The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Modeling >> Behavioral Models >> LDO modeling, nonlinear dynamics
https://designers-guide.org/forum/YaBB.pl?num=1213723670

Message started by srikanthpam on Jun 17^th, 2008, 10:27am

Title: LDO modeling, nonlinear dynamics
Post by srikanthpam on Jun 17^th, 2008, 10:27am

I am trying to model an LDO at a top level in a behavioral fashion. I was able to model the static charecterisitcs added with the input volt. to output volt., reference to ouput volt. and load current to output volt. transfer functions, using the laplace functions provided by verilog-a.
I realized in a real LDO (esp with a PMOS device) the load, input to output and reference to output transfer functions change with load current. But verilog-a doenot allow to use variables or functions as parameters in lapace funcions.
Can anyone plese help in how to include load/input voltage depended dynamics ( variable transfer functions).

Title: Re: LDO modeling, nonlinear dynamics
Post by Eugene on Jun 18^th, 2008, 12:04am

If you are trying to model static nonlinear elements, you can enter the dependency directly. However, if you trying to model a transfer function that varies with load current, I would try to model the RC elements directly instead of the resulting Laplace transfer function. I think it will be easier to make the RC elements depend on the desired variable.

Title: Re: LDO modeling, nonlinear dynamics
Post by srikanthpam on Jun 18^th, 2008, 10:29am

Thanx Eugene, for your reply.

Since its a closed loop system, the effective outpt impedence and other transfer functions may not be possible to represent as RC network. (May be it can be done, but definelty looks more complex)

I tried to describe using time domain (differential ) equations insted of freequency domain (transfer functions) description.

Here we can describe the dependence of the coefficients of the differential equations with load.

From what I understand in verilog-a only linear dynamics can be described in frequency domain. This is not true with all simulators, for example matlab/simulink supports the transfer functions with variable coefficents.

Title: Re: LDO modeling, nonlinear dynamics
Post by sheldon on Jun 19^th, 2008, 6:22pm

Srikanthpam,

One other approach you might want to try is building a more structural
model. An LDO is "beautiful" analog circuit. It looks very simple but the
behavior is very complex. For this class of circuits it can be difficult to
create pure behavioral circuits since the behavior is "rich" . The other
issues is that the characteristics of interest for a designer are almost entirely
the complex behaviors that are hard to describe. Since the structural model
for an LDO is an op-amp with non-linear resistor in the feedback path, the
overhead of a structural model is minimal.

Best Regards,

Sheldon

Title: Re: LDO modeling, nonlinear dynamics
Post by Srikanth on Jun 23^rd, 2008, 1:41pm

Hi Sheldon,
Thank you.
I think I agree with you...to really capture the "rich" behavior...."structural" is the way to go....while we can have much simpler models when all such behaviors are not required to be modeled....while i was able to make behavioral level model, with the non-linear dynamics...using differential equations with the coefficients as functions of time....its not very readable....simple structural model captures all the intended behaviors and is very readable....without much simulation overhead.

Srikanth

Title: Re: LDO modeling, nonlinear dynamics
Post by buddypoor on Jun 24^th, 2008, 1:19am

Hi Srikanth,

I really don´t know if my reply can help. Nevertheless, I attach a document with my LDO model together with simulation results (time domain). I have modelled the npn-transistor output current as a function of Vce and the opamp as a single pol device. The simulation shows the response to a step change of the reference voltage, of the input voltage as well as of the load. The program I´ve used is VISSIM.