Designer’s Guide Community

dandelion

Community Member

Offline

Posts: 98

A question on the step response
May 08^th, 2007, 7:14pm

Hi,
I know I can do the open loop analysis to investigate the stability. But in some times, we prefer to do the transient step response to see if it is stable.

I want to know, in doing the step response analysis, how should I select the rise up time and especially the amplitude of the step pulse?

Pls. give me some advice.

Thanks
Dandelion

Back to top

IP Logged

sheldon

Community Fellow

Offline

Posts: 751

Re: A question on the step response
Reply #1 - May 9^th, 2007, 7:00am

Dandelion,

Usually you use a fast edge compared to the DUT, for example,
1ns for a 10MHz op-amp. The pulse levels need to be small enough
not to induce large signal behavior, i.e., slew rate limiting, and large
enough to produce numerically accurate results. Usually +/- 10mV at
the output is good enough.

Could you clarify why you want to use transient analysis for stability
analysis? The only times I have used transient analysis for stability
analysis is when there are conditions that small signal will not identify,
for example, there are large signal phenomena that may effect the
stability. The issue in that case is that these are large signal simulations
and the general guidelines for small signal stability characterization do
not apply.

Best Regards,

Sheldon

Back to top

IP Logged

dandelion

Community Member

Offline

Posts: 98

Re: A question on the step response
Reply #2 - May 9^th, 2007, 7:51pm

sheldon wrote on May 9^th, 2007, 7:00am:

Dandelion,

� Usually you use a fast edge compared to the DUT, for example,
1ns for a 10MHz op-amp. The pulse levels need to be small enough
not to induce large signal behavior, i.e., slew rate limiting, and large
enough to produce numerically accurate results. Usually +/- 10mV at
the output is good enough.

� Could you clarify why you want to use transient analysis for stability
analysis? The only times I have used transient analysis for stability
analysis is when there are conditions that small signal will not identify,
for example, there are large signal phenomena that may effect the
stability. The issue in that case is that these are large signal simulations
and the general guidelines for small signal stability characterization do
not apply.

Best Regards,

� Sheldon
� �

Thanks for your reply, Sheldon. It is helpful to me.

Yes, we will do the transient analysis for the stability when there are large signal behavior.

Another reson we prefer to do it is to hope to avoid the loading effect when we are not sure how to break the loop properly.

Dandelion

Back to top

IP Logged

wave3x Junior Member Offline Posts: 10 Shang Hai, China.	Re: A question on the step response Reply #3 - Jun 2^nd, 2007, 6:34am You can use "pss+pstb" for your large signal stability analysis, if you have spectreRF. of course, transient is alway the robustest method to check stability.
Back to top	IP Logged