It's Only a Model
The calculator only provides a rough model of typical linear
regulated power supply behavior. It does not attempt to model actual
device behavior, or take all the known sources
of variability into account. It's fair to say that its results
are just plain inaccurate.
This is not the same thing as saying that its results are useless.
It's my experience that when faced with the calculations required
to pick the proper transformer, many will give up and simply choose
something that's way over-specified to be sure it won't put out too
little voltage. This will result in a design that is at best
pointlessly wasteful, and possibly even unworkable. You only need
a good ballpark estimate to avoid this, not an accurate physical
model.
Likely Sources of Inaccuracy
The rectifier type selector simply uses 0.5 V per diode for a
Schottky bridge rectifier, and 1 V for a silicon rectifier. These
values are representative of power diodes of this type, but the
actual values will be somewhat different. The exact diode type and
the load current play into the actual voltage drop, and the calculator
doesn't even attempt to model this.
The calculation for ripple in AC-input supplies is very much an
approximation. First, the formula used assumes that you're using a
simple capacitor type filter, and that the cap is large enough that
the load current won't cause it to let through too much ripple.
Second, we make no attempt at modeling the nonlinear IR drops in
the transformer's output coil, or their interactions with the filter
stage.
The thermal resistance calculation assumes that you've tightened
the mounting bolt well, and used the correct amount of heat sink
compound. If you wanted to model these errors, you can do it by
increasing any of the thermal resistance numbers. (They're just
added together in the calculation, so it doesn't matter which one
you change.)
The calculated temperature isn't an absolute value. To get that,
you need to add that temperature to the temperature of the air
surrounding the heat sink, called the "ambient temperature." You
will almost certainly not know this accurately. Chances are that
you're going to box the regulator up, which allows the regulator
to heat the air up inside the enclosure. The only way to figure
this out accurately short of just building it and trying it requires
high-end fluid dynamics software. This simple little calculator can
only make an arbitrary guess about ambient temperature when deciding
whether your configuration makes sense or not. It may not even come
close to your actual situation.
References
The ripple voltage calculation comes from The Art of Electronics, second edition,
page 46.
Section 7.2.8 in The Circuit Designer's Companion (pp.
236-237) by Tim Williams was helpful in putting together the rest
of the calculations. Those calculations work the problem from the
other direction and have a different purpose, however, so I was
unable to use any of them exactly. Thus, all errors in application
are mine.
This space intentionally left
blank. :)