Meaning
The selective disconnection of power from specific functional blocks within an integrated circuit reduces the total energy consumption during idle periods. Dynamic power gating utilizes sleep transistors to cut off the current path to sub-systems that are not currently executing tasks. This technique targets the leakage current that otherwise flows even when a module is not switching.
It allows a system to maintain a high performance state in active blocks while keeping inactive sections in a near-zero power state.
Switching Logic
Control signals from a central power management unit determine when to apply or remove the supply voltage to individual domains. The implementation of dynamic power gating requires header or footer transistors that act as high-side or low-side switches. These switches must be sized correctly to handle the peak current requirements of the gated block without introducing measurable voltage drops.
Fine-grained control allows for specific radio components or processor cores to be isolated independently.
Leakage Suppression
Minimizing static power becomes critical as transistor dimensions shrink and sub-threshold conduction increases. By physically opening the circuit, the gated block stops consuming energy that would otherwise be lost to heat. This suppression extends the operational life of battery-powered sensors and mobile devices.
Latency Impact
Re-activating a gated domain introduces a delay known as wake-up time. The system must wait for the internal capacitances to charge and the voltage to stabilize before the block can resume processing. Excessive switching frequencies might result in energy penalties that outweigh the savings from the idle period.
Careful scheduling of power states ensures that the performance requirements of the application are met without wasting power.