From: morris.slutsky on
Hi everyone,

I don't know if this is a suitable newsgroup for this sort of thing.
I've been posting in alt.guitar.amps mostly, but I am interested in
hearing from more of an electronics point of view. I guess I'm just
wondering - what do you guys think of this circuit? Workable,
buildable?

Anyway I was simulating a circuit in SPICE tonight and I just might
end up building it. It's based on an idea I saw on John Broskie's
TubeCad site, a push-pull output stage with floating supply. Desired
application - guitar amp power stage. It's by no means a "perfect"
audio circuit but I think it might be well-suited for this particular
use. And it's got a very, very low parts count.

Absolutely nothing exotic, the most bare bones simple transistors
ever. Radio Shack used to sell all of these. Don't think they do
anymore, though.

Schematic:
http://img213.imageshack.us/img213/2480/10wpowerstageym6.png

Resistors R1-R4 work very hard here. They set gain, linearity, input
impedance, output impedance, and bias. This thing idles along at
about 150 mA through Q3 and Q4, according to SPICE. A reasonably warm
class AB.

SPICE thinks that this is actually fairly linear. At 10W output into
8 ohms, 2nd harmonic is 32 dB down, 3rd harmonic predominates at 20
dB down, and there's a whole series of evens and odds after that at
much lower levels. The even harmonics are, I guess, because the
transistors aren't truly perfectly complementary.

The feedback is all local, so it clips smoothly. And even harmonics
aren't bad too here. Guitar amps are supposed to clip smoothly, and
even harmonics are considered to add some richness to the sound.

Damping factor is pretty horrible - the output impedance of this seems
to be about 6 ohms, according to SPICE, basically undamped. But a lot
of guitar amps, classic pentode ones, are undamped as well.

The floating power supply and using the filter caps as output coupling
caps looks goofy, but SPICE thinks it will work. And it would save me
the trouble of worrying about DC offset. Guitar amps really shouldn't
have response down to DC anyway - bad sounds can happen that way. I
do have R9 and R10 to sort of 'center' the supply in the case of real
world, non-identical caps, but the DC load of speaker to ground would
do the real balancing work here I hope.

I am really, really tempted to build this.

I'm not trying for any elusive 'tube sound' goal, I think that stuff
is way overrated. I just hope it would work, and maybe even sound
good when driven hard.

Do you guys think it would work? Does anyone use a circuit like this
for anything?

From: morris.slutsky on
On Dec 15, 3:39 am, D from BC <myrealaddr...(a)comic.com> wrote:
> I wouldn't do a discrete amplifier at 10W..
> Use an amplifier module.
> There's many from Digikey and Mouser.
> Use the application notes on the datasheets.
>
> D from BC
> myrealaddress(at)comic(dot)com
> British Columbia
> Canada

Hi D from BC.

I appreciate your advice, and I am aware that monolithic audio
amplifier IC's or modules are available in powers up to 100W or so.
However, this is a special purpose application.

I fully understand that these modules can provide absolutely wonderful
THD, frequency response, efficiency (class D etc.), damping factor, et
cetera. But they sound absolutely horrible if simply used in this
application. People have tried it. It sounds horrible. Guitar
amplifiers are generally operated partially or entirely into
distortion. Even a 'clean' guitar sound is pushing various stages
into the red on the transients. This is part of the sound of an
electric guitar. Vacuum tubes are not necessary for this, but are
well loved in this application. I've built vacuum tube amplifiers
myself, and enjoy using them for this.

Yes, these distortions can be simulated in many different ways, analog
or digital, and then fed to a perfectly clean power amplifier. Many
people, in fact, do this. For example, I've seen bands who have no
amplifiers on stage whatsoever, but prefer to run their guitars and
basses through such modeling devices directly into the house PA. One
problem with doing this is that your modeling setup will actually
require a much more powerful amplifier in order to prevent it from
ever clipping. Which, in turn, means bulkier speakers. You end up
with a much heavier and more complicated setup which doesn't really
provide much of an improvement in performance. That's why they still
make guitar amplifiers with imperfect amplification stages.

Audiophiles don't like the 'canonical' transistor amplifier either,
and I can see their point even if I don't agree with the 'magic wooden
knob' school of thought. The standard design - the one that looks
like a great big op-amp - doesn't handle being pushed into clipping
very gracefully. If you listen to recordings with lots of dynamics,
like most classical, you will probably be clipping on transients.
Just like a guitar player would. The question is - how does that
clipping sound? Gentle, or harsh? Does the amp take a few
milliseconds to recover equilibrium afterwards?

Some people would prefer their harmonic distortion to come in more
gradually, and are willing to tolerate/enjoy a little bit even at low
levels. Audiophiles and guitarists are such people.

And there's nothing wrong with that, is there? We're allowed to build
more than one kind of audio amplifier, at least I'd hope so. I mean,
wow, there are people building single ended triode amps in this power
range too.

I was just hoping people would check out the circuit and let me know
what they thought of it.
From: Dan Kuechle on

<morris.slutsky(a)gmail.com> wrote in message
news:8411097c-03f5-4299-bde7-5c095a778aba(a)w34g2000yqm.googlegroups.com...
> Hi everyone,
>
> I don't know if this is a suitable newsgroup for this sort of thing.
> I've been posting in alt.guitar.amps mostly, but I am interested in
> hearing from more of an electronics point of view. I guess I'm just
> wondering - what do you guys think of this circuit? Workable,
> buildable?
>
> Anyway I was simulating a circuit in SPICE tonight and I just might
> end up building it. It's based on an idea I saw on John Broskie's
> TubeCad site, a push-pull output stage with floating supply. Desired
> application - guitar amp power stage. It's by no means a "perfect"
> audio circuit but I think it might be well-suited for this particular
> use. And it's got a very, very low parts count.
>
> Absolutely nothing exotic, the most bare bones simple transistors
> ever. Radio Shack used to sell all of these. Don't think they do
> anymore, though.
>
> Schematic:
> http://img213.imageshack.us/img213/2480/10wpowerstageym6.png
>
> Resistors R1-R4 work very hard here. They set gain, linearity, input
> impedance, output impedance, and bias. This thing idles along at
> about 150 mA through Q3 and Q4, according to SPICE. A reasonably warm
> class AB.
>
> SPICE thinks that this is actually fairly linear. At 10W output into
> 8 ohms, 2nd harmonic is 32 dB down, 3rd harmonic predominates at 20
> dB down, and there's a whole series of evens and odds after that at
> much lower levels. The even harmonics are, I guess, because the
> transistors aren't truly perfectly complementary.
>
> The feedback is all local, so it clips smoothly. And even harmonics
> aren't bad too here. Guitar amps are supposed to clip smoothly, and
> even harmonics are considered to add some richness to the sound.
>
> Damping factor is pretty horrible - the output impedance of this seems
> to be about 6 ohms, according to SPICE, basically undamped. But a lot
> of guitar amps, classic pentode ones, are undamped as well.
>
> The floating power supply and using the filter caps as output coupling
> caps looks goofy, but SPICE thinks it will work. And it would save me
> the trouble of worrying about DC offset. Guitar amps really shouldn't
> have response down to DC anyway - bad sounds can happen that way. I
> do have R9 and R10 to sort of 'center' the supply in the case of real
> world, non-identical caps, but the DC load of speaker to ground would
> do the real balancing work here I hope.
>
> I am really, really tempted to build this.
>
> I'm not trying for any elusive 'tube sound' goal, I think that stuff
> is way overrated. I just hope it would work, and maybe even sound
> good when driven hard.
>
> Do you guys think it would work? Does anyone use a circuit like this
> for anything?
>

I am just a hobbyist. I am not am amplifier designer, but I don't see how
this amp would work. If you move the ground symbol to the other side of the
speaker you get something that looks more like an amp to me. I thought you
had put the gnd in the wrond spot until I read the rest of your post.
As drawn, it seems to me that when the input swings positive and Q1/Q3 turn
on they will be trying to short the positive voltage rail to ground thru .47
ohm (R7) and when the input swings negative and Q2/Q4 turn on they will be
trying to short the negative voltage rail to ground thru .47 ohm (R8).
Furthermore, if this amp did work, why wouldn't there be a huge 120 hz noise
component going through the speaker from the 60 hz full wave rectified power
supply? Are you sure you have the ground in the right place?

Dan


From: morris.slutsky on

> Are you sure you have the ground in the right place?
>
> Dan

Yep. I'm pretty sure that's where I want it.

It's based on a MOSFET design of John Broskie's from his Tubecad.com
site. I didn't believe it when I saw it either. But SPICE seems to
think it should work, and I like the simple feedback network. As Q1/
Q3 turn on, (and Q2/Q4 turn off) they pull down the positive voltage
rail. This pulls UP the negative voltage rail. This change in rail
voltage is coupled to the speaker through the filter caps C1/C2. As
far as ripple - equal and opposite ripple components will be coupled
to the speaker through C1 and C2, which should cancel. The feedback
of the darlingtons tends to 'buck' the ripple as well. This whole
thing simulates a lot better than it reasonably should.

And, as a design with low damping and without overall feedback, it
might well be a nice thing to have overdriven by a roaring guitar.

At least I think it might sound nice. I just have to try this
someday. Maybe soon.

From: Bob Eld on

<morris.slutsky(a)gmail.com> wrote in message
news:8411097c-03f5-4299-bde7-5c095a778aba(a)w34g2000yqm.googlegroups.com...
> Hi everyone,
>
> I don't know if this is a suitable newsgroup for this sort of thing.
> I've been posting in alt.guitar.amps mostly, but I am interested in
> hearing from more of an electronics point of view. I guess I'm just
> wondering - what do you guys think of this circuit? Workable,
> buildable?
>
> Anyway I was simulating a circuit in SPICE tonight and I just might
> end up building it. It's based on an idea I saw on John Broskie's
> TubeCad site, a push-pull output stage with floating supply. Desired
> application - guitar amp power stage. It's by no means a "perfect"
> audio circuit but I think it might be well-suited for this particular
> use. And it's got a very, very low parts count.
>
> Absolutely nothing exotic, the most bare bones simple transistors
> ever. Radio Shack used to sell all of these. Don't think they do
> anymore, though.
>
> Schematic:
> http://img213.imageshack.us/img213/2480/10wpowerstageym6.png
>
> Resistors R1-R4 work very hard here. They set gain, linearity, input
> impedance, output impedance, and bias. This thing idles along at
> about 150 mA through Q3 and Q4, according to SPICE. A reasonably warm
> class AB.
>
> SPICE thinks that this is actually fairly linear. At 10W output into
> 8 ohms, 2nd harmonic is 32 dB down, 3rd harmonic predominates at 20
> dB down, and there's a whole series of evens and odds after that at
> much lower levels. The even harmonics are, I guess, because the
> transistors aren't truly perfectly complementary.
>
> The feedback is all local, so it clips smoothly. And even harmonics
> aren't bad too here. Guitar amps are supposed to clip smoothly, and
> even harmonics are considered to add some richness to the sound.
>
> Damping factor is pretty horrible - the output impedance of this seems
> to be about 6 ohms, according to SPICE, basically undamped. But a lot
> of guitar amps, classic pentode ones, are undamped as well.
>
> The floating power supply and using the filter caps as output coupling
> caps looks goofy, but SPICE thinks it will work. And it would save me
> the trouble of worrying about DC offset. Guitar amps really shouldn't
> have response down to DC anyway - bad sounds can happen that way. I
> do have R9 and R10 to sort of 'center' the supply in the case of real
> world, non-identical caps, but the DC load of speaker to ground would
> do the real balancing work here I hope.
>
> I am really, really tempted to build this.
>
> I'm not trying for any elusive 'tube sound' goal, I think that stuff
> is way overrated. I just hope it would work, and maybe even sound
> good when driven hard.
>
> Do you guys think it would work? Does anyone use a circuit like this
> for anything?

Yes it will work, But, This amplifier has NO voltage gain. In fact it is the
output current gain stage of a more complete circuit. Look at other designs
and see that this is just the last stage, typical of many circuits. There is
little point in making this without adding the front end stages as well to
make a more thorough amp with both gain and feedback. This circuit has no
feedback, it's nothing more than unity gain Darlington followers.

To get ten watts into eight ohms, the input voltage will have to be 13 volts
peak or about 9 volts RMS. That seems a tad high for a line input from a
guitar. You won't be satisfied with it's underperformance, don't bother
making it.

If you want tube sound, make a tube amp.