help needed,Stability analysis and noise analysis

[ASICK]

I need help with the Stability analysis and noise analysis, as how to run these using Spectre in cadence.
1. During stability analysis, what to choose for probe instance, and
how does this analysis work? if you could explain that will be great.
2. During noise analysis, there are 2 options, like input noise and output noise and for each there is option of choosing prove, voltage nd current. how does this analysis work? and what are these options about? I must have asked a lot at one time, but will really appreciate ur help and time,
thanks

 

[sunking]

1. open loop
The phase margin(Gain =0)
normal PM>45
2. read the candence open book

 

[starcoming]

1 it is better to be sure a pm>60 at the Unit GB point.

 

[victoriya]

stability for what kind of circuit?
meanwhile the pm have to be>70

 

[threekingtiger]

There are two ways to test the stability of the loop. One is that you can open your loop by inserting the testing elements, which are a very very large ideal cap, a very very large inductance and an AC small signal source, then perform AC analysis to see the AC maganitude and phase. You will get the BODE plot in which the PM, GM and GBW can be easily obtained. The other way more simple is using the STB analysis in spectre. Break the loop by inserting an iProbe from analoglib library. Perform the stb anlysis. Initiate the waveform window and PM GM and GBW can be read. Also this analysis can give you the bode plot. The latter method is what I used often in practice.

 

[leo_o2]

Insert a voltage source into the loop. The voltage source is used to break the loop in stb analysis.
Output noise can be converted into an equivalence of input noise. And input noise can be converted into an equivalence of output noise too.

I need help with the Stability analysis and noise analysis, as how to run these using Spectre in cadence.
1. During stability analysis, what to choose for probe instance, and
how does this analysis work? if you could explain that will be great.
2. During noise analysis, there are 2 options, like input noise and output noise and for each there is option of choosing prove, voltage nd current. how does this analysis work? and what are these options about? I must have asked a lot at one time, but will really appreciate ur help and time,
thanks

 

[jjx]

It could also be mentioned that rather than using the large L and C approach as per above, you have a couple of sptXswitch cells available in the analogLib. They are quite powerful (once one learns how to understand what the position buttons mean ...)

And indeed, the definition of phase margin is quite system dependent... There are quite a few rule-of-thumb's out there, but traditionally they consider only the two-pole idealized model. For example, certain comparators and regulators would follow other "laws".

 

[LvW]

..................
And indeed, the definition of phase margin is quite system dependent... There are quite a few rule-of-thumb's out there, but traditionally they consider only the two-pole idealized model. For example, certain comparators and regulators would follow other "laws".

Hopefully, nobody in the forum believes this.
There is only one single and general definition for the phase margin, which is applicable for any system with feedback.
This definition - and several methods to determine the margin - was presented and discussed several times in this forum.

 

[jjx]

Yes, you're right, I was referring to the target value of the phase margin (45, 60, 70, etc), not the definition as such.

 

[LvW]

Yes, you're right, I was referring to the target value of the phase margin (45, 60, 70, etc), not the definition as such.

OK, that's right. There are, of course, different requirements as far as the target specification for the PM is concerned.