TR (Transient) Options

• Integration formula
  

  Euler	Damping effects may cause false results for energy relationships.
  Adaptive Trapezoid-Euler	Default integration formula. Used for oscillating LC systems or analog oscillators. This is especially useful for investigating energy problems.
  Trapezoid	Pure trapezoidal integration formula (no switching to Euler like the Adaptive Trapezoid-Euler formula). For purely oscillatory circuits with no discontinuities, this method gives accurate results and may be desirable. The method is less stable particularly in cases with discontinuities or non-zero derivatives at time=0.

  
  
• Local truncation error[%] - LDF

  Defines the acceptable truncation error value. This value controls the time step depending on the dynamic nature of the circuit. A small error increases the precision but also the calculation time. This number itself does not control the precision of the overall result. The default value is 1.

• Maximum number of iterations - IteratMax

  Maximum number of iterations allowed per simulation step. If convergence problems occur, the calculation for the active simulation step stops when the maximum value is reached without consideration of other error limits. The simulation then proceeds to the next step. The default value is 40.

• Absolute tolerance - RHS

  Absolute tolerance for RHS vector (residual vector). Used to determine the acceptable error (convergence criteria) for the RHS vector in the Newton-Raphson method. A lower value leads to more accurate results, but convergence problems might be possible in the Newton-Raphson iterations. The default value is 1m, where m stands for milli(10^-3).

• Absolute tolerance - LHS

  Used to determine the acceptable error (convergence criteria) for the solution vector update in the Newton-Raphson method. A lower value leads to more accurate results, but convergence problems may occur using the Newton-Raphson iterations. The default value is 1m, where m stands for milli(10^-3).

• Relative tolerance[%]

  If enabled, the simulator uses the specified relative tolerance percentage along with the absolute values (Maximum Current Error and Maximum Voltage Error) to determine acceptable error, and to determine convergence.

  Note:

  In case of non-convergence during transient simulation (that is, the maximum number of iterations has been exceeded), see Fixing Non-Convergence in Twin Builder.

• Apply operating-point convergence scheme

  Achieving convergence at the transient operating point (at t = 0) can be difficult for circuits with highly nonlinear models. In situations where convergence fails at the operating point, this option applies the continuation methods Gmin-stepping and Source-stepping to improve convergence.

  The Gmin-stepping method adds a large conductance value to ground for each node so that the nonlinear behavior is damped and sparse matrix becomes well-conditioned. It then gradually lowers the conductance value until the conductance becomes negligibly small.

  The Source-stepping method starts solving with zero value for all independent voltage and current sources, and continues to solve by slowly ramping their values until they reach their full values.

  Note:

  Gmin-stepping is available only for the default sparse matrix solver (not for new sparse matrix code).

  Source-stepping is applied only for SML type independent sources.

• Analog/Digital synchronization

  Mixed-signal synchronization ensures timely exchange of values between the analog and digital sub-simulators when simulating systems with both digital (VHDL) and analog (VHDL-AMS and SML) models or constructs. This simulation option controls the synchronization strategy used. Although each of the strategies below will yield correct results, correctly selecting the best synchronization strategy for the system to be simulated yields the fastest simulation.

  
  

  Hybrid	Default setting. Employs an optimistic digital and a conservative analog synchronization strategy. Results in faster and more efficient simulations than Conservative synchronization by reducing the number of unnecessary analog solution points. Yields the fastest simulations when the digital sub-system is smaller than the analog sub-system. Digital simulation events occur at a higher frequency than analog steps, and not every digital event affects the analog sub-system.
  Adaptive	Adaptively controls the amount of optimism in the optimistic digital simulation intelligently depending on the system. Yields the fastest simulations for systems with small fast digital systems. When the digital sub-system is larger than the analog sub-system, digital simulation events occur at a higher frequency than analog steps, and not every digital event affects the analog system.
  Conservative	Synchronizes at the minimum time step requested by each sub-simulator. Requires that the analog system simulation be solved for every digital system event, whether the analog system is affected. Yields the fastest simulations for systems with tight feedback loops requiring frequent synchronization (that is, when every digital event affects the analog sub-system).

• Advanced step mode

  Switches between the advanced and standard step size algorithms. By default, the improved algorithm is used. Clear the check box to use the standard algorithm. With Advanced step mode selected, the step size is more dynamic – increasing and decreasing at a faster rate compared to the standard algorithm. Using Advanced step mode leads to faster simulation speed especially when a small HMin is necessary for good accuracy, but only at limited points throughout the simulation. For best results with this algorithm, there should be sufficient separation between HMin and HMax.

  If the system needs a small timestep at a lot of points, the standard algorithm could be the better choice. This is because the advanced step mode causes the simulator to first overrun the required simulation points, then rollback to hit it. This increases the total number of iterations and slows down the simulation speed.

  Number of equal steps	When Advanced step mode algorithm is selected, this parameter sets the number of consecutive time steps of equal size that must be taken before a larger step size is used. Default value is zero. An integer value greater than zero should be used.
  Step acceleration damping [%]	When Advanced step mode algorithm is selected, this parameter sets the rate of damping of the step size acceleration for the next time step. Default value is zero. A value between 0-100% should be selected. A lower value for this parameter allows higher simulation speed; use a higher value to increase simulation stability.

• Samanskii factor

  The Update Jacobian after ___ iterations setting is an optimization method where the simulator only re-evaluates the Jacobian Matrix after the specified number of iterations. Default is one (1) which indicates that optimization is turned off. A value between 1 and 10 is allowed. In case of systems requiring a very large number of iterations per time step for convergence, a value greater than 1 can improve the performance of the simulator by avoiding Jacobian evaluations and matrix factorizations.

  Note:

  A very high value may lead to non-convergence. Ideally use a value between 2 and 6.

• Multithreaded Solver

  Uses multiple threads to perform model evaluation for all the models in the circuit (analog) solver. Model evaluation involves computing model equations with updated solution vector, and Jacobian and residual computations. Each thread processes one model at a time. As a result, if there are N threads, N models can be evaluated in parallel in a multi-core machine. Given the overhead of processing multiple threads, the real performance benefit can usually be realized when there are large numbers of models involved, or when there are complex models that require significant computational resources.

  Enable multithreaded solver	Uses multiple threads for the solver computation.
  Auto compute number of threads	Determines how many threads required – based on physical CPU cores available and number of circuit simulator models. If your machine has fewer than four cores, the solver disables multithreading for this option.
  Specify number of threads	If the above option is not desired, you can specify how many threads to use. The actual number is limited by circuit simulator model count.

   

  Note	Several models such as VHDL and SPICE are processed in serial in this version due to thread safety issues. However, in future releases, more models will be available for parallel processing.
  Note	By default, all user-defined C-Models are processed in serial mode unless an explicit C-Model API function call, SetAsThreadSafe, is made. For more details, see CModUser Object Methods.