Monday, September 19, 2011

MakerFaire NYC 2011

I went to MakerFaire at the NY Hall of Science, Queens, NY on September 17th and 18th, and took plenty of photos. Like last year, the location was the site of the 1964 World's Fair. Even though I grew up pretty close to New York, I didn't get to see the World's Fair as a child, so I'm glad to have these opportunities to see what's left of it.
As was true last year, there were lots of 3D printers and CNC milling machines. My impression this year was that a much larger percentage of them were hobbyist efforts rather than high-end commercial projects. I think that's a good thing. There seems to me to be a maturing of the 3D printer hobbyist effort in general, and the gradual emergence of more small businesses like Bre Pettis's Makerbot. The hard technical challenges (the big one being getting the extruder nozzle to work just right) have pretty much been identified now.

As I looked at some of the products, which have been improving in resolution, it occurred to me that an interesting approach would be, instead of going with increasingly fine nozzles, to use a coarse nozzle to place a slightly oversized drop of plastic, let that drop cool and harden, and then bring in a milling tool to shape it. This would mean moving back and forth frequently between the extruder nozzle and the milling tool, so it would need some tinkering and might not end up being an improvement.

There were lots of other tools, things on display, and cool stuff to see. I was really impressed with an elegant (if low-res) volumetric display called Lumarca. Essentially the guy uses a projector to project colors onto lengths of monofilament fishing line in the viewing volume, and by carefully controlling the projected image, he individually controls what colors appear along each length of monofilament.

There were a few interesting vehicles, like motorized skateboards and a Segway clone. Those were fun.

One thing I found interesting was that in addition to the expected Arduino stuff (which has the full weight of O'Reilly Publishing behind it), there were a good number of boards with more advanced microcontrollers, particularly ARM Cortex-M3 controllers. This interests me because with their larger address spaces and fuller feature sets, ARM processors can run Linux OSes or Python interpreters or other big pieces of code  beyond the itty-bitty programs that will fit on an Arduino. Teho Labs had a nice line of Cortex M3 boards. They shared space with Dangerous Prototypes who were showing off their Web Platform board.

I did get pretty tired and sore walking around so much, and needed some Advil. But it was definitely worthwhile. Maybe I'll have some project next year so that I can have a booth of my own.

Sunday, September 11, 2011

We need to build more educational computers

When I was a kid, I had this absolutely wonderful educational computer called Digi-Comp. It was very simple, with only three bits of state, and that was fine for learning an awful lot of basic stuff about computers. And it was sturdy as anything. I must have disassembled and reassembled it hundreds of times and it never broke and never stopped working. Somebody needs to design a 21st century mechanical computer using an inexpensive service like Ponoko or Shapeways or 100kGarages or Big Blue Saw to do laser cutting or 3D printing.

When I was in college, I saved up money to buy a KIM-1 single-board computer with a 6502 microprocessor, a hexadecimal keypad, and six seven-segment digit displays. I learned a lot of what became my career playing with that thing, writing little assembly language programs, soldering TTL chips to it, and generally having a great time. When I left school, I bought an old ASR-33 teletype from the school's department for retiring obsolete junk, and used it to give the KIM-1 a 300 baud line printer. Back in those days we had extraordinarily low thresholds of entertainment. Still, it was educational.

We need to be building more of this kind of stuff today. The things we build need to be easily hackable and easy to form user communities around. I guess you could say we already have something like the KIM-1 with today's Arduino, but it never fills me with intrigue like the KIM-1 did, where you were literally typing in machine opcodes onto that hex keypad, and they're showing up in the LED digits. It gives you a real sense of intimacy with the entire process of computation. Compiling C code never gets you quite that close to the action.

If you regard the Digi-Comp and the KIM-1 as two points along a spectrum of sophistication, we probably ought to plan on a third more advanced point, given that the KIM-1 is about 30 years old now. I've been doing some puttering with ARM-7 and ARM-9 boards of various kinds, some so capable as to run Linux, and I think that's a good third point because that's the kind of hardware that appears in modern consumer devices.