From: email@example.com (carltons)
Subject: Re: How to interpret AC Analysis?
User-Agent: NewsWatcher-X 2.2.3b2
NNTP-Posting-Date: Sun, 22 Sep 2002 22:06:25 GMT
Date: Sun, 22 Sep 2002 22:06:25 GMT
In article , "Helmut Sennewald"
> "carltons" schrieb im Newsbeitrag
> > In article , John Woodgate
> > wrote:
> > ....
> > I may be missing something in this thread, but AC analysis in Spice is
> > always done with a 1v pk drive
> Hello Steve,
> it is not necessary to use 1V. You can use the voltage level
> you want. e.g. 1mV. It is just convient to have 1V, because
> the dB scale then gives the amplification factor directly.
> > and an assumption of linearity.
> That's the basic idea behind the .AC analysis.
> > For
> > example, you put in 1v into a gain stage of 10,000 and you get out a 10KV
> > pk sine wave.
> > It also assumes an infinite supply voltage.
> The .AC analysis assumes no DC-supply. It uses a linearized model
> of all your components. Any DC-voltage supply is treated as a short(0V).
> Any current source is treated as isolated. These sources are used
> only for calculation of the DC-operating point which is necessary
> to calculate the linearized models.
> > If you want
> > reality, then you put in correct models and do the analysis in the time
> A filter designers will always use the .AC analysis first, but in
> parallel he/she should think about slewrate limitations and output
> voltage swing, especially of the active devices.
> > mode or am I missing something here?
> I agree with you that the .TRAN timing analysis is very important
> if it comes to the reality check. Quite often, it is not possible
> to overlook the stability of a circuit with the .AC analysis.
> It is also nearly impossible to decide wether you have overshoot
> or not in the time domain when you only have an .AC analysis.
> The .TRAN analysis will give you the answer.
> Best Regards
I agree with you completely, but something in this thread just wasn't
making much sense. AC analysis is fine by itself if you're sure that the
circuit is stable and that the voltages in and out are going to make it
linear. To believe otherwise is a waste of simulation time. BTW, does
anyone ever check the zeta of the filter that they design using time
domain analysis? I do, but I'm weird.
Not a baseball player. Just a old ex-Motorola engineer. I did play third
plate when I was younger and am a right hander, however.