In fully differential ring oscillator , I ma using the delay cell as shown in attached image.
In this the top two mosfets will be in the triode region so that by varying the vcontrol we can vary the resistance. The MOS at the bottom, which is acting as constant current source should be in saturation across PVT.
What's about the differential input pair mosfets? It should be saturation across the PVT or what?
I don't think it matters a lot whether the devices
are in, or straddling, particular region(s). Of course
the diff pair devices will be toggling between "off"
and either linear or saturation (-ish).
Curious why you'd choose to vary the load and not
simultaneously the tail source. You could run into
more corners trouble if these are let wander
independently. I'd be setting Itail to be Iload*2
at the bias reference.