ccs (composite current source)

CCS (Composite Current Source) is Synopsys's advanced waveform-based timing and noise model that represents cell output behavior as time-varying current sources rather than simple delay/slew tables — providing significantly more accurate timing, noise, and power analysis than NLDM, especially at advanced process nodes.

Why CCS Is More Accurate Than NLDM

  • NLDM: Models the output as a single delay value and a linear ramp (one slew number). The actual waveform shape is lost.
  • CCS: Models the output as a current waveform that interacts with the actual load network — capturing the real voltage waveform shape, including non-linear transitions and load-dependent behavior.
  • This matters because at advanced nodes:
  • Waveforms are not linear ramps — they have distinct shapes that affect downstream cell switching.
  • The interaction between driving cell and load (Miller effect, crosstalk) depends on the actual waveform.
  • Setup/hold timing is sensitive to waveform shape, not just arrival time.

CCS Model Components

  • CCS Timing: Current source model for output driving behavior.
  • Stores output current vs. time waveforms for each (input_slew, output_load) combination.
  • The STA tool convolves this current with the actual RC load network to compute the precise output voltage waveform.
  • Result: More accurate delay and transition time that accounts for the specific downstream network.
  • CCS Noise: Noise immunity and propagation model.
  • Models how noise glitches on inputs propagate to outputs.
  • Captures the cell's noise rejection characteristics.
  • Used for signal integrity analysis to determine if crosstalk-induced glitches cause functional failures.
  • CCS Power: Current-based power model.
  • Provides more accurate dynamic power estimation than NLDM's energy tables.
  • Captures the actual current draw profile during switching.

CCS vs. NLDM Accuracy

  • Delay: CCS is typically 2–5% more accurate than NLDM for single cells, with larger improvements for cells driving complex RC networks.
  • Setup/Hold: CCS can be 10–20% more accurate for setup/hold time computation — critical for timing closure at advanced nodes.
  • Noise: NLDM has no noise model. CCS provides full noise analysis capability.
  • Waveform: CCS produces realistic non-linear waveforms; NLDM produces only linear ramps.

CCS in the Design Flow

  • CCS data is stored in Liberty (.lib) files with additional CCS-specific sections.
  • Characterization: More data must be extracted during library characterization — current waveforms in addition to delay tables.
  • File Size: CCS Liberty files are 3–10× larger than NLDM files — more data per timing arc.
  • Runtime: CCS-based STA is 10–30% slower than NLDM due to more complex calculations.
  • Sign-Off: CCS is the recommended (or required) model for sign-off timing at 28 nm and below in Synopsys flows.

CCS is the state-of-the-art timing model for Synopsys-based design flows — it provides the waveform accuracy needed for reliable timing closure at advanced semiconductor nodes.

Go deeper with CFSGPT

Get AI-powered deep-dives, save terms, and run advanced simulations — free account.

Create Free Account