of the office is empty, the back hallways dark. Outside, rain is pouring down on Shenzhen.

Mr Fujimoto holds out his hand, and Ms. Wu casually places a knife into it, blade facing away from Fujimoto. The movement is silent, rehearsed. They’ve done this countless times. Fujimoto slides the blade under the tape of one of the cardboard packets, slicing the cover free. He places the knife to the side, reaches into the packet and draws out a silicon wafer.

I’ve never seen solar cells packed for distribution before. The first thing that strikes me is how incredibly dense the stuff is. A box containing 400W of silicon is about the size of a box of hot pockets. The table is stacked with boxes. Leaning back in his chair, Fujimoto describes a hidden world in solar.

These solar cells all come from semiconductor foundries in Taiwan. The foundries make silicon crystals, slice them into wafers, dope them with semiconductors and print conductive traces onto the wafers. The production process is pretty good, and 99% of the cells come out perfect. These are called grade A cells, and they go straight to the automated solar production lines that crank out the modules for rooftops and solar power plants. These cells go for $.65/watt. But one percent of the cells have a small flaw. Perhaps there’s a chip, or an uneven coloration, or a burn mark from the processing. These are called grade B cells, and they can’t go into the automated production lines. They’re not total scrap–they still produce 4W of power per cell, but there’s enough variation between cells that they wouldn’t be perfectly matched. They’re functional, but they can’t be sold through traditional channels, and this is where people like Mr. Fujimoto come in.

Fujimoto runs a clearinghouse for grade B cells. The world produces 30GW of solar a year, and the ~1% defect rate means that 300MW of solar cells get sold through clearinghouses to people who don’t care about the cosmetic defects. That small defect knocks 50% off the price of the panel, dropping it down to an incredible $.32/W. That’s not quite the cheapest solar in the world–the really beat-up stuff, with big dings, goes for ~$.25/W. All of these numbers feel surreal compared to the Solyndra-tinted world of US solar, where companies strive to reach the holy grail of $1/W silicon cells. On the other end of the world, we couldn’t feel farther removed from American’s automation and venture-capital-laden solar ecosystem. Instead of a world dominated by a few domineering (and hard-falling) giants, Asian solar production feels like a thriving ecosystem, full of companies of different shapes and sizes squeezing out a living for themselves anywhere they can fit. A rejected cell is a boon, a resource. Like a carcass in the African savanna–no part of the animal goes to waste. The dinged up solar gets cut up, the bad bits sliced out, and the good pieces turned into microsolar modules. Life finds a way.

Mr Fujimoto assures us that he has 200KW of solar in the office at all times, ready to ship out at a moment’s notice. I get that drug-deal feeling, again. If the need is urgent, Fujimoto says, he will hand-deliver whatever quantity of solar is needed within twenty-four hours. We ask him about international shipping and customs. Ms. Wu titters quietly behind her hand. 10KW of solar, Mr. Fujimoto says, is very small. He can fit 10KW of solar cells into my backpack. His customers run microsolar factories all around the world, and they fly in with a duffel bag, check the merchandise on the spot, pay in cash, and walk out an hour later with all the solar they need for the next six months. It shouldn’t be so surprising to us–like everything else we’re finding in the world of microsolar production, it’s shady, ad-hoc, and there’s a thriving industry flying well below the radar.

We ask about quality, and Fujimoto says it’s time for a demo. He grabs a shrink-wrapped package of solar cells and we head to the backroom, where a man is testing wafer after wafer on a semi-automatic wafer testing machine. Fujimoto whispers a few words in Mandarin, and the man pops open the package and slides a wafer into the tester. There’s a flash, and the I-V curve of that wafer pops up on a screen. Rows and rows of statistics follow, showing the power output, voltage, current, input flux levels, grid resistance and a million other factors I never even thought existed. They guarantee all their cells, Fujimoto explains, to a minimum output of 4W. Meanwhile, the worker keeps loading more and more cells into the tester. Test, FLASH, test. Fujimoto points out that the cell is producing at least 4W. Test, FLASH, test. He does it again. After ten cells or so, we say we’ve seen enough. Fujimoto keeps the man testing cells for another couple minutes, until he’s sure the point of the demonstration has sunk in. It has. Microsolar is an under-the-radar, ad-hoc world, but Fujimoto is no crook. His product might be someone else’s waste, but businesses run on these waste cells, and he’s serious about providing the best rejected cells he can. It’s not a scam and it’s not glamorous–it’s just business. It’s the business of microsolar, and it all takes place out of the light of day.

by Alex at Mantis Shrimp Invention

Four months ago, I started working with Shawn to help make small, low-cost solar products for developing parts of the world.  In the process, we visited some solar factories in Southern China to source small micro-solar modules, and we were amazed to find out that EVERYTHING THEY’RE DOING IS WRONG.

Micro-solar is our own term for solar panels that put out less than 25 Watts.  Micro-solar is a big idea about a small thing–by cutting up solar cells and rearranging them, it’s possible to make very low-cost, low-power solar to embed in a product.  In particular, we’re interested in the ~1W solar panels, the ones that are small and cheap enough to be everywhere.  Over the last decade, we’ve become excellent at making ultra low-power electronics, from cell phones that last weeks in standby to LED lights to wireless sensor nodes.   Before 2000, we really didn’t have many electrical devices that could make use of a .4W power source.  Now, our engineering is more advanced, and there’s a slew of low-power electronics that we can couple with these panels to make all kinds of battery-less electronics.  Micro-solar is responsible for everything from solar powered garden lights and portable iphone chargers to home power systems for the developing world.  It’s incredibly versatile and useful, enabling a whole class of new, ambiently-powered products, and once there was enough micro-solar production online, a Billion-dollar industry emerged, practically overnight, to make products powered by micro-solar cells.  The scale of the industry means that the stakes are high, and there’s a lot of potential imapct and money for the groups that can make improvements in the production of micro-solar and micro-solar-powered products.  And that just makes it all the more extraordinary that China is filled with thousands of micro-solar plants, all running the exact same, stone-age process for making micro-solar, and struggling just to stay afloat.

Micro-solar is a Chinese innovation, originally.  Nobody even realized that micro-solar was possible until a chinese factory cranked one out one inauspicious day sometime around the turn of the century.  All the money and attention had been going to large scale rooftop modules, made at insane speeds by precision german machines.  The ‘innovation’ that made micro-solar possible was really just cheapness–cheap people and cheap materials.  Micro-solar factories could put together a slew of low-paid workers to slog through the production of a panel, manually working through each of the process steps.  The process, as you’ll see later, is dirt-simple, and is rubber-stamped all over Southern China, giving rise to thousands of factories all doing the same thing.

I’ll talk more about this in a bit–for now, let’s take a look at how these panels are made:

All micro-solar cells begin life as a standard 4 Watt silicon solar wafer just like the ones that the girl is reaching for in the picture.  Each piece of silicon only puts out a very limited voltage, around .65V with no current, and to make an electrically useful solar cell, the wafer must get sliced up, and the slices wired up in series to increase the voltage to a useable level.  This girl is sitting at a laser cutter that traces out patterns on the wafer, scoring the silicon so that it can be broken later.

The girl then breaks each of the slivers off the solar wafer by hand, stacking them up on a tray.

This manual, imprecise process inevitably creates buckets of ruined solar shards.

Once the cells are cut into slices, they need to be joined together.  The first step is to solder on thin, tinsel-like conductors onto each cell.  This is done entirely by hand.  This guy fluxes and solders solar pieces constantly for his 10-hour shift.

The pieces are then backed with pieces of double-sided tape.

After the pieces are backed, another worker peels off the paper backing from the tape and sticks the pieces next to each other on a plastic or cardboard backing.

This worker goes over the cells and solders each cell to the next one, forming an electrical series string.  This manual step introduces the bulk of the errors in the end product.

In preparation for epoxy coating, this worker brushes off each piece with isopropyl alcohol, to clean it.

Solar panels, ready for epoxy.

The epoxying is done in a somewhat dusty-free room.  Modules are placed on holders, and a transparent glue is gooped over each module and spread about by hand until the module is covered.

The epoxied panels are placed in a rack and loaded into a heating chamber where the epoxy gets cured.

After the panels are complete, another worker uses a halogen lamp to test their current and voltage output.  Panels that pass the test are shipped.

So, that’s the process.  What’s extraordinary to me about micro-solar production is that it’s a huge operation to run for a tiny margin.  A line of ~100 workers will crank out about 300,000 micro-solar modules in a year.  Competitive price pressures force factories to trim their margins down to the bone.  The largest single cost to the manufacturer is labor, accounting for about 50% of the shipped cost of a panel.  But somehow, nobody has refined the process to reduce the labor and increase their margins.  About 10% of the end product ends up with a flaw, due to the imprecise, manual production, and has to be trashed, adding to the manufacturer’s operating expenses.  But every manufacturer bears the same burden.  Manual handling and breaking of cells also produces huge losses of raw silicon, but again, everybody does it.

On top of all this, the end product isn’t that great.  Epoxy-covered panels only last three years before the ultraviolet light from the sun yellows and cracks the epoxy covering and tanking the electrical output from the panel.  Over three years, panels don’t produce enough energy to make up for the energy that goes into their production.  They just become rotting piles of yellowed e-waste.  The irony, of course, is that the raw silicon trapped inside the panel is still good–that silicon will last for twenty-five years.   The usefulness of the product is limited by the rest of the process, particularly the epoxy covering.  But despite the innumerable flaws, inefficiencies, and incentives to improve, micro-solar works well enough to get products into a Wal-mart, and as long as that’s the case, none of the manufacturers out there are touching it.

Seems like somebody oughta fix that.

by Alex at Mantis Shrimp Invention