The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> Fourier Analyzer of Spectre > analogLib
https://designers-guide.org/forum/YaBB.pl?num=1274212958

Message started by samiran on May 18^th, 2010, 1:02pm

Title: Fourier Analyzer of Spectre > analogLib
Post by samiran on May 18^th, 2010, 1:02pm

Dear all,

I am new to the fourier analysis through a circuit simulator. According to my study (I have gone thru Ken Kundert's book), I have learn that analogLib's fourier component is more accurate (at least form anti-aliasing point of view) than the waveform calculator's dft function. And for my need I want to use fourier analyzer. But I am facing some problem in getting things on right path.

As initial startup I just took a sinusoidal voltage source of 1mVp-p, 1KHz and connected it to the fourier component in single-ended fashion. I have set the fourier components parameter values as follows:

Model name: test
Number of Harmonics: 1
Fundamental Frequency: 1KHz
Minimum no. of time points: 100
Active: Yes
Order of interpolation: 10
No. of reference Harmonics: 1
Scale factor: 1

I have set these values based on my assumption as I don't know the relationships.

I am totally confused as how I should progress? I have run a transient simulation of 1mS but encountered a fatal error.

Please help me!

Regards
Samiran

Title: Re: Fourier Analyzer of Spectre > analogLib
Post by sheldon on May 18^th, 2010, 8:04pm

Samiran,

Could you share the error message? It would be useful to understand
what the simulator thinks is happening.

Best Regards,

Sheldon

Title: Re: Fourier Analyzer of Spectre > analogLib
Post by samiran on May 18^th, 2010, 10:44pm

Sheldon,

Following is the error I got:

Error found by spectre during circuit read-in.
input.scs: FA1 is an instance of an undefined model test.

spectre terminated prematurely due to fatal error.

FA1 is the instance name I have given.

If possible, could you please explain what should be the values of different model parameters? And also explain after setting the parameter values, what are the steps (I know I have to do a transient simulation in order to acquire the time points) ?

Cheers!
Samiran

Title: Re: Fourier Analyzer of Spectre > analogLib
Post by sheldon on May 20^th, 2010, 7:34am

Samiran,

　　　Maybe you should delete the model name.　You have specified
that you want
1) One harmonic, the fundamental
2) The fundamental frequency is 1kHz
--> so the transient analysis needs to be longer than
1ms
3) You have specified 10th order interpolation
--> I would leave this blank
4) The maximum number of time points is 100

Setup seems reasonable for the current test conditions.
Sorry do not see a parameter like number of reference
harmonics in the on-line help.

Best Regards,

Sheldon

Title: Re: Fourier Analyzer of Spectre > analogLib
Post by samiran on Jun 1^st, 2010, 3:25am

Hi Sheldon,

I have tried the same you suggested but without any success! It still gives the same error. I can't remove the name FA1, because it's the instance name for the Fourier Analyzer. Please help me.

Basically my intention is to do a linearity analysis of a filter circuit. Please, please let me know the steps to do it in cadence. And how to collect the relevant data. Or if you can redirect me to a good reference it will also help, I guess!

Regards
Samiran

Title: Re: Fourier Analyzer of Spectre > analogLib
Post by Ken Kundert on Jun 1^st, 2010, 10:41pm

The problem is with 'test'. Spectre is saying that you never defined the model 'test'. So you either need to stop telling the Fourier Analyser to use 'test' or provide a model statement for 'test'. Perhaps you should try it with out specifying the model name as 'test'.

-Ken

Title: Re: Fourier Analyzer of Spectre > analogLib
Post by samiran on Jun 7^th, 2010, 8:21am

Thank you Ken! It just worked fine without the model name! Thank you so much!

Could you please explain the role played by model name?

Regards
Samiran

Title: Re: Fourier Analyzer of Spectre > analogLib
Post by Ken Kundert on Jun 7^th, 2010, 1:01pm

The idea is that if you would like to perform Fourier analyses on many signals in your circuit, you could specify the common parameters on a single model statement rather than many instance statements. This is generally not practical in a design environment like Artist because it is difficult to incorporate model statements, but it works nicely when you are running outside a design environment from a netlist.

-Ken

Title: Re: Fourier Analyzer of Spectre > analogLib
Post by samiran on Jun 8^th, 2010, 6:24am

Hi,

The result of the fourier analyzer gets written in spectre.out inside psfdirectory. How do I access a particular data from the result for example, Total Harmonic Distortion by value through a OCEAN script?

Regards
Samiran.

Title: Re: Fourier Analyzer of Spectre > analogLib
Post by samiran on Jun 10^th, 2010, 6:16am

I have one more question regarding this: Does stop time of transient simulation have effect on the THD value? I mean, I know that for Fourier Analysis to be executed properly, stop time should be ≥ time period of the fundamental frequency, but if the stop time if more multiples of time period then would this change THD value.

Actually, I have observed such kind of thing happening. When I increase the stop time of transient analysis, then THD value decreases. Could you please explain why this is happening?

Regards
Samiran

Title: Re: Fourier Analyzer of Spectre > analogLib
Post by sheldon on Jun 19^th, 2010, 9:54pm

Samiran,

You can use the THD function from the calculator in an OCEAN script.
Could you provide more detail on the relationship between THD and stop
time that you are seeing? Does this occur for the circuit with the sine
wave input or for the filter? ...?

Best Regards,

Sheldon