In low power design, we may shut down main parts of chip power, so only one or serval block is still power on. In order to no floating input pins in powered block, we must use a isolation cells to lead input pins to logic '0'. We also say isolation cell as clamp cell.
In Low power design an isolation cell is required where each logic signal crosses from a power domain that can be powered down to a domain that is not powered down. The cell operates as a buffer when the input and output sides of the cell are both powered up, but provides a constant output signal during times that the input side is powered down (prevents short ciruit current due to folating node of always on block). An enable input controls the operating mode of the cell.
A cell that can perform both level-shifting and isolation functions is called an enable levelshifter
cell. This type of cell is used where a signal crosses from one power domain to
another, where the two voltage levels are different and the first domain can be powered
down.
When you power down a domain, the charges get accumulated in the intermediate nodes (floating nodes ). If this intermediate node is input to the always on block will cause short ciruit current.
hi vikarm
as far as i understood,
isolation cells are used to avoid the crowbar currents, as the other power domain is off so i feel we dont require to put isolate cells for power down block,
In multiple-supply designs and multivoltage with shut-down designs, the outputs from the shut-down partition into the active partition must be maintained at predictable signal levels.This signal isolation is achieved by using an isolation cell. The isolation logic ensures that all inputs to the active partition are clamped to a fixed value.
(my understanding)Moving higher technology nodes above 90nm, Even though the Shut off region is not conducting, Logic 0/1 coming constantly will affect leackage in the Shut off region interface cells.