Flavonoid

From TechkwondoResearchWiki

1 Synopsis
2 Prototypes
3 prior and influential art, technology, articles, research, etc.

Synopsis

Flavonoid links 1st Life and 2nd Life worlds so as to provide a mechanism for translating embodied, kinesthetic activity into 2nd Life actions. A homebrew, Internet-enabled kinesthetic sensor, conceptually similar to a traditional pedometer, is being designed as a networked object that bridges the geophysical worlds (1st Life) and online digitally networked worlds (2nd Life). By providing data feeds about the kinesthetic activities of the person wearing Flavonoid, various embodiments representing that data can be created in 2nd Life, such as the appearance of online avatars, or that avatar’s wealth or capabilities. Flavonoid is designed with health and electronic gaming in mind, as well the ways in which action and activity can be transacted between geospatial, physical worlds and digital online worlds — between 1st Life and 2nd Life. Both of these worlds — 1st Life and 2nd Life — are conceived of without rigid definition. Suffice it to say that there is an embodied world, and one in which embodiment can only be achieved through digital mechanisms. The translations and linkages between these worlds occurs through a prescribed set of transactions, typically using a computer of some sort, a network connection, and a semantic media ecosystem. One such example of such transactions would be an online social environment such as World of Warcraft. For the purposes of this project, I would categorize World of Warcraft as a 2nd Life world with no direct 1st Life embodiment of the sort that concerns this project

C5 "Quest for Success" RFQ Submission

Accepted for the C5 "Quest for Success" Talk Competition

[edit]

Prototypes

[edit]

1st Flavonoid Prototype

Slating Flavonoid. Here the device is communicating its current readings in real-time to my laptop. I devised a bit of code that simply keeps count (four bytes, two words, high and low) the current number of steps. My idea is that these figures would be automatically slated with a timestamp anytime the device is linked to the Flavonoid mothership. One idea I'm doing presently is to have that happen over Bluetooth (very easy, expensive in the context of what I think the device should cost) or USB (also easy, and cheap, but more awkward and you can't really do it easily if you configure the system to use a phone, like tne N70, to automatically upload the data while on the go). I wore the device throughout the day and slated it when I got back (total steps - 0x0B87 in 8 hours and 1 minute)

This one was a prototype using a Basic Stamp, Bluetooth radio, and a Memsic 2125 dual-axis accelerometer. It's basically a DIY pedometer, that counts how many steps you've taken. The data is sent over the Bluetooth link if one is up. Otherwise, it stores the data in a nugget of EEPROM on the Basic Stamp.

What I was hoping to accomplish is just get a little work in creating a movement sensor in a very simple way. The prototype got me thinking about more complicated motion rather than just pedometric, for instance, the motion that might be involved when snowboarding or surfing, for instance. That kind of motion is more fluid in a way than walking, which has a predictable sort of cadence and beat.

[edit]

2nd Flavonoid Prototype (SPI)

Here's a board layout for the second Flavonoid prototype. It's a lot more sophisticated electrically and just conceptually. I've included a 3-axis accelerometer that's rather expensive ($15), a real-time clock (RTC), touch sensor and a bit of flash memory. All these devices run along a four-wire interface (SPI) and are controlled by a microcontroller that's off board. (Using Atmel's, like the ATmega8 and ATmega16 — although I may end up using an ATmega32 as the code seems to be ballooning beyond 16kb.)

The clock is there to design in a way to encourage sustained activity. It'll hold time even without power, so the frame of reference can be on the order of hours or days, rather than just minutes. The touch sensor — well, I have ideas for that, but one pragmatic reason is to help mitigate "cheats." I've tried this sensor and it'll meausre body capacitance through a hunk of wood and fabric. The flash memory is to contain data from the readings. I'll need to use some sort of circular buffer to account for the fact that these EEPROM can withstand 100k read/write cycles and I suspect these I'll need a few orders of magnitude more headroom.

Top of sample from PCB Fab Express

Bottom of sample from PCB Fab Express

And the manufactured prototype board.

(More information on the boardhouses and a very modest review I wrote is here.

Here's an assembled prototype using the SPI interface. The image documentation is here.

[edit]

2nd Flavonoid Prototype (I2C)

This is an I2C prototype. You can see it's smaller, largely because my skills with the PCB layout tools are refining. But, also — I2C requires fewer nets between the parts.

This particular variation is using a dual-axis accelerometer — the MXC6202xJ from Memsic. This prototype taught me a few things, including confirming my opinion that I should really use a tri-axis accelerometer. I also learned a few things about laying out the footprints for small pitch surface mounted parts: you must design the landing pads slightly longer than specified. (I should come up with a percentage oversize as a design rule.) Having a slightly longer pad would make the reflow go easier, it seems, as well as proving a bit more of a visual indicator as to how well the reflow went. Intuition also suggests to me that the device might adhere better with the increase of a bit of alignment due to surface tension of the solder and the pad on the device.

[edit]

3rd Flavonoid Prototype (I2C, ADXL330, On-board Microcontroller)

I jumped a bit ahead and went and designed a board with an ADXL330 tri-axis ratiometric accelerometer, using the on-board A2D converters on an Atmel ATMega32L. It has a USB connection as well, along with a few extra pins connected to the PWM pins of the ATmega32L that I could attach to some three color LEDs. I'm still building the board..by hand. I switched to all 0603 surface mount discretes (at least most of them), which takes small to a whole new level, especially for hand-constructing these things. But, it's fun and a great exercise in careful, detailed craft work. No reflow with the skillet here. I decided to try applying a bit of flux to the tin, and then floating a thing layer of solder on the pads. Placing the super small 0603 surface-mount devices on the pads and applying just a little heat is enough to get a decent solder joint to hold the device on the board. Now, if only everything will work once it's all done!

This board came from Advanced Circuits — I thought I'd try out their $33 board special for the first prototype. Mostly because I wanted something quickly. You can see it comes back bare-bones — no solder mask, no silk..nothing. The lack of solder mask is funky — makes me nervous about areas of the board that are susceptible to solder ball-style shorts.

Here's the board on a test harness to make sure the FT232RL (that chip there) is working. It's the vascular interface between the Flavonoid and the rest of the world — it pumps out the Flavonoid data when you upload yourself to the mothership.

The microcontroller on board, no crystal yet, but it runs using the internal RC oscillator. I was pretty surprised it worked, but I fracked the fuse settings on the first chip. Turns out I can probably set it straight by hooking this chip up to an external clock source — which is how I set the timing fuse, darnit. Anyway, I replaced it and set the fuse correctly to clock from a fast external crystal and it worked like a charm. Flashed the chip with my little test command prompt thing and saw it breathing.

This is basically all finished here. Prototype v03. What did I learn? The touch sensor is a little problematic because it may need to be carefully tuned to respond to human-body capacitance rather than the capacitance of any large object. There are some design notes on tuning the thing, but this will take time. Also, it will be easier to design for the sensor once it is in some sort of enclosure, perhaps wooden, for the aesthetics. It is entirely possible to build the code on both the Mac and on Windows, which is a great boon. Right now, I'm compiling on my Mac laptop and flashing the device on the PC. The Windows PC has the laptop drive mounted as a shared volume, and I use AVR Studio's AVR Programming interface, which is connected to my STK500 development board. I also have a AVR ICE MkII programmer which _should_ work from the laptop directly, although that'll take some fiddling to get going. I can leave the Flavonoid connected over USB on the laptop and just hook it up to the programmer and leave it that way, so that I can compile, flash the device, and then see the new output over USB on the Mac without having to undo any connections, move things around, etc. I created a simple bit of code to do basic testing of the I/O, test each of the little "subsystems" (EEPROM, the real-time clock, accelerometer) individually. The small 100mA 3.7 volt battery didn't last long enough. I replaced it with a 900mA 3.7 volt battery that's almost exactly the same footprint as the PCB. Next Steps: send off the slightly modified board design to get fabricated. Start thinking about the firmware to support the device's purpose. Talk to someone about enclosure fabrication. Incremental Prototyping: Testing the QT113H Proximity Sensor

[edit]