Positive parallel feedback oscillator

I have designed and built an oscillator using positive parallel feedback using a microstrip ring resonator. I designed the feed back and amplifier circuits separately. The amplifier is conditionally stable at the oscillation frequency. When power is applied oscillations do not start up. However when I touch the microstrip circuit the oscillation does start up.

What I want to know is if oscillators of the parallel positive feedback type require the amplifier section to be unstable. I have read "Microwave Transistor Amplifiers, Analysis and Design" by Gonzalez and have not found an answer to this question. In my opinion I believe the amplifier can be unconditionally stable and oscillations should still build up due to the noise being amplified and then filtered through the feedback network.

Any advice would be appreciated.
-Brandon

I have built the feedback type but with a coaxial resonator. The active device does not have to be stable, but you do need an excess of open loop gain (3 dB or a little more) at zero phase to get the oscillation started. Does the oscillation start with your finger on, and then the power supply on? I assume yor are coupling it to 50 ohms somehow (through small cap). Does it start up without loading? Disconnect it and use a loop probe (made from coax) right above the resonator to sample the signal.