Reply-To: "Kevin Aylward"
From: "Kevin Aylward"
Subject: Re: How to interpret AC Analysis?
X-Newsreader: Microsoft Outlook Express 6.00.2800.1106
Date: Sun, 22 Sep 2002 17:25:58 +0100
NNTP-Posting-Date: Sun, 22 Sep 2002 17:28:57 BST
"Terry Pinnell" wrote in message
> "Kevin Aylward" wrote:
> >"Terry Pinnell" wrote in message
> >> CircuitMaker, like most other Spice programs I've seen, includes an
> >> Analysis (or 'Bode Plot') in its repertoire. But, until Win Hill
> >> John Woodgate set me straight in a recent thread, I'd assumed it
> >> something of a no-brainer to use. I'd thought that, for any
> >> could simply select an AC Analysis on the output, set the x and y
> >> scales respectively to 'Log' and 'Magnitude in Decibels', and I
> >> be able to read off the circuit's '-3dB point' (or points) with
> >> cursors.
> >> However, as John puts it, I now see that "You must refer the output
> >> levels to the output voltage, not the input voltage."
> >I have not seen this post, and don't know to what he is refereeing.
> >> So, how exactly do you go about doing that in practice please?
> >> pretty extensive reading, I've still seen no documentation that
> >> this.
> >Set your AC input voltage to 1V. The plot at any node is the gain
> >referred to that input voltage.
> >> Is it perhaps a matter of first using Transient Analysis to
> >> the circuit's output at the 'base frequency' (say 1 Hz in a low
> >> filter), and then applying this in a calculation after the AC
> >> Analysis? Or is there some reasonably straightforward procedure
> >> professionals follow, albeit not quite as simple as I thought?
> >I think they're leading you up the garden path. It is as you stated
> >the bigging. Its a no-brainer, although a decent spice automatically
> >the correct defaults scales modes for dc, ac, trans and noise
> I may not have explained it too well, but I don't think there's any
> doubt about it!
> Maybe after reading the relevant posts from about
> onwards, you'll see. And I posted an illustration of the 'no-brainer'
> approach I was using at
> It clearly gave the wrong answer - as I soon found when I breadboarded
> it. With hindsight it then became embarrassingly obvious that the
> cut-off point should be independent of the gain setting.
> But I'm sort of pleased to see that maybe it's not so obvious after
Not really, if the issue is what it seems to be. Its trivial.
Note that above I said that the *gain* is plotted when the input signal
is 1V, *not* the output level. Its obvious that if the gain is say
24.7db, the "3db" point is 3db *down* from 24.7db, i.e. 21.7db!
You seem to be confusing -3db as an absolute value, rather then as 3db
down from a reference.
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.