Please only people who are interested in this topic reply.

Now I have the power supply to hold the rails at 65-68 volts, but I
can't pump that much into my speakers. They are 2.33 ohms. I have
plenty of woofers which I like, but I might add some tops in the
future. In the mid and treble regions they are more like 6 ohms right
now.

Now alot of talk and engineering has been given to protecting woofers.
In fact the relay that cuts out on a normal amp does so when there is
a DC offset, the DC would never make it to the tweeters. But I have
blown some anyway.

Now I envision the circuit, three output pairs per channel, and
paralell FETs as switching regulators. I have no problem designing the
PWM circuit or the amplifier stages, what the problem is, is how
exactly to control the power.

If I detect the output power and set the regulators to be able to
provide a short burst of power for a number of milliseconds, perhaps
12 Db or so above the base output level, the question becomes how many
milliseconds ? And what "rechanrge" time ? And then how to integrate
time to avoid overheating the speakers.

I know this is not easy, even if it reduced to the simplest terms, but
I know this mich, when it is cranked, if you edge the power down by
20% it is likely noone will notice, as long as it doesn't distort. The
anti distortion circuit is going to be in place. The best way then, as
long as sufficient headroom exists is to regulate the rails down.
Otherwise things will run hotter.

I'm also looking for ideas on cooling. I am starting to think that the
heatsink and fan for a modern microprocessor might do well. If this is
feasible, the bulk is cut down. Instead of heat sinks there could be
an array of these HS/fan combos. One for each transistor. OK that
might mean 24 of them, but I want what I want.

I also intend to do this without anything digital except for maybe a
few gate ICs. You can't buy sometrhing like this.

I do not put digital in the audio path, a view shared by some
manufacturers who still use motoized volume controls. Or used to. The
EQ will be analog, as will the gain stage for the gain limiting. When
the EQ is set flat and the unit is not overdriven it will function as
a straight line. A wire with gain.

So any ideas other than going and ****ing myself are welcome.

JURB

In article . com,
says...
(snip)
If I detect the output power and set the regulators to be able to
provide a short burst of power for a number of milliseconds, perhaps
12 Db or so above the base output level, the question becomes how many
milliseconds ? And what "rechanrge" time ? And then how to integrate
time to avoid overheating the speakers.

I know this is not easy, even if it reduced to the simplest terms, but
I know this mich, when it is cranked, if you edge the power down by
20% it is likely noone will notice, as long as it doesn't distort. The
anti distortion circuit is going to be in place. The best way then, as
long as sufficient headroom exists is to regulate the rails down.
Otherwise things will run hotter.
(snip)

So, a long time-constant (sub-audio) continuous sensing circuit, or
a digital sample at a constant (say,.25sec) interval, with presets for
boosted, normal, and reduced levels. On the face of it, the digital
sample would be less likely to cause distortion in the output.
Interesting concept, Jurb
This makes me think of a high power compressor!

wrote:

Please only people who are interested in this topic reply.

Now I have the power supply to hold the rails at 65-68 volts, but I
can't pump that much into my speakers. They are 2.33 ohms. I have
plenty of woofers which I like, but I might add some tops in the
future. In the mid and treble regions they are more like 6 ohms right
now.

Now alot of talk and engineering has been given to protecting woofers.
In fact the relay that cuts out on a normal amp does so when there is
a DC offset, the DC would never make it to the tweeters. But I have
blown some anyway.

Now I envision the circuit, three output pairs per channel, and
paralell FETs as switching regulators. I have no problem designing the
PWM circuit or the amplifier stages, what the problem is, is how
exactly to control the power.

If I detect the output power and set the regulators to be able to
provide a short burst of power for a number of milliseconds, perhaps
12 Db or so above the base output level, the question becomes how many
milliseconds ? And what "rechanrge" time ? And then how to integrate
time to avoid overheating the speakers.

I know this is not easy, even if it reduced to the simplest terms, but
I know this mich, when it is cranked, if you edge the power down by
20% it is likely noone will notice, as long as it doesn't distort. The
anti distortion circuit is going to be in place. The best way then, as
long as sufficient headroom exists is to regulate the rails down.
Otherwise things will run hotter.

I'm also looking for ideas on cooling. I am starting to think that the
heatsink and fan for a modern microprocessor might do well. If this is
feasible, the bulk is cut down. Instead of heat sinks there could be
an array of these HS/fan combos. One for each transistor. OK that
might mean 24 of them, but I want what I want.

I also intend to do this without anything digital except for maybe a
few gate ICs. You can't buy sometrhing like this.

I do not put digital in the audio path, a view shared by some
manufacturers who still use motoized volume controls. Or used to. The
EQ will be analog, as will the gain stage for the gain limiting. When
the EQ is set flat and the unit is not overdriven it will function as
a straight line. A wire with gain.

So any ideas other than going and ****ing myself are welcome.

JURB

with out going into great details. It seems that you are interested
in a class D amplifier ?
That type of amp is a PWM systems where you must recreate the sine
wave via the PWM system. The problem with all of this is, you need a
high carrier to be able to cover the complete audio spectrum.
The last one I looked at which was only a low power unit, used a
500 Khz carrier. It would be nice to have something in the line of
2 mhz and up how ever, you'll run into issues with generating R.F.
illegally.


--
"I'm never wrong, once i thought i was, but was mistaken"
Real Programmers Do things like this.
http://webpages.charter.net/jamie_5

In article ,
says...
wrote:

(snip)
with out going into great details. It seems that you are interested
in a class D amplifier ?
That type of amp is a PWM systems where you must recreate the sine
wave via the PWM system. The problem with all of this is, you need a
high carrier to be able to cover the complete audio spectrum.
The last one I looked at which was only a low power unit, used a
500 Khz carrier. It would be nice to have something in the line of
2 mhz and up how ever, you'll run into issues with generating R.F.
illegally.




OP wants to implement the amplifier itself in
analogue. I think he will accept some digital controls.