Fractional-N PLLs for Performance Computing

Overview

Perceptia’s DeepSub™ pPLL03F is a family of all digital PLLs featuring low-jitter and compact area suitable for clocking applications with critical timing requirements at frequencies up to 4GHz. It is suitable as a clock source for performance computing blocks and ADCs/DACs with moderate SNR requirements.

Perceptia’s second generation pPLL03F family is available on technologies from 5nm to 40nm and across multiple foundry partners. We are continually expanding the range of technologies where it is silicon proven and can quickly port it to other technologies or foundries upon request.

To give SoC designers the maximum flexibility in building complex multi-domain clock systems, pPLL03F is very small (< 0.005 sq mm) and low power (< 2.1mW in GF 22FDX). It is well suited to applications with many clock domains where each is driven by their own PLL. To simplify system design, PLL03 has an integrated power supply regulator which allows multiple instances of PLL03 to share common power supplies. Alternatively instances of pPLL03 can share supplies with the blocks that use its output clock.

pPLL03 integrates easily into any SoC design and includes all the views and models required by modern back end flows.

The pPLL03F is built using Perceptia’s second generation all digital PLL technology. This robust technology delivers identical performance across many processes, regardless of PVT conditions. It consumes a small fraction of the area of an analog PLL whilst maintaining comparable performance.

pPLL03F can be used as an integer-N PLL or as a fractional-N PLL. The fractional-N mode provides a high flexibility to choose the best combination of input and output clock frequencies at the system level.

Perceptia further provides integration support and offers customization and migration services.

Key Features

  • Low jitter, suitable for many clocking applications, including high speed digital, ADC, DAC, medium-speed PHY
  • Extremely small die area (< 0.005 sq mm), using a ring oscillator
  • Output frequency can be from 1 to 400 times the input reference, up to 4GHz
  • Reference clock from 10MHz to 500MHz
  • Second-generation digital PLL architecture, providing integer and fractional multiplication
  • Two PLL outputs via separate postscalers
  • Post-scalers programmable from 1 to 2,040
  • Lock-detect output
  • Can generate a spread-spectrum clock from a clean reference
  • Oscillator output duty cycle better than 48 / 52% on CK_PLL_OUT0 and CK_PLL_OUT1
  • Highly testable using industry standard flows
    • ATPG vectors provided
    • Specification of functional tests to supplement ATPG testing
  • Industrial operating conditions (-40 to 85°C) with junction temperature up to 100°C

Benefits

  • Fractional multiplication with frequency up to 5GHz (12nm and below) or 5GHz (16nm and above)
  • Low jitter (< 10ps RMS)
  • Small size (< 0.005 sq mm)
  • Low Power (< 2.1mW in GF 22FDX)
  • Support for multi-PLL systems with tens of PLLs per SoC
  • Easy integration

Block Diagram

Fractional-N PLLs for Performance Computing Block Diagram

Applications

  • Performance Computing
  • Complex SoCs with many clock domains
  • Low-jitter applications
  • Moderate SNR ADC/DAC

Deliverables

  • Datasheet
  • Detailed Verilog behavioral model
  • Timing models
  • LEF5.6 abstract for floor planning/chip assembly
  • Integration Guide
  • DFT Guide
  • Integration support
  • Characterization report
  • GDSII layout macrocell
  • CDL netlist for LVS
  • DRC, LVS and SI verification reports
  • Netlist model with accompanying documentation allowing integration of the module in scan chains

Technical Specifications

Foundry, Node
Portable to all CMOS processes 65nm and smaller. See specific product listings
Maturity
Silicon Proven
Availability
Depends on process - please ask
GLOBALFOUNDRIES
Pre-Silicon: 12nm , 14nm , 14nm LPE , 14nm LPP , 20nm LPM , 28nm , 28nm FDSOI , 28nm HPP , 28nm LPH , 28nm SLP , 32nm , 40nm LP , 55nm , 55nm LPX , 65nm , 65nm LP , 65nm LPe
SMIC
Pre-Silicon: 14nm , 28nm , 28nm HK , 28nm HKC+ , 28nm PS , 40nm LL , 55nm G , 55nm LL , 65nm LL
Samsung
Pre-Silicon: 10nm , 28nm FDS , 28nm LPH , 28nm LPP , 32nm LP , 45nm LP , 65nm LP
TSMC
Pre-Silicon: 5nm , 10nm , 12nm , 16nm , 20nm , 22nm , 28nm , 28nm HP , 28nm HPC , 28nm HPCP , 28nm HPL , 28nm HPM , 28nm LP , 40nm G , 40nm LP , 45nm GS , 45nm LP , 55nm FL , 55nm G , 55nm GP , 55nm LP , 55nm NF , 55nm ULP , 55nm ULPEF , 55nm UP , 65nm G , 65nm GP , 65nm LP
UMC
Pre-Silicon: 14nm , 28nm , 28nm HLP , 28nm HPC , 28nm HPM , 28nm LP , 55nm , 65nm LL , 65nm LP , 65nm SP
×
Semiconductor IP