This topic entertains discussion of the challenges presented by the audio channel of a high-power audio generation device.
I
started by choosing the smallest, most efficient and highest power high
frequency capable "transducer" (tweeter) that I could find, then worked
backwards from its maximum power handling capability to design a
solution that would be able to push it right up to and past its maximum
power handling limit.
This is the BOSS BPT70 300 Watt Super Compression Tweeter. A pair of them can be purchased from Amazon for just over $50.
http://www.amazon.com/Boss-BPT70-Super-Compression-Tweeter/dp/B0032FOK06The
BOSS website has distressingly little to say about this tweeter, but we
do know that its frequency response is flat out past 25k Hz and extends
down to 3500 Hz. So it ought to work for our purposes. It is also the
right size (though this sucker is surprisingly solid and weighty) for
suspending out in front of a parabolic mirror.
As I have
mentioned elsewhere (under the root of the battery management discussion
topic), driving this sensitive and high-power 4 ohm impedance tweeter
to its limits will require a peak-to-peak voltage swing of 110 volts at a
current of 12 to 15 amps.
To minimize battery consumption, we
don't want to be wasting power heating up audio amplifier heat sinks,
which dictates a high-efficiency class-D (digital) switching amplifier
design. I settled upon a n H-bridge output structure so that we only
need to provide half the peak-to-peak voltage with each side of the
tweeter being switched between ground and 44.4 to 55.5 volts.
(Note
that I haven't quite settled upon the operating voltage for the audio
[amplification] system. We get 55.5v from having 15, 3.7v
lithium-polymer cells connected in series. This is what I would
prefer. But I found a nice 12-cell battery management chip that could
provide charge management and battery cell balancing for up to 12 cells.
Twelve cells at 3.7v per cell gives us 44.4v total. So, obtaining
55.5v would require two of the 12-cell-maximum charge management chips
and electronics, effectively doubling the complexity of battery
management just to handle and provide the voltage from an additional
three cells (11.1v). The question is... would that difference be
audible?? Would/does it really matter? We'll see.)
I have three
favorite high-voltage power-handling class-D switching amplifier chips,
all from Texas Instruments. Their TAS5261, TAS5630 and TAS5631:
http://www.ti.com/ww/en/analog/tas5630/index.shtmlhttp://focus.ti.com/docs/prod/folders/print/tas5630.htmlhttp://focus.ti.com/docs/prod/folders/print/tas5631.htmlhttp://focus.ti.com/docs/prod/folders/print/tas5261.htmlEach
of these lovely class-D switching amplifiers is less than $10 in low
quantities, making them a good choice for our application. The 5630/1's
are stereo amps, but them can be strapped together to double their power
handling capability. One takes an analog input whereas the other
accepts a pulse-width-modulated (PWM) source. The TAS5261 is a
monophonic 315-watt amplifier which accepts a PWM modulated source.
Since
the Cortex-M3 chip has multiple PWM outputs, as well as a single 10-bit
DAC output, I plan to experiment with both approaches to see which one
makes the most sense. A high-frequency PWM output can be readily
converted into an analog voltage using a simple 2-pole LC filter since
the PWM repetition rate can be many times the highest audio frequency we
plan to generate.
All three of these devices have very expensive
evaluation boards available which, while totally impractical for
production use, will make quick prototype decision making much easier.
And they can be used as reference designs for the use of those chips to
save on prototype iterations.