One use of the eStarling frame would be to prototype the Twitter Sampler aio.

The basic idea is a widget that sends your friend’s Twitters to an ambient display.  In keeping with the principles of the aiosphere, this object should not look like a computer display.

I imagine a graphics system with 1. a background image of coarse-weave fabric, and 2. a text overlay system which uses an embroidery-like font.

The display would be framed in an ornate wood picture frame.The result would be a dynamic ‘needlework’ sampler. If you’re not familiar with this historical practice, this Wikipedia article is a good start.

In final form this device would follow aiosphere principles and send very small amounts of data, offloading processing complexity to the devices at either end.

For a proof of concept one could write PC-side code that adds text to an image and emails it to the eStarling.

Not really an aiosphere project because of the bandwidth, but worth mentioning.  eStarling makes some cheap wifi picture frames that you can even email pics to.  Such a system is a great foundation for intriguing time-based systems.  A remote cam somewhere, set up to periodically send an image to the frame, makes a wonderful real-time window onto another place in the world.  Imagine a gallery of frames, all windows onto other places in the world?

I imagine outdoor shots, so it could get complex and expensive.  But to test such a system, why not use Panasonic’s low-cost net cam, which can email pics?

Lunch with my friend Ed today, another of many discussions over the years about the aiosphere. He is of the opinion that the whole thing is simple. I am of the opinion that the whole thing is hard. I guess it might be because we have different ideas about what “the whole thing” is. He favors mocking up a pair of hardware objects, and throwing together some scripts and PHP code to make small amounts of data flow between them.

Thing is, I know that people have been doing this exact thing for over a decade now, and reinventing the wheel each time.  I believe that what is missing is a universal architecture; hardware, firmware, software and script; which supports both the rapid creation of new hardware, and the easy creation of new linkages and behaviors for existing hardware. A universal ambient gadget system. This is the aiosphere.

But the distance between our perspectives simply serves to highlight the fundamental problem with my concept: I cannot easily articulate it, describe it, define it, its scope, its shape, its size. This, I think, will be my biggest challenge. I keep telling Ed, making objects is easy; coding point to point communications is easy. Making a demo from those ingredients is easy, and people do it every day.  But creating a new protocol, that overlays the global information network, and provides dead-simple connections between objects in different places, that is the grail (and I will be happy if it turns out to be easy too, but I’m not counting on it).

I am always searching for smaller, cheaper microcontrollers! Especially with UART.

NEC makes some compelling devices because of package size (2×2mm), but their feature set is very limited. No UART, no comparator.

Microchip doesn’t put the debug engine on their smaller, lower-cost micros. If you don’t mind debugging on a different micro, their F687, in 4×4mm QFN, is an attractive target for some designs, about a buck fifty in small qtys. Of course, they also make some very tiny, very low cost devices (10Fxx), but they are just too awful to program.

Zilog makes some 5×6QFN parts with UART, and they have a free dev environment. Hardware multiply! Lots of memory. About a buck and a half in small qtys.

NXP also make some small footprint and low cost (about a buck) micros, but only 1k mem max. Still, if you can make that work, the price is a powerful argument. High speed architecture, nice UART features, 3×3QFN.

Silicon Labs makes some useful micros for designs requiring lots of power and small footprints. Hardware multiplier, big memory options, fast clock, 3×3QFN, starting at about a buck seventy-five in small qtys. Can migrate to an OTP part for higher volumes, which cuts the price about in half.

Atmel makes one small, low-cost micro with UART, but it is too memory-constrained for many of my designs. Still, cheap.

As the creator of the MSP430 user group, I feel compelled to mention TI’s offerings. 6×6QFN, just over two bucks in small qtys. Buckets of memory, extremely low power, hardware multiplier.

And honorable mention goes to Freescale’s RS08 family, very few peripherals, not C programmable, but less than fifty cents even in small qtys.

Let’s make tiny smart objects!