sp1tswitch equivalent for pac simulations

Hi Nicola,

No, this is not possible (nor would it be a good idea as it would break the analysis). What are you trying to do? Are you using harmonic balance or shooting mode of pss?

Andrew.

Hi Nicola,

No, this is not possible (nor would it be a good idea as it would break the analysis). What are you trying to do? Are you using harmonic balance or shooting mode of pss?

Andrew.

It's a shooting pss. I'd like to check the voltage gain in pac sampled simulation between nodes N1-N2 as in the sch drawn here https://community.cadence.com/cadence_technology_forums/f/custom-ic-design/43477/noise-jitter-transfer-function-along-clock-driven-inverter-chain
I'm not sure that with a PXF sampled (with Stimuli option "nodes_and_terminals" selected) in Voltage Gain mode in the Direct Plot Form, when I select the output terminal of the first inverter, this is giving me the voltage gain between node N1 and node N2 (provided that in PXF sampled form I put the node N2 as output)...

Hi,
any update on this topic? How should I perform a simulation like this?

If this is urgent, you should contact customer support. My availability is rather limited this week (I'm on vacation most of it, and at a customer yesterday and today).

Did you try using the voltage gain on the direct plot form for sampled pac? Similarly I think that using the voltage gain with nodes and terminals (you'd have to make sure you're saving the output terminal of the inverter connected to node N1) would do the right thing.

Actually thinking about it, I'm not sure measuring the voltage gain with PAC would be correct here, because that works by measuring the voltage at the numerator and denominator nodes, which would be at the sampling instant. The gain is not really instantaneous, and so the signal level at the N1 output when the output of N2 is switching is probably not terribly meaningful.

Andrew.

Hi Andrew,

sorry for the late reply, I'm back on the topic since a few days. 
The basic goal was to check the transfer functions between various noise sidebands from one inverter output to the following one, all superimposed e.g. in 1kHz - 15GHz bandwidth. You are saying that is not possible to do? What kind of calculation spectre does to get output spectrum at each node? I thought it was calculating the gain by linearizing the inverrter at the transition points (and then there had to be a method to check this gain).

Thank you
Nicola

Hi Nicola,

I think that using pxf with nodes and terminals should do the right thing. That was what I was asking you whether you'd tried. I don't think you can do this in a single simulation though, because the sampled analyses are going to strobe the output at a particular time, and so you'd need to strobe the output of each inverter at the threshold crossing - plus for pxf, it would need the output to be for each inverter. 

The simulation does not work by linearising the inverter at the transition points. It linearises the inverter across the entire period (so it's a time-varying linearisation), and it's just the output that is sampled at the particular time when the output signal crosses a threshold.

Andrew.