m3h@psuecl.BITNET (05/18/86)
Some food for thought: Is it feasible to construct a 1.5 kW audio amp
for modest bucks ???
I think that the following scheme will do the trick for dynamic
loudspeakers.
+110 -110
.......... ...Rf... : :
: : : : :d s:
: : \ : : : \ : : \ ;[buf]|[...]|''''''''';
:..:- \..:.Rin:.:- \.:.....:+ \..: s : d ;
in.:+ / ...:+ / ...:- / ''''''''':''''''[not]''
: / : : / : : / {L}
V :.................:
Rs
A1 A2 C1 : OUTPUT STAGE
V
This schematic is a little tough to read so I'll try to talk through
the circuit. The input stage A1 is merely a unity gain amp. In actual
practice, it would be a good idea to opto-isolate this stage from A2.
A2 is a variable gain inverter with gain = Rf/Rin. A2's signal is fed to
C1, a comparator. The output of C1 drives two opto-isolated logic gates with
complementary states whose output voltages float with respect to the source
terminal of the power mosfets which they drive. These power mosfets switch
the power supply through the speaker (symbolized as {L}), charging it when
C1 is high and discharging it when C1 is low. The amount of current passing
through {L} is monitered through Rs, a sensing resistor. This voltage is fed
back to C1. Thus, the mean current passing through {L} must be proportional
to the input voltage. Current sensing probably provides better control over
driver excursion than voltage feedback (recall that F=ilXB).
Note that this scheme assumes that our switching circuitry operates
at a much higher frequency than the charge and discharge times of {L} to the
upper and lower hysteresis levels of C1. Furthermore, the load is assumed to
be inductor-like, i.e. this amp will not work with piezo drivers etc. The
power mosfets should be an array of mosfets in parallel since the ON
resistances of high voltage mosfets are typically ~1.5 ohms. The high voltage
supply is derived by rectifying 110 a.c. voltage and filtering (hence the
sleaziness) this eliminates the need for bulky transformers. Make sure that
your life insurance is paid up....
The design above can be easily modified for voltage feedback instead
of current feedback. Note that I haven't shown the hysteresis loop on the
comparator. Furthermore, I intend to add some protection circuits.
If anyone has ideas on this amp please reply before I build it.
P.S. Is this how Carver does it ???