Process · Physical Design & Signoff
Shielding
Routing grounded or supply-tied wires beside or above and below a sensitive net so that coupling to switching neighbours is intercepted, limiting crosstalk delay and noise on that net.
- Also known as
- Net shielding, Coaxial shielding
- Type
- Process
- Domain
- Physical Design & Signoff, Signal integrity
Notes
Full shielding costs roughly three tracks per shielded net, so it is reserved for clocks and analogue-sensitive signals.
Overview
Shielding, in chip physical design, is the practice of routing wires tied to ground or to the supply alongside a sensitive signal so that switching neighbours couple to a quiet wire instead of to the signal. It is applied selectively, to clocks and long critical wires where crosstalk would do the most harm, because every shield takes routing tracks. It is a different thing from electromagnetic shielding of a package or board, which uses metal enclosures or layers to block radiated interference.
How it works
A shield is an ordinary wire on the same metal layer as the signal, placed in the neighbouring track and connected to the power or ground network. Because a quiet supply wire now sits between the signal and its neighbours, almost all of the capacitively coupled noise that would have reached the signal is intercepted. Delay becomes more predictable too: without a shield, a neighbour switching in the opposite direction makes the signal behave as if its coupling capacitance were roughly doubled, while with a shield the signal always sees the nominal value. A nearby supply wire also gives the signal's current a short return path, which reduces inductance on fast global wires.
In practice designers attach shielding requirements to particular nets through routing rules, often together with extra width or spacing, and the router lays the shield wires in as it routes those nets. Shields may run on one or both sides, and one shield can sometimes be shared by two critical nets routed either side of it.
The cost is area and congestion. Each shield consumes a track that could carry a signal, shields must be tied into the power grid, and the number of nets needing protection tends to grow at higher frequencies. Alternatives include simply spacing wires further apart, resizing drivers and, in research, active shields that switch along with the signal instead of being tied to a supply.
Examples
- A clock tree whose main trunks are routed with grounded wires on both sides so that nearby data switching cannot disturb clock edges.
- Shield sharing, studied by Intel engineers for a gigahertz-range microprocessor, in which one supply wire protects two critical nets and the shields also help form the power grid.
- A sensitive analogue reference line in a mixed-signal chip routed between supply wires to keep digital switching noise off it.
- Active shielding, a research technique in which the wires next to a critical net are driven in step with it rather than tied to a supply.
Related terms
Part of the Physical Design & Signoff glossary, with every term in the topic and the lifecycles, tools and standards that use them.
Where it appears on the Hub
Used on 7 pages of the Hub. In each group, the pages that mention it most come first.
Suggest an improvement
Suggest an improvement to this entry
A clearer definition, an example, a correction or a source an editor can check. Editors review every suggestion before the entry changes.
Last reviewed . Written by The Semiconductor Hub editorial team. The full glossary