Level shifters or isolation cells landed in the wrong voltage area. How do you debug it?

Short Answer

The strategy -location decides which domain a cell belongs to, and the domain decides which voltage area it may be placed in, so a cell in the wrong VA is almost always a location or domain problem, not a placer bug. Start from check_mv_design (ICC2), then trace the cell with report_mv_path -cell (ICC2) to see its strategy, supplies and domain. Fix the strategy or make the missing supply available, then re-insert rather than hand-moving the cell.

Technical Explanation

• Location rules: for set_level_shifter (UPF), -location self uses the domain whose port is shifted, parent its parent; ICC2 adds automatic as its default, IEEE 1801 defaults to self.
• Supply rule: whatever location you pick, the cell needs its supplies available there; a low-to-high shifter needs both rails.
• Shared VA: with shared_voltage_area (UPF) several domains with equivalent primaries sit in one VA, so a cell can legally land in a VA named after another domain.
• Placement follows domain: ICC2 places a cell inside the VA of its power domain, so a cell in an unexpected VA was usually inserted into an unexpected domain.
• Trace: report_mv_path -cell (ICC2) prints the strategy, supply, inputs, outputs and domain, or the reason insertion or association failed.
• Second opinion: check_lp -stage design (VCLP) checks the implemented cells against the UPF independently of ICC2.

# [UPF]  design.upf
set_level_shifter LS_DSP_OUT -domain PD_DSP -applies_to outputs -rule low_to_high -location parent
# [ICC2]  icc2_shell
check_mv_design
report_mv_path -cell U_DSP/LS_DSP_OUT_0
report_power_domains
# [VC LP]  vc_static_shell
check_lp -stage design
report_violations -app LP

What To Check

• The strategy -location matches where the supplies actually exist.
• The cell domain matches the domain of the VA it sits in, or the VA is shared.
• Secondary PG for dual-rail cells is present in that VA.
• ICC2 and VC LP agree after the fix.

Healthy, Suspicious & Hard-stop Results

• Healthy (illustrative): check_mv_design is clean and every LS_DSP_OUT cell reports PD_MYCHIP as its domain.
• Suspicious (illustrative): check_mv_design is clean, but VC LP reports a shifter whose output supply differs from the UPF.
• Hard stop: check_mv_design reports an unavailable supply on LS_DSP_OUT, or cells sit in VA_DSP with no VDD1p0 strap.

What The Interviewer Is Testing

• Do you know that location controls domain, and domain controls VA?
• Would you check supply availability before moving cells by hand?

Follow-up Question & Model Response

"When would you deliberately put a level shifter in the switchable domain?"

Candidate Model Response: When the receiving side has no room or no route, and the switchable VA already carries the needed rails. For example, a high-to-low shifter into PD_DSP can sit in VA_DSP if the input rail is present there. The cost is secondary PG in that VA and a cell that shares the fate of PD_DSP when it is off. That is acceptable here because the shifter only feeds PD_DSP logic, which is off at the same time.