I have a ring osc design project. The specs are

1.  span: 2.4G-2.424G
2. Kvco,60Mhz/V
3. Phase noise: <-85dbc/Hz@ 1Mhz offset
4. Voltage swing, bigger than 0.25V but cannot saturate
5. 2 stage

Cmosp18 tech. I am not sure about which kind of delay cell to choose, from 4, it should be an unsaturated one. Besides, in case I have choosed some circuit structure, how to size the transistors? Any suggestions? Thanks in advance

I don't have much experience with oscillators but I think using a ring osc. with a triode load of the delay stage will result in improved linearity as the voltage swings are lower than for a load in saturation (because the output swing is limited by the overdrive voltage of the load transistors). However, this means also transitions will be slower and the phase noise will be higher than for a saturated load delay stage where the output swing is higher. I might go with a differential inverter with a Maneatis load.  Judean1, may I ask why you would choose a triode load as I would like to learn more about different osc. topologies.

Sven

Thanks for all your reply!

To clarify a few points:

Actually I have tried both the maneatis and the diode loaded delay cell. However, both of them would only oscillate when there are at least three stages. In order to make the Ring oscillator oscillate at 2 stages, some positive feedback mechanisms have to be introduced, for example, a cross-coupled latch.  However, in this way, the osc will saturate and that is the tricky point. So what I am looking for is the delay cell that can introduce at least the 90 degree of phase shift in order to oscillate while without using latch in order to be unsaturated. Sound contradictory, ur

BTW, this is a research project, so the specs might not make any sense from the practical point of view.