russian PDP-11 like K1801BM2 CPU Startup reading w/o address?

I once built a phase-shift oscillator out of three RC phase-shift stages and an op-amp. But I wanted something as close as feasible to a sine wave. So I put a variable resistor in the op-amp fedback, to control the gain. Then I fiddled with that. Crank the gain up and I got a severely clipped waveform; crank it down and it wouldn't oscillate (neither of which was surprising in the least). I fiddled with it until I found a setting that made the amplitude just barely grow, growing until it just barely clipped, at which point it stabilized. (It grew fairly slowly; it took multiple seconds to reach the point of clipping and stabilizing.) Finding the exact resistor value theoretically would have required knowing precise values for the components in the R-C network, the other resistor in the op-amp feedback, loading from parasitic capacitance (actually, the frequency - something in the audio range, I think - was low enough that one might have been ignorable), etc. I considered trying to build some kind of AGC for it, but never actually got the round tuits for that.