Re: dc/ac nomenclature

["John S. Denker" <jsd@MONMOUTH.COM>]

At 02:28 PM 5/31/01 -0400, Carl E. Mungan wrote:

> (1) Do you use "dc" and "constant current" as synonyms or not? eg. Is
> I0*exp(-t/T) direct, since it never reverses direction? If so, the
> sum of two dc values is not in general dc. That seems odd to me, but
> if's that the accepted convention I can live with it.
>
> (2) In light of your answer to (1), what does the term "quasi-dc" mean?
>
> (3) Do you use "ac" to refer to things like voltage and the like
> which are not currents?

Wow, what interesting questions.  I don't recall ever having though about
this issue carefully before.

Here's my take on it:

1) In circuit analysis, I use DC and AC as shorthand for "DC component" and
"AC component".  In general, every signal has some of each.

1b) It is common for the AC component to be small compared to the DC
component, in which case the standard procedure is:
  *) Start with a DC analysis to find the operating point.  This will
generally be a nonlinear analysis because of nonlinear circuit elements
such as diodes.
  *) Then do the AC analysis, using linearized (small-signal) models for
the nonlinear elements.

2) As a consequence of the foregoing, I would say that an AC signal is one
where the DC component is small or irrelevant.  The electric power in the
wall socket beside me is an example.

Similarly I would say that a DC signal is one where the AC component is
small or irrelevant.  As an extreme example, a diode bridge used as a
battery charger might produce a signal where the AC component was
comparable to the DC component, but the AC component is irrelevant to the
battery-charging application.  Therefore I would (just barely) be willing
to call this a DC power supply.  I might add a qualifier such as "ratty DC".

If both the AC and DC components are significant, I would say that it is
*neither* an AC nor DC signal.  I would speak of the "whole" signal or the
"AC+DC" signal.

3) According to this viewpoint, the exponential waveform Carl asked about
could be classified either AC or DC or neither, depending on whether its
time-dependence was important to the application or not.

In particular, if I set my oscilloscope to "AC coupling", that waveform
might get through fully, partially, or not at all, depending on the timescale.

Aperiodic transients like this are often not well suited to analysis in
terms of AC and DC components.  You are not _required_ to resolve a signal
into its AC and DC components if you don't want to.  If it doesn't make
sense, don't do it.

4) I freely use AC/DC terminology to refer to things other than
currents.  I routinely speak of AC or DC voltages, and even the AC and DC
components of the position of a fluttering aileron.

5) I don't care whether the current actually reverses direction, except as
follows:  If there is a zero crossing, it indicates that the AC component
not small compared to the DC component, so the technique described in item
(1b) -- small-signal AC analysis relative to a DC operating point -- is not
going to suffice, and a full nonlinear AC+DC analysis will be
required.  But that might be required even if there are no zero crossings,
so still I don't really care about zero crossings per se.

6) I don't usually speak of quasi-DC.  To me, DC does not mean zero
frequency, as in exactly 0 Hertz.  Anything that is restricted to exactly 0
Hertz is uninteresting.

A slowly-varying DC signal is still a DC signal, as long as the variations
are small enough and/or slow enough that they don't cause trouble.

As an example, the DC power supplies in my audio amplifier have small (!)
but measurable amount of hum and other AC components.  I call them DC power
supplies, because that's what they are "supposed" to be doing.  But that
doesn't stop me from measuring the AC components.