FOR HIGHER OUTPUT
BY FERNANDO GARCIA
This project started in a roundabout way. I had been interested in assembling a
Class D amplifier for quite some time, but had not found the time or motivation
to do it. When my subwoofer amp died and I found it too expensive to repair,
I was motivated to finally delve into Class D. Searching the Internet, I found
some pretty good and reasonably priced kits, which I promptly bought.
By now, you may be wondering why this article is not
named a Class D amp project instead. The reason is
that all amplifiers require a good power supply. Class D
amps in particular require a tightly regulated supply for
two reasons. The first is to maximize power output; it must
get as close to the amp’s maximum voltage rating without
exceeding it and thus destroying it. The second is that
these amps operate with very low feedback which makes
them quite susceptible to powerline ripple (otherwise
known as hum) — not a good thing on a sub amp.
Again searching the Internet, I found many opinions
on suitable power supplies, but they all agreed with my
original findings: a tightly regulated supply is a must. Some
people advocated the use of a linear regulator, but I felt
that its low efficiency would negate the Class D’s major
strength: its high efficiency. Thus, I decided to use a
switchmode regulator. NOTE: If you feel daunted by the
complexities of a switchmode supply, you may try the
simpler approach discussed in the sidebar, which employs
common adjustable three-terminal regulators.
So, what controller type should I use? I decided on
National Semiconductor’s line of Simple Switchers as they
are relatively simple to design, reliable, and widely available.
Next question: what type of switchmode topology? At
first, I thought about using a buck regulator. However,
simulating the circuit with National’s software, it became
readily apparent that a buck regulator simply was not a
feasible option. The output voltage level the project
required ( 32 VDC) — with enough margin for low and
high line conditions — would exceed the family’s ratings.
Then I considered a boost topology, as it requires a
48 May 2009
lower voltage input which will be stepped up. The output
voltage may be easily realized with an input voltage in
the 25 volt whereabouts — easily met with off-the-shelf
transformers. Unfortunately, the boost regulators place a
substantial current stress on the main transistor switch and
I could not get the desired power level (150 watts) with
any single IC from National’s lineup.
Back to the drawing board. In these circumstances,
the solution is frequently to use a boost controller IC
coupled to an external MOSFET switch. That would have
worked, but I wanted to attempt something different.
How about two simple switchers in a master/slave boost
configuration? That was intriguing, as I had always desired
to test my skills at current sharing schemes. Additionally,
since two devices are employed, how about synchronizing
them with a 180 degree phase shift, such that we have a
two phase controller with half the ripple and a step
response twice as fast of a single switcher? The idea
became very appealing as the lessons learned from this
experiment could be applied later in high power supplies.
For a high performance switchmode power supply,
a printed circuit board (PCB) is a must to accommodate
not only large ground planes but also a few strategic
components in SMT format. Also, a lot of thought was
given to design a single layout that would accommodate
either the master or slave configurations, depending on
how the board is stuffed.
The result is shown in Figure 1 which is drawn on