Did you think of using a gyroscope, or a MEMS chip with analog outputs? I think this will be unstable because it will start to swing.

Hello, thanks for the question.
Yes, I do. I planning applying one in future, but at the moment my main goal is checking analog control conception and this is reason why I use my own simple tilt sensor.
About stability of my tilt sensor, actually, you right, but balance consist of three M3 nut, and this give a very small mass on short shoulder, plus ball-bearing have a lubricant with own viscosity which compensate the inertia of the balance. This few things reduce swinging risks. Of course, this features of this construction, and I can't guarantee what even my next similar tilt sensor will have same stability. But for the moment, this construction is suits me.