# For this particular Ring oscillator topology, will the circuit prefer to latch up or oscillate?

This circuit is chosen from Razavi's textbook Design of CMOS Phase-Locked loops
In this circuit, where the outer inverters are weaker than the inner ones, does the circuit prefer to latch up or oscillate? What is the reason for this preference?

• Why not use a simulator and dig deeper into the reasons. I mean, as it stands, it's a pretty-much unbounded question which will likely result in answers that aren't much practical use. Commented Mar 8 at 8:11

Has this textbook covered a 2 inverter latch?

This circuit is essentially 4 of these in series.

By describing the outer inverters as weaker, I am presuming that each strong inverter output can overdrive the weak inverter output so that the next strong inverter can change state.

This seems like a thought experiment that would be a good homework question, which I'm assuming this is.

Understand how a single cross coupled inverter pair will work, then the answer is evident.

• which textbook covers this material? May u give some references? Really appreciate it! Commented Mar 8 at 2:50
• I was assuming the textbook you referenced would have this covered, but perhaps if the book focuses on PLLs it might not. Searching for 'cross coupled inverter latch' gave many good links. The key is that one inverter is 'strong', the other is 'weak'. A strong inverter output can overdrive the output of a weak inverter - a weak inverter putting out a '0' can have it's output forced to a '1' state by a strong inverter with a '1' output. Commented Mar 8 at 3:11
• Why do this? If you assume the input logic to the cross coupled inverter pair can become high-Z, the weak inverter will then hold the strong inverter at the last binary logic value it was driven to. Commented Mar 8 at 3:13

Treated independently, each ring can settle into a stable state:

simulate this circuit – Schematic created using CircuitLab

This is only my intuition, but it seems to me that there is a condition in which both rings are stable, and neither disagrees with the other. If they were then joined at their corners, nothing would change, regardless of their relative "strengths".

I think this would likely latch into a stable state, but this treatment does not consider propagation of changes. It's not clear to me what would happen if one corner is momentarily forced to a different level.