Fast and Accurate Design Methodologies using Fully Event-Driven Simulations for non-ideal Mixed-Signal Systems

针对非理想混合信号系统使用完全事件驱动仿真的快速、准确的设计方法

• 批准号: 406675560
• 负责人: Professor Dr.-Ing. Ulrich Hilleringmann
• 金额: --
• 依托单位: Abteilung Advanced System Engineering
• 依托单位国家: 德国
• 项目类别: Research Grants
• 财政年份: 2019
• 资助国家: 德国
• 起止时间: 2018-12-31 至 2022-12-31
• 项目状态: 已结题
• 来源: https://gepris.dfg.de/gepris/projekt/406675560?language=en
• 关键词: Fast; Accurate; Design; Methodologies; using

Nowadays wireless communication is becoming more important and need the implementation of power efficient and high density ICs. Hence, mixed-signal systems like DC-DC converters or phase-locked loops (PLL) are increasingly integrated.These systems mix analogue and digital parts and exhibit a nonlinear pulse-width modulated (PWM) switching behaviour. Thus their design cannot rely on general feed-back theories and simulations at transistor or behavioural levels are often used. Transistor-level (TL) models are very accurate, whereas behavioural models allow an easier handling of parasitic effects. The main bottlenecks of these approaches are the long processing times, the huge computational consumption and the overall design costs. Thus a quick insight into the system behaviour is made difficult endangering the guaranty of a robust design. In addition, at TL all non-ideal effects are considered at once, making the analysis of their singular and combined influences on the performances even more complex.To get a quick overview of the system properties, design methods based on linear models are used. Because of the early linearization, nonlinear and non-ideal effects are not considered. However, it is crucial to take them into account for a robust design, since they affect the stability, frequency purity and transient dynamic behaviour. Additionally, the validity of these modelling approaches is limited to the domain close to the operation point, discarding them for transient pull-in predictions which are essential when considering frequency synthesis.To combine the accuracy of TL simulations and the efficiency of linear models, a mixed-signal system can be represented by an event-driven (ED) modelling approach. Based on the example of a charge pump (CP) PLL, an ED model considering the major non-ideal effects is introduced here. This model evaluates only the triggering edges of the PWM procedure, reducing drastically the simulation time and the amount of produced data. Speed-up factors up to 100000 are expected with a very good accuracy compared to TL simulations.In this project, to overcome the design challenges of mixed-signal systems, i.e. switching behaviour, non-ideal and nonlinear behaviour, the efficient ED model will be advanced to cover the CP-PLL, clock and data recovery PLL and Delay-Locked Loop applications. All major parasitic and nonlinear effects will be extracted at TL and modelled within the ED representation. In addition, an interface between the ED model and Cadence will be implemented in order to realise a practical parameter extraction. Once the ED model is enriched by these parasitic effects, their influence on the loop performances will be systematically explored, leading to an exhaustive and more robust design methodology of mixed-signal PLLs and DLLs. To show the feasibility of this modelling and design methodology, the gathered knowledge will be extrapolated to other mixed-signal systems like the DC-DC converter.

在无线通信日益重要的今天，需要实现低功耗、高密度的集成电路。因此，混合信号系统，如DC-DC转换器或锁相环（PLL）越来越集成。这些系统混合了模拟和数字部分，并表现出非线性脉宽调制（PWM）开关行为。因此，他们的设计不能依赖于一般的反馈理论和模拟在晶体管或行为水平经常使用。晶体管级（TL）模型非常准确，而行为模型允许更容易地处理寄生效应。这些方法的主要瓶颈是处理时间长、计算量大和总体设计成本高。因此，对系统行为的快速洞察很难危及稳健设计的保证。此外，在TL中，所有的非理想效应都被同时考虑，使得分析它们对性能的单一和综合影响变得更加复杂。为了快速了解系统属性，使用了基于线性模型的设计方法。由于早期线性化，没有考虑非线性和非理想效应。然而，将它们考虑到稳健设计是至关重要的，因为它们影响稳定性，频率纯度和瞬态动态行为。此外，这些建模方法的有效性仅限于工作点附近的域，在考虑频率合成时至关重要的瞬态拉入预测中丢弃它们。为了结合TL仿真的准确性和线性模型的效率，混合信号系统可以用事件驱动（ED）建模方法来表示。以电荷泵（CP）锁相环为例，介绍了考虑主要非理想效应的放电模型。该模型仅评估PWM过程的触发边缘，大大减少了仿真时间和产生的数据量。与TL模拟相比，预计加速因子高达100000，精度非常高。在本项目中，为了克服混合信号系统的设计挑战，即开关行为，非理想和非线性行为，将推进高效ED模型，以涵盖CP-PLL，时钟和数据恢复PLL以及延迟锁相环应用。所有主要的寄生和非线性效应将在TL提取，并在ED表示中建模。此外，为了实现实用的参数提取，将实现ED模型与Cadence之间的接口。一旦这些寄生效应丰富了ED模型，它们对环路性能的影响将被系统地探索，从而导致混合信号锁相环和dll的详尽和更稳健的设计方法。为了展示这种建模和设计方法的可行性，所收集的知识将被外推到其他混合信号系统，如DC-DC转换器。