The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> THD simulation
https://designers-guide.org/forum/YaBB.pl?num=1165375335

Message started by linkernzhang on Dec 5^th, 2006, 7:22pm

Title: THD simulation
Post by linkernzhang on Dec 5^th, 2006, 7:22pm

Hi ,all
Now I'm doing a simulation of a variable gain amplifier .I meet a problem in running THD simulation .I 'm not familiary with the THD simulation methord. Could anyone give me some suggestion on the function ?
Thx!
Regards
Linkern

Title: Re: THD simulation
Post by sheldon on Dec 6^th, 2006, 6:23am

Linkern,

There two basic non-idealities that analog designers need to consider
when designing: noise and non-linearity. THD is a measure of non-linearity
or distortion. To measure it, input a sine wave and measure the ratio of
the power in the fundamental to the power in the distortion tones. If you
have access to ADE, there is a calculator function to measure THD.

Best Regards,

Sheldon

Title: Re: THD simulation
Post by linkernzhang on Dec 7^th, 2006, 6:50pm

Thank you Sheldon!
Yesterday I have tried to simulate the THD in cadence. I iput a sine wave , then plot the output waveform with the DFT function in the calculator. But I got two different results when I use different timing step . May be I should go back to re-consider the Fourier transform concept to figure the reason. Thank you again!

Regards!
Linkern

Title: Re: THD simulation
Post by sheldon on Dec 8^th, 2006, 10:59pm

Linkern,

Assuming that the simulator tolerances have been set appropriately, then
you are probably seeing interpolation error. The DFT samples the waveform
at regular time intervals. In general, the time point that the simulator choses
are not the same time points that the DFT needs. So the DFT interpolates the
simulator results to estimate the time points required for the DFT. By using
different times steps you are reducing the error in the estimation, interpolation
error. There are many things to try

1) Easy
Set max_step to less than 1/2 the time step size of the DFT, this can be
computationally expensive, it makes simulation time longer but usually
gives good results.

2) More accurate
add a zvcvs to the output to sample the output and then perform the
DFT on the sampled output.
zdft (sampled 0 out 0) zvcvs td=10u ts=10n

You will need to add a little extra time to the simulation and shift the
DFT times to compensate for the delay in the zvcvs.

3) Use the Fourier integral it is not sensitive to interpolation error
fourTransform ( out 0 ) fourier fund=1k points=256

Good Luck,

Sheldon