mini mosfet audio amp design

[[email protected]]

Somebody is going to have to explain page 3 of this data sheet to me.
Please don't tell me it is a misprint, as National has had six years to
correct.

At a supply voltage of 15 volts, the curve shows 10 watts out.

Last I looked no single-supply design can drive to below zero or above
supply rail without the use of some inductive tricks that the test circuit
doesn't show.

So, converting a 15 volt peak-to-peak output signal (assuming that the
device can swing to both rails) gives about 5.3 volts RMS.

The data sheet specifies an 8 ohm load, so given that Power = volts^2 / load
yields about 3.5 watts out.

What gives?

Jim

Only thing I can think of - they're assuming (-15V) and (+15V)
available.

If you go from 0-15V, you're right;
15 / SQRT(2) / 2 = 5.3V RMS; V^2/R = 3.5 W.

If we go (-15) to (+15), we take
30 / sqrt(2) / 2 = 10.6V; (10.6V)^2 / 8 ohm = 14 W out possible. With
some going to heat.

Michael

 

[Joerg]

Here's an interesting schematic:

http://www.redcircuits.com/Page123.htm

I was wondering what R7 would be connected to. (It's not clear in the
picture.)

What do you guys think of the schematic, overall? A good project for
a beginner?

To be honest, no. First, a circuit that needs potmeters where setting
them wrong can cause stuff to blow up isn't my cuppa tea. Second, an
audio power amp that requires a regulated supply isn't my cuppa tea either.

If I'd spend the time building an audio amp I'd go all out and build the
biggest honking monster class-D amp the neighborhood has ever seen. The
kind when one twang on the E-guitar causes the plaster to fall off the
ceiling. That's what I did as a kid, except it wasn't class D and had
tubes (and the laws came after some playing on the E-guitar ...)

 

[Rich Grise]

It's better than many discrete designs for sure but not as good as some really
excellent ones.


That's more like it.


That looks fairly OK too.


It seems to have redeemed itslf now.

I notice that all three have a "crossover" pot, but it's kind of
a set-it-and-forget-it adjustment.

Either is WAY WAY better than that first one. Like John said,
if the biasing was designed right, then R3 could be a fixed R,
and R4 could be a much lower value pot in series with a fixed
R, to make it less ridiculously touchy.

Cheers!
Rich

 

[mng]

On Mar 24, 2:53 pm, [email protected] wrote:

I meant class D. Brain fart. [Dan Quayle: "To waste a mind is
terrible."]

I see these chips advertised from time to time, but never did a
research project on who has good silicon. Much like switcher design, I
suspect the BS runs deep on such amplifier chips.

I wonder how the Class D chips sound compared to the LM3875...

Ah well, my soldering ability won't yet allow me to play with Class D
chips. (Unless they make DIPs that will fit on a breadboard...?)

Michael

You could try the AMP6 kit from 41hz.com. I have their AMP3. Toroids
are a pain to wind, but it's a lot of fun overall. And it sounds good
(IMHO).

 

[Dave Platt]

Somebody is going to have to explain page 3 of this data sheet to me.
Please don't tell me it is a misprint, as National has had six years to
correct.

At a supply voltage of 15 volts, the curve shows 10 watts out.

Last I looked no single-supply design can drive to below zero or above
supply rail without the use of some inductive tricks that the test circuit
doesn't show.

Assuming that you're looking at the graph in the middle of the left
column... the labelling of the X axis indicates that this graph is
based on the use of a bipolar power supply. The 10-watt number comes
at the point where you're using a +15 and a -15 supply, for a
peak-to-peak voltage limit of 30 volts.

The chip can run on either a single- or double-sided power supply (the
former is shown on Page 2, the latter on Page 1). The overview text
on Page 1 refers to a double-sided supply, and that's the version they
used to derive the graph on Page 3.

What gives?

The datasheet suffers from bipolar disorder?

 

[Don Klipstein]

Somebody is going to have to explain page 3 of this data sheet to me.
Please don't tell me it is a misprint, as National has had six years to
correct.

At a supply voltage of 15 volts, the curve shows 10 watts out.

Last I looked no single-supply design can drive to below zero or above
supply rail without the use of some inductive tricks that the test circuit
doesn't show.

So, converting a 15 volt peak-to-peak output signal (assuming that the
device can swing to both rails) gives about 5.3 volts RMS.

The data sheet specifies an 8 ohm load, so given that Power = volts^2 / load
yields about 3.5 watts out.

What gives?

That IC uses a split supply, and those curves (I see them on Page 4)
show output reaching 10 watts for 4 ohm load and a little short of 10
watts into an 8 ohm load with +/- 15 volt supply - that is 30 volts
rail-to-rail.

- Don Klipstein ([email protected])

 

[Eeyore]

I notice that all three have a "crossover" pot, but it's kind of
a set-it-and-forget-it adjustment.

Either is WAY WAY better than that first one. Like John said,
if the biasing was designed right, then R3 could be a fixed R,
and R4 could be a much lower value pot in series with a fixed
R, to make it less ridiculously touchy.

Agreed.

Welcome back again btw.

Graham

 

[Glenn Gundlach]

This one looked really good:http://www.swtpc.com/mholley/ PopularElectronics/Jul1969/PE_Jul1969.htm

until I realized I'd have some trouble getting some of the parts.

Thanks y'all,

Michael

I built a pair of those my junior year in high school. The biasing was
the pits. When I rebuilt it a 2 years later I used bigger heatsinks -
1 per channel and changed the bias a bit by removing the DHD-800 and
using a low value pot.

The 40408s are no longer available but a 2N5655 from ON Semi would
probably be suitable. The rest should be available.
It would be better to run in on split 30 supplies but you'd do well to
check out this guy. I believe Phil Allison is a contributor.

http://sound.westhost.com/projects-1.htm

GG

 

[Paul E. Schoen]

Bob

It seems that the upper bank of transistors are on constantly with a
combined current of 3.6 amps, which is also drawn by the lower bank with no
input. So with 15 volt supplies you are blowing 108 watts without any
output. The good news is that the efficiency gets better as you approach
maximum output. Maybe as good as a "real" amplifier.

Also, look what happens with a 3.2 ohm load. Nasty clipping and spikes.

This does not seem to be a good design. And you really should use models of
actual parts, rather than generic op-amps and NPN and PNP transistors.

There are many simpler and much better circuits you can make.

Paul

 

[Rich Grise]

Agreed.

Welcome back again btw.

Oh, I've been around, just doing more reading than posting. I've
given up on the politics crap, it seems you always get these
long boring pissfests, with the neocons being neocons and the
socialists being socialists, ho hum.

Of course, being a Libertarian, both types of extremists hate me. ;-)

Cheers!
Rich

 

[Kevin Aylward]

It's not the way I'd draw a potnetiometer, and it wouldn't pass muster
with the drafting office of any of the European firms I've worked for.


You'd want to start off with no bias voltage between the MOSFET gates
(pure class B operation), and monitor the no load current drawn by the
amplifier as you used the pontentiometer R4 to increase this voltage
(to bring you into class AB).

It is the crudest possible biasing scheme. The MOSFET gate-to-source
voltage is temperature dependent

http://www.powermanagementdesignlin...ionid=KD1E2OVGD5J4QQSNDLPSKH0CJUNN2JVN?pgno=4

and tends to decrease with increasing temperature at "low" drain
currents, leading to a risk of thermal runaway.

However....if one is using the lateral Hitachi (2SK50 2SJ135) mosfets, they
have a pretty close to zero Tc at around 100ma bias, and so not much of a
problem.

I think the most serious problem with this circuit is the lack of a gate
source zener clamps. An output S/C will cause up to the full 48 volts to be
applied to said gate source, frying said gate source.