The Principle of Automatic Sweeping

All "auto" means is that it will take 50 steps (so 51 points). From the "spectre -h dc" help this is the same as not specifying the number of steps or step size, and it picks either Lin or log depending on how far apart the start and stop are:

In addition, you can specify a
step size parameter (`step', `lin', `log', or `dec') and determine whether the
sweep is linear or logarithmic. If you do not specify a step size parameter,
the sweep is linear when the ratio of stop to start values is less than 10 and
logarithmic when this ratio is 10 or greater.

There's no comprehension of how abrupt the output change is - that's pretty much irrelevant for a dc sweep anyway (it doesn't help the analysis to know this). So "auto" is actually quite simple...

Andrew.

All "auto" means is that it will take 50 steps (so 51 points). From the "spectre -h dc" help this is the same as not specifying the number of steps or step size, and it picks either Lin or log depending on how far apart the start and stop are:

In addition, you can specify a
step size parameter (`step', `lin', `log', or `dec') and determine whether the
sweep is linear or logarithmic. If you do not specify a step size parameter,
the sweep is linear when the ratio of stop to start values is less than 10 and
logarithmic when this ratio is 10 or greater.

There's no comprehension of how abrupt the output change is - that's pretty much irrelevant for a dc sweep anyway (it doesn't help the analysis to know this). So "auto" is actually quite simple...

Andrew.

Dear Matsuge,

Matsuge said:

I've already known the rough value of the threshold(which is under 1V), so the sweeping steps need to be fine only around the threshold.

However, I also have to check the leak current at the input voltage = 0V/3.5V/5.5V.

So I want set the sweeping range from 0V to 5.5V,

but if I run this simulation with linear sweep of 1mV step from 0V to 5.5V, it takes over 1 hour.

If you are truly interested in saving simulation time, I might suggest you use Assembler or Explorer and set the DC voltage in a corner simulation using an "inclusion" sweep. In this fashion, you can set the inverter input voltage in coarse steps at its extrema, but include a much smaller voltage step over the input voltage range where you expect the inverter threshold voltage.

Shawn

If you were using a dc sweep, you could also use the "Add specific points" checkbox to add a few points around the expected threshold. However, you have to list each additional point explicitly.

The benefit of the method that Shawn proposed is that you can have multiple (overlapping) sweep ranges and ADE will combine them. So you could do (and you can have more than two sweep definitions and they don't have to be the same type):

Note that in general a sweep directly within the simulator (e.g. a dc sweep) will be faster than doing it in Explorer/Assembler, but if it's taking 1hr for a 50 point sweep, then the overhead of running them separately is unlikely to be an issue.

Also with Explorer/Assembler you then have the option of running multiple points in parallel which could bring the simulation down (if you have enough compute resource and simulation licenses available.

Andrew

Thank you Andrew and Shawn.
I didn't expect that automatic sweep principle is just about 50 steps sweep.
Adding specific points seems to suit this case.

I will also keep in mind about Inclusion method and the speed difference between simulating within and out the simulator.

Thank you so much.