how to test stability for transient analysis

[surianova]

Hi!

I have a problem on how to test the stability in transient analysis.
Let say i have a opamp with 10 degree phase magin for Ac analysis
in open loop, which is consider unstable when use in negative feedback.
How i want to test it in transient so that i can see oscillation? Is it put a pulse at the input? what frequency to test? Pls advice.. .Thanks in advance..

Suria

 

[VSWR]

Stability check using transient analysis is a good way to verify the open loop AC analysis. Insert a step at the input of the loop and watch how the loop settles, you will probably see a damped sinusoidal response.

I've done this in SPICE many times for verification.

But 10 deg phase margin is very low. Aim for 60 deg phase margin and 6 dB amplitude margin to be safe. Otherwise you may have problems during temperature variations, tolerance and long time drift of your design.

 

[surianova]

Stability check using transient analysis is a good way to verify the open loop AC analysis. Insert a step at the input of the loop and watch how the loop settles, you will probably see a damped sinusoidal response.

I've done this in SPICE many times for verification.

But 10 deg phase margin is very low. Aim for 60 deg phase margin and 6 dB amplitude margin to be safe. Otherwise you may have problems during temperature variations, tolerance and long time drift of your design.

Ya. I have try with a step input of 1 ps and it show the oscilltion with a phase margin of 10 degree. The oscillation frequency is ard 200 Mhz.
And unity gain bandwith frequency is 1GHz. How is the oscillation frequency related to gain bandwidth frequency? Thanks.

 

[santhoshv78]

The unstable closed loop system oscillate in its natural frequency, Wn.
The -3db bandwidth will be arround two times the natural frequency depending on the damping factor.
what is the gain of the opamp.

 

[surianova]

any formula or equation? My opamp gain is 40dB.

 

[zareh]

put opamp in unity-gain configuration,then apply a pulse with small amplitude because large amplitude bring another limitation of opamp (slewrate).a nominal amplitude can be 100mv
if opamp has small phase margin then output pulse rings and finally settles.
a good design should have at least 67 degree of phase marign and do not rings in step response.

you can also apply a sine input and measure(see) output delay with respect to input which in this case(10 degree phase margin) output is close to inverted input ( delay=.472T , T=perid)

 

[layes]

just run it

/Warning #7 - Do not post useless messages. (klug)/

 

[drasta]

Best way is pulse on the input and you can see the ringing because under 60 degrees phase margin is too low. You can also do furier analyses and see spectrum of the output signals and if you will have peaks no only on working frequency but also on others.