One testbench for PSS and Transient Simulation with different stimuli.

 Dear Thomas,

> . Is there a way, to setup both in one testbench ? For DC-simulations, I can just set the DC value

> of the vwpl source to 0 and everything works as expected. I haven't found a way to do the same for

> PSS simulations which leads to me always having to make a copy of the testbench where I then

> statically connect the powerdown node to ground. This has lead to mistakes in the past and I would

> prefer a single testbench that just works and can be reused.

I had a thought that I often use in my test benches to allow them to be used for many analysese that might be helpful and allow you to use the same test bench for a PSS simulation. When there are multiple stimuli, I include a design variable for each source to define its delay. For example, in your test bench with a powerdown piecewise linear source, I might invlude a delay variable "delay_powerdown" in the definition of the source as shown in the attached example. In your pss simulation, if the value of delay_powerdown is set to a value that far exceeds the value of TSTAB,  the source will remain at 0 throughout your PSS simulation. For a transient simulation, you might set delay_powerdown to something less than the TSTOP time of your simulation to study the impact of powerdown.

I hope I understood your queston!

Shawn

Shawn,

Actually that won't work. The tstab is additional time added on to Tonset (the onset of periodicity). It will simulate for as long as needed for the sources to be periodic, and then the tstab. So with a tstab of 1u, if you have a source with a delay of 1m, it will simulate 1m+1u for the initial transient.

Thomas - a couple of approaches I'd suggest:

1. Put a design variable on the "stretch" parameter of the vsource (a PWL parameter). This allows you to scale the time values in the PWL. So for example, I have this netlist (easier to show a netlist than a picture of the parameters):
   
   parameters high=0 timefactor=1e-15
   
   v1 (n1 0) vsource type=pwl wave=[0 0 1u 0 1.01u high 2u high 2.01u 0 3u 0 3.01 high] stretch=timefactor
   vclk (n2 0) vsource type=sine freq=500M ampl=1
   
   pss pss fund=500M
   
   By setting high=0 and timefactor=1e-15 it scales all the times to be tiny, and the amplitude variation to be 0. You can't set timefactor to 0 unfortunately. If you set high to be your desired peak value and timefactor=1, the PWL will behave as you want it for the tran analysis.
2. Add two sources in your testbench - in parallel. One that you'll use for tran and one for PSS. Then create two config views for the same schematic, and use the instance binding "bind to open" for one of the sources in one config, and for the other source in the other config. Then you simulate using the config views. Bind to open means that the instance will be omitted from the netlist. One schematic to maintain, and two very simple configs. You can use the same approach to switch in and out other testbench components.

Regards,

Andrew.