The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> Spectre's PAC: Frequency response of an amplifier
https://designers-guide.org/forum/YaBB.pl?num=1126179766

Message started by achim.graupner on Sep 8^th, 2005, 4:42am

Title: Spectre's PAC: Frequency response of an amplifier
Post by achim.graupner on Sep 8^th, 2005, 4:42am

Hi all,

I want to simulate the frequency reponse of an amplifier having a small non-linearity. Hence, the operating point depends on the input's amplitude. The shift in the operating point in turn changes the gain of a stage. This shifted gain in turn causes a move of the resonant frequency. This can be analyzed with multiple transient simulations folled by a fourier transform.
However a PAC analysis should yield the same information with much less effort. I use a vsin-source as stimulus with "PAC magnitude = 1". I vary the amplitude of this source as necessary but I have no clue what frequency to choose. The result of the PAC analysis changes when changing the frequency.
Does someone has an idea or can recommend a good book or paper on PSS and PAC, I am quite new to these analyses but found no good information source yet.
Thanks, achim

Title: Re: Spectre's PAC: Frequency response of an amplif
Post by Ken Kundert on Sep 8^th, 2005, 8:01am

Typically you measure the frequency response of an amplifier with an AC analysis. Why is this not sufficient in your case?

-Ken

Title: Re: Spectre's PAC: Frequency response of an amplif
Post by pancho_hideboo on Sep 8^th, 2005, 8:18am

Hi.

You should sweep large signal amplitude in Master PSS analysis, then do PAC analysis with pacmag=1.

Here is very simple example.

vin (in 0) vsource type=sine freq=fin ampl=amp pacmag=1 fundname="Fin"

sweeppss sweep param=amp start=1 stop=10 step=1 {
pss pss fund=100M harms=8 method=gear2only annotate=status
pac pac start=10 stop=50k dec=200 maxsideband=0 annotate=status
+ freqaxis=absout save=allpub
}

Title: Re: Spectre's PAC: Frequency response of an amplif
Post by achim.graupner on Sep 8^th, 2005, 10:52pm

Ken Kundert wrote on Sep 8^th, 2005, 8:01am:

Typically you measure the frequency response of an amplifier with an AC analysis. Why is this not sufficient in your case?

-Ken

Well, it happens that the circuit is slightly non-linear. In measurements it can be observed, that the resonant frequency depends on the input amplitude. I want to reproduce this behaviour in simulations.
So far I did transient simulations and varied both the source's frequency and amplitude and determined the output's rms-value after the circuit has been settled, a quite lengthy simulation.
- achim

Title: Re: Spectre's PAC: Frequency response of an amplif
Post by achim.graupner on Sep 9^th, 2005, 1:53am

pancho_hideboo wrote on Sep 8^th, 2005, 8:18am:

This is exactly what I have done so far.
You have set the soucre's frequency freq=fin -> what value to choose?
You have set pss-fund=100MHz but the pac-sweep from 10Hz to 50kHz, why those values?
- achim

Title: Re: Spectre's PAC: Frequency response of an amplif
Post by Ken Kundert on Sep 9^th, 2005, 6:56am

All amplifiers are slightly nonlinear. AC analysis will characterize the behavior when signals are small. If you want to characterize the behavior for large signals you should run a swept PSS analysis. The combined PSS/PAC suggested by Pancho Hideboo is another possibility, but it answers a slightly different question, which is "how does the circuit respond if I apply two signals, one large at a fixed frequency, and one small at a swept frequency?".

-Ken

Title: Re: Spectre's PAC: Frequency response of an amplif
Post by achim.graupner on Sep 16^th, 2005, 5:03am

Hi Ken,

I tried to use the swept PSS, but I suppose I did not have understood what is going on during a PSS: I tried to use a very simple circuit: an ideal resistor, an ideal capacitor and an ideal inductor with a series resistor that are connected in parallel.

V0 (net1 0) vsource dc=5 mag=1 type=sine ampl=1 freq=frq
L0 (net1 net8) inductor l=18m
C0 (net1 0) capacitor c=10p
R1 (net8 0) resistor r=1
R0 (net1 0) resistor r=100K
ac ac start=10k stop=10M annotate=status
sweeppss sweep param=frq start=10k stop=10M dec=20 {
pss pss fund=frq harms=0 errpreset=moderate annotate=status

pac pac sweeptype=relative start=frq maxsideband=0
+ annotate=status
}

now I've done both: an AC-analysis and a swept PSS and measured the current flowing out V0. In AC-simulation the current is, as expected at minimum at about 280kHz.
The PSS results in an output current of 5A which is comparable to the DC-operating point. As I am interested in the AC amplitude is run a subsequent PAC.
For PAC I've ploted the source's current, as sweep I've used "frequency" and as signal level I' used peak. In this case the results of the AC and the swept PSS/PAC analysis are identically. However, if I use "variable" as sweep value, the resonant frequency is to be found at a different location. Any explanations for this behaviour?

Thanks in advance,
- achim

Title: Re: Spectre's PAC: Frequency response of an amplif
Post by Ken Kundert on Sep 16^th, 2005, 8:28am

I cannot understand your question, and your netlist does not exhibit the behavior you describe.

-Ken