The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> automated simulation of hysteresis
https://designers-guide.org/forum/YaBB.pl?num=1236026290

Message started by analogue_guy on Mar 2^nd, 2009, 12:38pm

Title: automated simulation of hysteresis
Post by analogue_guy on Mar 2^nd, 2009, 12:38pm

hi

how do you simulate hysteresis behaviour automatically?

cheers

Title: Re: automated simulation of hysteresis
Post by Andrew Beckett on Mar 2^nd, 2009, 10:40pm

The conventional way is to do a DC sweep in one direction, and then a second in the opposite direction. Spectre can do this simply by adding hysteresis=yes to the dc analysis (there's a checkbox on the dc form in ADE for this too). You can also just do two dc analyses, specifying the sweeps in different directions (from a netlist).

Regards,

Andrew.

Title: Re: automated simulation of hysteresis
Post by analogue_guy on Mar 3^rd, 2009, 4:35am

Andrew Beckett wrote on Mar 2^nd, 2009, 10:40pm:

Thanks for the answer, formerly I used always transient simulation with a voltage ramp and I wanted a setup to run plenty of monte carlo runs and just keep the hysteresis window value after every runs.

Title: Re: automated simulation of hysteresis
Post by ywguo on Mar 3^rd, 2009, 10:04pm

Andrew,

I simulated hysteresis as analogue_guy did. A slowly rising/falling slope was applied to the input of DUT.

The DC hysteresis sweep is new to me. In my mind, the simulator solves the operating point of DUT in each step of a DC sweep independently. For eg., let the simulator sweep from 1V to 2V by 0.1V step. It will solve the operating point of 1.0V, 1.1V ... 1.9V, 2V respectively and independently. As well known, the hysteresis is not memoryless. The output is strongly decided by the previous state. How does spectre simulator solve this problem by DC hysteresis sweep?

Yawei

Title: Re: automated simulation of hysteresis
Post by boe on Mar 4^th, 2009, 1:18am

Yawei,
Quote:

... It will solve the operating point of 1.0V, 1.1V ... 1.9V, 2V respectively and independently. As well known, the hysteresis is not memoryless. The output is strongly decided by the previous state. How does spectre simulator solve this problem by DC hysteresis sweep?

Using solution of previous step as initial guess for next step reduces computation time and provides previous state.
BOE

Title: Re: automated simulation of hysteresis
Post by Andrew Beckett on Mar 4^th, 2009, 1:30am

That's right - the starting point for each point in the sweep is the previous point, and so you will generally see the hysteresis. Of course, if the hysteresis is based upon some capacitative charge storage, then DC won't do - but in many cases it's sufficient (and quick!).

Regards,

Andrew.

Title: Re: automated simulation of hysteresis
Post by analogue_guy on Mar 4^th, 2009, 7:57am

Andrew Beckett wrote on Mar 4^th, 2009, 1:30am:

Yes, the DC sweep works, hysteresis function can be seen. But comparing the results with the transient results I can see that the circuit is less sensitive for statistical variations (using the MC).

But could you recommend me any technique to translate these results (either the DC sweep or transient) to only a number in every run (basically the difference between the two values) and I could deal only with these number afterwards.

At the moment I save the whole waveforms in both cases (DC sweep and transient). But it required more time, also post processing, so it is not that comfortable.

Cheers

Title: Re: automated simulation of hysteresis
Post by boe on Mar 5^th, 2009, 1:25am

Analogue_guy,
You can use function "cross" to determine where the comparator output changes.
Is that what you are looking for?
BOE

Title: Re: automated simulation of hysteresis
Post by analogue_guy on Mar 5^th, 2009, 12:16pm

boe wrote on Mar 5^th, 2009, 1:25am:

Analogue_guy,
You can use function "cross" to determine where the comparator output changes.
Is that what you are looking for?
BOE

Thanks, I looked for this kind of solution. I will try to do so.
Cheers