My "theory" is that a Quad requires either an accellerometer or human input to keep from flipping. Any other possibilities?
The IR sensor only works in some environments.
I'm pretty sure that at some point an accelerometer (or accel/gyro combo) is needed to tell the copter to balance itself. I haven't seen any IR ones before (only ones iv seen were built by a couple friends of mine in college). Guess if you really wanted to get pointless though you could make them balance off the signals of a few rangefinders (pointed at teh ground on different spots of your copter). But now thats just me being idiotic.
You're correct it can measure fast rotation. That's about 6 deg in a Hertz! ;-) Sorry 6 degrees in 1/50 sec. So we can stop it from rotating violently. Let's talk about Yaw for now. If it is spinning slowly, say 30 deg / minute, the ENC-033r will not be able to measure it. It will return a zero value. If we can't measure it, how can we stop it from happening? MarkT said this sensor is for short term control only. So we need another sensor. But some copters don't have another sensor. Am I correct?
You might end up needing a second more accurate sensor to operate when the copter isn't moving around to keep it in line.
Yes this is total suck because it means you need another one (or few) of your ports on your arduino(or other controller) to accomplish the same task, or you could build a circuit to amplify the output of the gyro at its lower sensing range (basic instrumentation and controls), and have a digital port control when the "high speed low accuracy", and "low speed high accuracy" circuits are turned on. (when HSLA starts to get low switches over until an input is given or if the LSHA signal starts to get too high)