Quantitative Modeling and Optimization of VRM Filter Parameters Under Current Mode Constant On-Time Control

This paper proposes a time-domain computational model to optimize filtering parameters in Current Mode Constant On-Time (CMCOT) controlled Voltage Regulator Modules (VRMs) for high-performance processors. CMCOT, being a nonlinear control technique with distinct steady control and transient control mechanisms, complicates the precise quantification of required filtering parameters. The proposed model establishes an exact time-domain behavioral representation of the VRM, solved via numerical methods. It enables the quantitative determination of the minimum filtering capacitance satisfying dynamic performance specifications under given CMCOT parameters. This approach reduces design iterations and prevents capacitor overdesign. Validation results demonstrate 0.4% error against SIMPLIS simulations and 2% error against a 16-phase buck prototype, confirming the model's effectiveness.