Marcy Rye
To achieve the level of functionality I wanted it was mandatory to find some way to control the individual brightness of each LED in the matrix. It took a long time, and I only found something that worked by a stroke of luck.
You can see the chip I eventually found in the image on the right. It's located where the red and yellow wires meet on the surface of the breadboard. On the left, is the 2-Stamp network where one Stanp gets the sensor data, and sends the numbers to the other, which feeds instructions to the MAX7219 based on the sensor input.
I later switched to a single Stamp since the serial communication was causing more problems than it was solving.
I searched for weeks trying to find a chip that was designed to simultaneously control the individual brightness of 64 LEDs, and that would also interface with the basic stamp. I found many chips that would allow for individual on/off control of a matrix of LEDs, but even these usually were limited to 8 or 16 LEDs. I found chips that would control brightness, but one chip per LED was required, and I have 64 LEDs, so that would have made the circuit too large.
I talked to engineers all over the place, at companies in England, Canada, California, Texas, and New York. Most of them were very helpful and gave me whatever information they could, but none of them were aware of a chip that would do what I wanted. Some knew of products in development that might be able to do it but were unable to discuss it due to nondisclosure agreements.
I started working out a solution by way of a very large network of something like 16 Stamps such that each LED would be on a seperate Stamp pin. (The Stamp's PBASIC language has a built in pulse-width-modulation command that easily controls LED brightness.) However, once I setup a test network of 4 Stamps and only 16 LEDs with the PWM command, I found that while I could in fact control indivdual brightness that way, it was very very slow.
One day, I was talking to a tech at Parallax (that sells the Stamps) about how to maximize the speed of my network (using serial protocols and pwm). On a whim, I asked him if by any chance he knew of a single chip that could control the indivdual brightness of a grid of LEDs. Without a single hesitation, he answered, "MAX7219." I grilled him. Would it control brightness? Would it control the brightness of each LED separately and simultaneously? Could I have 64 LEDs on at the same time, each with a different level of brightness? He said, "sure."
I had talked to so many people who told me it was impossible that I asked him my questions over and over, and he very kindly answered them over and over. Yes, yes, yes. All was possible. The MAX7219 was designed as an LCD driver chip and as such it permitted individual control of an 8x8 grid of separate LED units. The specs for the chip were designed to st it up with an LCD display to show numbers or letters (like on your VCR clock that forever says "12:00"), not individual LEDs. All I had to do was to create my own display by soldering together an 8x8 matrix of LEDs.
So I got one, and connected it to the Stamp, and after much trial and error, was able to control the on/off state of 8, and then 16 LEDs with it. I kept rereading the specs and messing around with the code until a couple weeks later I hit on a way to control the brightness of each LED by sort of hacking the way it was designed to address the segments of the digits in a numeric (or alphabetic) LCD display. After that, it was just a matter of getting the serial settings right to maximize the speed of the lights changing brightness. And if you really felt like going nuts, it's possible to cascade the 7219s such that you could create a grid of 256 LEDs, or as many as you can afford to buy. (LEDs are EXPENSIVE! especially if they are blue.)
Once I hooked up the 64 LEDS, it turned out that the whole circuit worked better with a single Stamp, eliminating serial control altogether. This let the lights change brightness very fast, and made the rate of change extremely responsive to sensor input. The result was a deep engagement of the participants since they could clearly see the effect of their interaction.