BeginnerQuestion 15 of 50

What is the "primary power net" of a power domain?

From PDVerse Low-Power Physical Design Mentor Guide, part of the pdVerse Mentor Guide

By Tabish Iqbal, who has 9+ years of VLSI Physical Design experience

Short Answer

The primary power and ground of a domain are the default supply for every cell in it: any instance without an explicit supply connection gets them automatically. In IEEE 1801-2015 you give them as the domain's primary supply set; older UPF named the nets directly. Physically, they are the rails that the standard-cell rows in that domain sit on.

Domain PD_COP with standard-cell rows tapping the primary rail VDD1p0_SW and VSS through follow-pins, while an always-on strap VDD1p0 runs alongside to a retention flop and an isolation cell only. — Technical Reference DiagramWhat is the "primary power net" of a power domain?

Technical Reference DiagramWhat is the "primary power net" of a power domain?

Technical Explanation

Primary supply: IEEE 1801 implicitly connects the primary set's power and ground to every instance in the domain that has no supply ports of its own.
Modern form: create_power_domain -supply (UPF) with {primary SS_COP} names the set directly.
Legacy form (still accepted by ICC2/PT): set_domain_supply_net -primary_power_net -primary_ground_net (UPF).
Physically, the primary nets become the row rails, so ordinary cells in the domain need no extra routing.
Special cells, such as retention or isolation cells, take a second supply through a secondary PG pin from an always-on strap.
Get the primary wrong and whole blocks simulate as unpowered, or cells land on a rail that switches off unexpectedly.
Run check_mv_design (ICC2) after loading the UPF; it reports supply connectivity problems before they reach placement.

# [UPF]  design.upf
create_supply_set SS_COP -function {power VDD1p0_SW} -function {ground VSS}
create_power_domain PD_COP -elements {U_COP} -supply {primary SS_COP}

Common Mistake

The Trap: Using the always-on net as the primary of a switchable domain, or the switched net as the isolation and retention supply.

In the first case the domain never really powers down; in the second, isolation and saved state vanish the moment the switch opens.
Before any shutdown test, confirm each switchable domain's primary is the switched net and each strategy supply is always-on.

Follow-up Question & Model Response

"Can a domain have more than one primary power net?"

Candidate Model Response: No. Each domain has exactly one primary supply set, meaning one primary power and one primary ground. Extra supplies attach through strategies, such as the isolation or retention supply, or through explicit connections to secondary PG pins. If part of the logic needs a different primary, it belongs in a different domain.

Practical Example

Design Scenario: MYCHIP: PD_COP's primary is SS_COP, whose power function is the switched net VDD1p0_SW; the always-on VDD1p0 feeds the rest of the chip. In legacy form the same intent reads set_domain_supply_net PD_COP -primary_power_net VDD1p0_SW -primary_ground_net VSS (UPF). Rows in PD_COP sit on VDD1p0_SW, and retention flops reach up to a VDD1p0 strap for their shadow latch. Check: PD_COP has one primary set, and none of its strategies name SS_COP as their supply.

Low-Power & UPF Handbook