Discussing audio amplifier design -- BJT, discrete

[Jon Kirwan]

I'd like to take a crack at thinking through a design of an
audio amplifier made up of discrete BJTs and other discrete
parts as an educational process.

I imagine this will be broken up into three sections; input
transconductance, transimpedance VAS, and output driver. But
other arrangements (such as combining the VAS and output
driver using a signal splitting BJT) would work for me, in
learning.

I said "BJT" and "discrete" but I'm also open to the idea of
using BJT pairs, such as the BCV61 and BCV62. In the case of
current mirrors, that may make sense. But not high-priced,
elite and/or hard to get, or obsolete. And no FETs. This is
to be about learning to design with BJTs.

SMT vs through-hole isn't an issue for learning about a
design, I suppose. If I need to build up some section and
test it with a signal, I'll probably want to do it quickly
and without having to buy services every step of the way. So
I may 'dead bug' SMT parts to get there. (The basic idea
here is to learn, not to make something tiny.)

Although I have some other applications, right now I'd like
the target use to be as a computer speaker system (not unlike
those dirt cheap, sadly almost all of them 10% THD, systems
sold today into this market. Except that I'd like to work
through the design on my own, from start to end.

Given what I understand right now from a very short search on
the topic, the input should be taken as a maximum of 1.0Vrms
and the input's load should appear to be something like 10k
ohms. If someone knows different from that, I'll accept the
criticism and change that spec.

I'd like to consider a tone control and a volume control to
be included.

Output is to be into a small 8 ohm speaker. With that
maximum 1.0Vrms at the input and the volume control set to
maximum the wattage into 8 ohms should be around 10 watts.
Since human hearing won't tell much difference between 8
watts and 12 watts, this is a bit of a sloppy spec and I'm
open to anything in the area of 5-20 watts... though I'm
really wanting to keep the rail voltages down to something
modest and the BJTs not having to tolerate hugish Vce.

Now that I say this, an odd idea comes to mind because the
CFL light bulbs include two TO-220 BJTs that can handle quite
a high Vce on them. I could cannibalize those. But to be
honest, I'm still not needing high watt outputs. So there's
no reason to think about scavenging such parts.

I would like to design it to work into 4 ohms as a margin
bound and not as a design goal, but even 5.6 ohms would be
acceptable.

I'm not looking for this to be done quickly, either. If it
takes months of only occasional back-and-forth, I'm fine with
that. Also, I expect to do my work and don't expect someone
else to hand-hold me from complete ignorante to complete
enlightenment. I just need someone to slap my face when
I say something terribly stupid and/or point in a truely
useful direction when I need it. Or else someone who is
wanting to explore this with me and willing to work for it.

Is anyone here willing to consider a sincere discussion?

Jon

 

[Jon Kirwan]

On Tue, 26 Jan 2010 12:57:13 -0800, Jon Kirwan
<jonk@infinitefactors.org> wrote:

... ignorante ...

.... ignorance ...

Jon

 

[David Eather]

Jon Kirwan wrote:

I'd like to take a crack at thinking through a design of an
audio amplifier made up of discrete BJTs and other discrete
parts as an educational process.

I imagine this will be broken up into three sections; input
transconductance, transimpedance VAS, and output driver. But
other arrangements (such as combining the VAS and output
driver using a signal splitting BJT) would work for me, in
learning.

I said "BJT" and "discrete" but I'm also open to the idea of
using BJT pairs, such as the BCV61 and BCV62. In the case of
current mirrors, that may make sense. But not high-priced,
elite and/or hard to get, or obsolete. And no FETs. This is
to be about learning to design with BJTs.

SMT vs through-hole isn't an issue for learning about a
design, I suppose. If I need to build up some section and
test it with a signal, I'll probably want to do it quickly
and without having to buy services every step of the way. So
I may 'dead bug' SMT parts to get there. (The basic idea
here is to learn, not to make something tiny.)

Although I have some other applications, right now I'd like
the target use to be as a computer speaker system (not unlike
those dirt cheap, sadly almost all of them 10% THD, systems
sold today into this market. Except that I'd like to work
through the design on my own, from start to end.

Given what I understand right now from a very short search on
the topic, the input should be taken as a maximum of 1.0Vrms
and the input's load should appear to be something like 10k
ohms. If someone knows different from that, I'll accept the
criticism and change that spec.

I'd like to consider a tone control and a volume control to
be included.

Output is to be into a small 8 ohm speaker. With that
maximum 1.0Vrms at the input and the volume control set to
maximum the wattage into 8 ohms should be around 10 watts.
Since human hearing won't tell much difference between 8
watts and 12 watts, this is a bit of a sloppy spec and I'm
open to anything in the area of 5-20 watts... though I'm
really wanting to keep the rail voltages down to something
modest and the BJTs not having to tolerate hugish Vce.

Now that I say this, an odd idea comes to mind because the
CFL light bulbs include two TO-220 BJTs that can handle quite
a high Vce on them. I could cannibalize those. But to be
honest, I'm still not needing high watt outputs. So there's
no reason to think about scavenging such parts.

I would like to design it to work into 4 ohms as a margin
bound and not as a design goal, but even 5.6 ohms would be
acceptable.

I'm not looking for this to be done quickly, either. If it
takes months of only occasional back-and-forth, I'm fine with
that. Also, I expect to do my work and don't expect someone
else to hand-hold me from complete ignorante to complete
enlightenment. I just need someone to slap my face when
I say something terribly stupid and/or point in a truely
useful direction when I need it. Or else someone who is
wanting to explore this with me and willing to work for it.

Is anyone here willing to consider a sincere discussion?

Jon

Yes, I've done a few amps... Truthfully it would be nice to do something
useful and it dovetails well with something else I am doing.

 

[Jon Kirwan]

On Wed, 27 Jan 2010 09:04:02 +1000, David Eather
<eather@tpg.com.au> wrote:

Jon Kirwan wrote:
I'd like to take a crack at thinking through a design of an
audio amplifier made up of discrete BJTs and other discrete
parts as an educational process.

I imagine this will be broken up into three sections; input
transconductance, transimpedance VAS, and output driver. But
other arrangements (such as combining the VAS and output
driver using a signal splitting BJT) would work for me, in
learning.

I said "BJT" and "discrete" but I'm also open to the idea of
using BJT pairs, such as the BCV61 and BCV62. In the case of
current mirrors, that may make sense. But not high-priced,
elite and/or hard to get, or obsolete. And no FETs. This is
to be about learning to design with BJTs.

SMT vs through-hole isn't an issue for learning about a
design, I suppose. If I need to build up some section and
test it with a signal, I'll probably want to do it quickly
and without having to buy services every step of the way. So
I may 'dead bug' SMT parts to get there. (The basic idea
here is to learn, not to make something tiny.)

Although I have some other applications, right now I'd like
the target use to be as a computer speaker system (not unlike
those dirt cheap, sadly almost all of them 10% THD, systems
sold today into this market. Except that I'd like to work
through the design on my own, from start to end.

Given what I understand right now from a very short search on
the topic, the input should be taken as a maximum of 1.0Vrms
and the input's load should appear to be something like 10k
ohms. If someone knows different from that, I'll accept the
criticism and change that spec.

I'd like to consider a tone control and a volume control to
be included.

Output is to be into a small 8 ohm speaker. With that
maximum 1.0Vrms at the input and the volume control set to
maximum the wattage into 8 ohms should be around 10 watts.
Since human hearing won't tell much difference between 8
watts and 12 watts, this is a bit of a sloppy spec and I'm
open to anything in the area of 5-20 watts... though I'm
really wanting to keep the rail voltages down to something
modest and the BJTs not having to tolerate hugish Vce.

Now that I say this, an odd idea comes to mind because the
CFL light bulbs include two TO-220 BJTs that can handle quite
a high Vce on them. I could cannibalize those. But to be
honest, I'm still not needing high watt outputs. So there's
no reason to think about scavenging such parts.

I would like to design it to work into 4 ohms as a margin
bound and not as a design goal, but even 5.6 ohms would be
acceptable.

I'm not looking for this to be done quickly, either. If it
takes months of only occasional back-and-forth, I'm fine with
that. Also, I expect to do my work and don't expect someone
else to hand-hold me from complete ignorante to complete
enlightenment. I just need someone to slap my face when
I say something terribly stupid and/or point in a truely
useful direction when I need it. Or else someone who is
wanting to explore this with me and willing to work for it.

Is anyone here willing to consider a sincere discussion?

Jon

Yes, I've done a few amps... Truthfully it would be nice to do something
useful and it dovetails well with something else I am doing.

Oh, my gosh! This is great to hear!!

Could you tell me what you know about the outputs of a PC
sound chip/board? As I gather the details right now, the
idea of 1Vrms max into a load of 10k ohm may be right. Do
you know any of the details?

Thanks,
Jon

 

[David Eather]

Jon Kirwan wrote:

On Wed, 27 Jan 2010 09:04:02 +1000, David Eather
eather@tpg.com.au> wrote:

Jon Kirwan wrote:
I'd like to take a crack at thinking through a design of an
audio amplifier made up of discrete BJTs and other discrete
parts as an educational process.

I imagine this will be broken up into three sections; input
transconductance, transimpedance VAS, and output driver. But
other arrangements (such as combining the VAS and output
driver using a signal splitting BJT) would work for me, in
learning.

I said "BJT" and "discrete" but I'm also open to the idea of
using BJT pairs, such as the BCV61 and BCV62. In the case of
current mirrors, that may make sense. But not high-priced,
elite and/or hard to get, or obsolete. And no FETs. This is
to be about learning to design with BJTs.

SMT vs through-hole isn't an issue for learning about a
design, I suppose. If I need to build up some section and
test it with a signal, I'll probably want to do it quickly
and without having to buy services every step of the way. So
I may 'dead bug' SMT parts to get there. (The basic idea
here is to learn, not to make something tiny.)

Although I have some other applications, right now I'd like
the target use to be as a computer speaker system (not unlike
those dirt cheap, sadly almost all of them 10% THD, systems
sold today into this market. Except that I'd like to work
through the design on my own, from start to end.

Given what I understand right now from a very short search on
the topic, the input should be taken as a maximum of 1.0Vrms
and the input's load should appear to be something like 10k
ohms. If someone knows different from that, I'll accept the
criticism and change that spec.

I'd like to consider a tone control and a volume control to
be included.

Output is to be into a small 8 ohm speaker. With that
maximum 1.0Vrms at the input and the volume control set to
maximum the wattage into 8 ohms should be around 10 watts.
Since human hearing won't tell much difference between 8
watts and 12 watts, this is a bit of a sloppy spec and I'm
open to anything in the area of 5-20 watts... though I'm
really wanting to keep the rail voltages down to something
modest and the BJTs not having to tolerate hugish Vce.

Now that I say this, an odd idea comes to mind because the
CFL light bulbs include two TO-220 BJTs that can handle quite
a high Vce on them. I could cannibalize those. But to be
honest, I'm still not needing high watt outputs. So there's
no reason to think about scavenging such parts.

I would like to design it to work into 4 ohms as a margin
bound and not as a design goal, but even 5.6 ohms would be
acceptable.

I'm not looking for this to be done quickly, either. If it
takes months of only occasional back-and-forth, I'm fine with
that. Also, I expect to do my work and don't expect someone
else to hand-hold me from complete ignorante to complete
enlightenment. I just need someone to slap my face when
I say something terribly stupid and/or point in a truely
useful direction when I need it. Or else someone who is
wanting to explore this with me and willing to work for it.

Is anyone here willing to consider a sincere discussion?

Jon
Yes, I've done a few amps... Truthfully it would be nice to do something
useful and it dovetails well with something else I am doing.

Oh, my gosh! This is great to hear!!

Don't get too excited, my brain has been badly scrambled. I often miss
things on a first pass - don't be afraid to check - I won't take offense.

Could you tell me what you know about the outputs of a PC
sound chip/board? As I gather the details right now, the
idea of 1Vrms max into a load of 10k ohm may be right. Do
you know any of the details?

Thanks,
Jon

 

[John Larkin]

On Tue, 26 Jan 2010 12:57:13 -0800, Jon Kirwan
<jonk@infinitefactors.org> wrote:

I'd like to take a crack at thinking through a design of an
audio amplifier made up of discrete BJTs and other discrete
parts as an educational process.

I imagine this will be broken up into three sections; input
transconductance, transimpedance VAS, and output driver. But
other arrangements (such as combining the VAS and output
driver using a signal splitting BJT) would work for me, in
learning.

I said "BJT" and "discrete" but I'm also open to the idea of
using BJT pairs, such as the BCV61 and BCV62. In the case of
current mirrors, that may make sense. But not high-priced,
elite and/or hard to get, or obsolete. And no FETs. This is
to be about learning to design with BJTs.

SMT vs through-hole isn't an issue for learning about a
design, I suppose. If I need to build up some section and
test it with a signal, I'll probably want to do it quickly
and without having to buy services every step of the way. So
I may 'dead bug' SMT parts to get there. (The basic idea
here is to learn, not to make something tiny.)

Although I have some other applications, right now I'd like
the target use to be as a computer speaker system (not unlike
those dirt cheap, sadly almost all of them 10% THD, systems
sold today into this market. Except that I'd like to work
through the design on my own, from start to end.

Given what I understand right now from a very short search on
the topic, the input should be taken as a maximum of 1.0Vrms
and the input's load should appear to be something like 10k
ohms. If someone knows different from that, I'll accept the
criticism and change that spec.

I'd like to consider a tone control and a volume control to
be included.

Output is to be into a small 8 ohm speaker. With that
maximum 1.0Vrms at the input and the volume control set to
maximum the wattage into 8 ohms should be around 10 watts.
Since human hearing won't tell much difference between 8
watts and 12 watts, this is a bit of a sloppy spec and I'm
open to anything in the area of 5-20 watts... though I'm
really wanting to keep the rail voltages down to something
modest and the BJTs not having to tolerate hugish Vce.

Now that I say this, an odd idea comes to mind because the
CFL light bulbs include two TO-220 BJTs that can handle quite
a high Vce on them. I could cannibalize those. But to be
honest, I'm still not needing high watt outputs. So there's
no reason to think about scavenging such parts.

I would like to design it to work into 4 ohms as a margin
bound and not as a design goal, but even 5.6 ohms would be
acceptable.

I'm not looking for this to be done quickly, either. If it
takes months of only occasional back-and-forth, I'm fine with
that. Also, I expect to do my work and don't expect someone
else to hand-hold me from complete ignorante to complete
enlightenment. I just need someone to slap my face when
I say something terribly stupid and/or point in a truely
useful direction when I need it. Or else someone who is
wanting to explore this with me and willing to work for it.

Is anyone here willing to consider a sincere discussion?

Jon

Back when transistors were young, and transistor manuals (GE, RCA)
were published, there were tons of such circuits around. They all
pretty much converged to a few forms, and haven't changed much since.

I could post some circuits from the old manuals, it that wouldn't
spoil what you want to do.

10 watts into computer speakers sounds like a lot. Most AM radios
didn't make one watt. You might experiment first to see how much power
you really need.

John

 

[Jon Kirwan]

On Tue, 26 Jan 2010 16:15:45 -0800, John Larkin
<jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Tue, 26 Jan 2010 12:57:13 -0800, Jon Kirwan
jonk@infinitefactors.org> wrote:

I'd like to take a crack at thinking through a design of an
audio amplifier made up of discrete BJTs and other discrete
parts as an educational process.

I imagine this will be broken up into three sections; input
transconductance, transimpedance VAS, and output driver. But
other arrangements (such as combining the VAS and output
driver using a signal splitting BJT) would work for me, in
learning.

I said "BJT" and "discrete" but I'm also open to the idea of
using BJT pairs, such as the BCV61 and BCV62. In the case of
current mirrors, that may make sense. But not high-priced,
elite and/or hard to get, or obsolete. And no FETs. This is
to be about learning to design with BJTs.

SMT vs through-hole isn't an issue for learning about a
design, I suppose. If I need to build up some section and
test it with a signal, I'll probably want to do it quickly
and without having to buy services every step of the way. So
I may 'dead bug' SMT parts to get there. (The basic idea
here is to learn, not to make something tiny.)

Although I have some other applications, right now I'd like
the target use to be as a computer speaker system (not unlike
those dirt cheap, sadly almost all of them 10% THD, systems
sold today into this market. Except that I'd like to work
through the design on my own, from start to end.

Given what I understand right now from a very short search on
the topic, the input should be taken as a maximum of 1.0Vrms
and the input's load should appear to be something like 10k
ohms. If someone knows different from that, I'll accept the
criticism and change that spec.

I'd like to consider a tone control and a volume control to
be included.

Output is to be into a small 8 ohm speaker. With that
maximum 1.0Vrms at the input and the volume control set to
maximum the wattage into 8 ohms should be around 10 watts.
Since human hearing won't tell much difference between 8
watts and 12 watts, this is a bit of a sloppy spec and I'm
open to anything in the area of 5-20 watts... though I'm
really wanting to keep the rail voltages down to something
modest and the BJTs not having to tolerate hugish Vce.

Now that I say this, an odd idea comes to mind because the
CFL light bulbs include two TO-220 BJTs that can handle quite
a high Vce on them. I could cannibalize those. But to be
honest, I'm still not needing high watt outputs. So there's
no reason to think about scavenging such parts.

I would like to design it to work into 4 ohms as a margin
bound and not as a design goal, but even 5.6 ohms would be
acceptable.

I'm not looking for this to be done quickly, either. If it
takes months of only occasional back-and-forth, I'm fine with
that. Also, I expect to do my work and don't expect someone
else to hand-hold me from complete ignorante to complete
enlightenment. I just need someone to slap my face when
I say something terribly stupid and/or point in a truely
useful direction when I need it. Or else someone who is
wanting to explore this with me and willing to work for it.

Is anyone here willing to consider a sincere discussion?

Jon

Back when transistors were young, and transistor manuals (GE, RCA)
were published, there were tons of such circuits around. They all
pretty much converged to a few forms, and haven't changed much since.

I remember reading in popular electronics about some audio
amps that I couldn't even come close to following at the
time. The series of them with the name 'tiger' in them.

I'm not so much interested in _circuits_, per se, as I am in
learning about topologies, various ideas built upon them, and
then the specific details of designing towards a specific
implementation. For example, I enjoyed learning about
bootstrapping as a general idea _and_ as it applies to a
couple of specific areas. Having both theory _and_ specific
practice helps firm up the ideas better.

I could post some circuits from the old manuals, it that wouldn't
spoil what you want to do.

It may serve as a point of discussion. Would you be willing
to discuss their details and the broader theories as applied?

10 watts into computer speakers sounds like a lot. Most AM radios
didn't make one watt. You might experiment first to see how much power
you really need.

Oh, I figure one watt is enough, too. As a practical matter
and as a consumer using a device and not as a designer trying
to learn something. That's what... 3Vrms? Into 8 ohms? A
voltage gain of 3, given 1Vrms input? I'm wanting to learn
some things, not place one BJT (okay, not really, but it
almost seems like that) down as an emitter follower and then
calling it good.

Up front, I thought I'd like to deal with perhaps something
on the order of about 10Vrms into 8 ohms. I figured that is
enough 'bad' that I'd have to cope with some interesting
corners along the way; but not enough 'bad' that I'd have to
deal with too much all at once.

For example, at around 10 watts or so, it's enough that I may
need to seriously consider avoiding class-A operation of the
output stage and move to class-B, instead. But it is low
enough that there is some room to discuss each, as well as
class-AB biasing, too. More power and I'm almost certain I'm
pushed into class-B. Less power and.. well, who cares that
much? At one watt or so, just class-A and be done with it? I
won't learn the reasoning behind trade-offs that way.

There's more. I just figured at about 10 watts I'm likely to
learn some things but not be forced to learn so much that I'm
overwhelmed.

I'm open to specific advice about all this, of course.

Jon

 

[Jon Kirwan]

On Wed, 27 Jan 2010 10:11:40 +1000, David Eather
<eather@tpg.com.au> wrote:

Jon Kirwan wrote:

snip of offer to discuss amplifier design

Oh, my gosh! This is great to hear!!

Don't get too excited, my brain has been badly scrambled. I often miss
things on a first pass - don't be afraid to check - I won't take offense.
snip of trailer unresponded to

hehe. Okay. It's just a topic I'd like to play with and I'm
glad there might be someone else out there to talk it over
with. What can I say? Except _thanks_!

Jon

 

[David Eather]

Jon Kirwan wrote:

On Tue, 26 Jan 2010 16:15:45 -0800, John Larkin
jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Tue, 26 Jan 2010 12:57:13 -0800, Jon Kirwan
jonk@infinitefactors.org> wrote:

I'd like to take a crack at thinking through a design of an
audio amplifier made up of discrete BJTs and other discrete
parts as an educational process.

I imagine this will be broken up into three sections; input
transconductance, transimpedance VAS, and output driver. But
other arrangements (such as combining the VAS and output
driver using a signal splitting BJT) would work for me, in
learning.

I said "BJT" and "discrete" but I'm also open to the idea of
using BJT pairs, such as the BCV61 and BCV62. In the case of
current mirrors, that may make sense. But not high-priced,
elite and/or hard to get, or obsolete. And no FETs. This is
to be about learning to design with BJTs.

SMT vs through-hole isn't an issue for learning about a
design, I suppose. If I need to build up some section and
test it with a signal, I'll probably want to do it quickly
and without having to buy services every step of the way. So
I may 'dead bug' SMT parts to get there. (The basic idea
here is to learn, not to make something tiny.)

Although I have some other applications, right now I'd like
the target use to be as a computer speaker system (not unlike
those dirt cheap, sadly almost all of them 10% THD, systems
sold today into this market. Except that I'd like to work
through the design on my own, from start to end.

Given what I understand right now from a very short search on
the topic, the input should be taken as a maximum of 1.0Vrms
and the input's load should appear to be something like 10k
ohms. If someone knows different from that, I'll accept the
criticism and change that spec.

I'd like to consider a tone control and a volume control to
be included.

Output is to be into a small 8 ohm speaker. With that
maximum 1.0Vrms at the input and the volume control set to
maximum the wattage into 8 ohms should be around 10 watts.
Since human hearing won't tell much difference between 8
watts and 12 watts, this is a bit of a sloppy spec and I'm
open to anything in the area of 5-20 watts... though I'm
really wanting to keep the rail voltages down to something
modest and the BJTs not having to tolerate hugish Vce.

Now that I say this, an odd idea comes to mind because the
CFL light bulbs include two TO-220 BJTs that can handle quite
a high Vce on them. I could cannibalize those. But to be
honest, I'm still not needing high watt outputs. So there's
no reason to think about scavenging such parts.

I would like to design it to work into 4 ohms as a margin
bound and not as a design goal, but even 5.6 ohms would be
acceptable.

I'm not looking for this to be done quickly, either. If it
takes months of only occasional back-and-forth, I'm fine with
that. Also, I expect to do my work and don't expect someone
else to hand-hold me from complete ignorante to complete
enlightenment. I just need someone to slap my face when
I say something terribly stupid and/or point in a truely
useful direction when I need it. Or else someone who is
wanting to explore this with me and willing to work for it.

Is anyone here willing to consider a sincere discussion?

Jon
Back when transistors were young, and transistor manuals (GE, RCA)
were published, there were tons of such circuits around. They all
pretty much converged to a few forms, and haven't changed much since.

I remember reading in popular electronics about some audio
amps that I couldn't even come close to following at the
time. The series of them with the name 'tiger' in them.

I'm not so much interested in _circuits_, per se, as I am in
learning about topologies, various ideas built upon them, and
then the specific details of designing towards a specific
implementation. For example, I enjoyed learning about
bootstrapping as a general idea _and_ as it applies to a
couple of specific areas. Having both theory _and_ specific
practice helps firm up the ideas better.

I could post some circuits from the old manuals, it that wouldn't
spoil what you want to do.

It may serve as a point of discussion. Would you be willing
to discuss their details and the broader theories as applied?

10 watts into computer speakers sounds like a lot. Most AM radios
didn't make one watt. You might experiment first to see how much power
you really need.

Oh, I figure one watt is enough, too. As a practical matter
and as a consumer using a device and not as a designer trying
to learn something. That's what... 3Vrms? Into 8 ohms? A
voltage gain of 3, given 1Vrms input? I'm wanting to learn
some things, not place one BJT (okay, not really, but it
almost seems like that) down as an emitter follower and then
calling it good.

Up front, I thought I'd like to deal with perhaps something
on the order of about 10Vrms into 8 ohms. I figured that is
enough 'bad' that I'd have to cope with some interesting
corners along the way; but not enough 'bad' that I'd have to
deal with too much all at once.

For example, at around 10 watts or so, it's enough that I may
need to seriously consider avoiding class-A operation of the
output stage and move to class-B, instead. But it is low
enough that there is some room to discuss each, as well as
class-AB biasing, too. More power and I'm almost certain I'm
pushed into class-B. Less power and.. well, who cares that
much? At one watt or so, just class-A and be done with it? I
won't learn the reasoning behind trade-offs that way.

There's more. I just figured at about 10 watts I'm likely to
learn some things but not be forced to learn so much that I'm
overwhelmed.

I'm open to specific advice about all this, of course.

Jon

I like 10 watts as a starting size - at this size you have to start
doing things the way the big amps do, but it is not so big as to be
outrageously expensive, for example you still use a relatively small
power supply, heatsinks, and inexpensive transistors, and at the end you
can use it with your PC and really blow those 320 watt PFPO (peak
fantasy power output)speakers away.

I like your hesitation on class A. You want an amp with some power
output and class A is very inefficient, never more than 25% and often
way less. This would add greatly to the cost - a 40 watt power supply,
heatsinks capable of getting rid of the same as heat while keeping the
transistor junction temperature low, and beefier transistors. You also
get to put up with a shorter service life from all that heat. The "big
thing" with class A is there is no crossover distortion, which can lead
to better overall distortion figures, but the cost is huge - a kit for 2
x 20 watt class A sells for $600.

My particular bias for an amp this size is to go class AB with a split
power supply. The majority of quality audio amps follow this topology
and this is, I think, I great reason to go down this design path (what
you learn is applicable in the most number of situations). I should hunt
down a schematics of what I'm seeing in the distance (which can/will
change as decisions are made) - some of the justifications will have to
wait

The first step is to think about the output. The basic equations are

(1).....Vout = sqrt(2*P*R)

With R as 8 ohms for a common speaker and 10 watts that is 12.7 volts -
actually +/- 12.7 volts with a split power supply.

(2).....Imax = sqrt(2*P/R)

This comes out to 1.6 amps. You should probably also consider the case
when R speaker = 4 ohms when initially selecting a transistor for the
output 2.2 amps - remember this is max output current. The power supply
voltage will have to be somewhat higher than Vout to take into account
circuit drive requirements, ripple on the power supply and transformer
regulation etc.

Are you OK with connecting mains to a transformer? or would you rather
use an AC plug pack (10 watts is about the biggest amp a plugpack can be
used for)? The "cost" for using an AC plug pack is you will need larger
filter capacitors.

I should also ask if you have a multi meter, oscilloscope (not necessary
but useful)and how is your soldering? But it would be wise to keep this
whole thing as a paper exercise before you commit to anything.

 

[Phil Allison]

"David Eather" <

** Learn to trimv- wanker!!

I like your hesitation on class A. You want an amp with some power output
and class A is very inefficient, never more than 25% and often way less.

** Class A amplifiers are up to 50 % efficient.

This would add greatly to the cost - a 40 watt power supply,

** 20 watts is all that is needed.

heatsinks capable of getting rid of the same as heat while keeping the
transistor junction temperature low,

** Junction temps can settle at 125 C with no problems.

and beefier transistors.

** Nonsense.

You also get to put up with a shorter service life from all that heat.

** Bull.

The "big thing" with class A is there is no crossover distortion, which
can lead to better overall distortion figures, but the cost is huge - a
kit for 2 x 20 watt class A sells for $600.

** Irrelevant what some unspecified kit sells for.

A 20 watt class A power stage only needs bigger heatsinks compared to usual
low bias, class AB operation.

My particular bias for an amp this size is to go class AB with a split
power supply. The majority of quality audio amps follow this topology

** Bollocks.

Only some very high powered hi-fi and pro-audio amps use additional DC
rails.

The power supply voltage will have to be somewhat higher than Vout to take
into account circuit drive requirements, ripple on the power supply and
transformer regulation etc.

** There is no variation in the DC current draw from a PSU with class A
amplifiers - in fact, this is the very definition of class A operation of
an audio amp. So the DC rails will remain steady from no drive to full
output.

Also, the heatsink will cool considerably when the amp is operated at full
sine wave power.

Why is SOOO much bollocks posted about something so very simple ??



..... Phil

 

[Jon Kirwan]

On Wed, 27 Jan 2010 17:31:00 +1000, David Eather
<eather@tpg.com.au> wrote:

Jon Kirwan wrote:
On Tue, 26 Jan 2010 16:15:45 -0800, John Larkin
jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Tue, 26 Jan 2010 12:57:13 -0800, Jon Kirwan
jonk@infinitefactors.org> wrote:

I'd like to take a crack at thinking through a design of an
audio amplifier made up of discrete BJTs and other discrete
parts as an educational process.

I imagine this will be broken up into three sections; input
transconductance, transimpedance VAS, and output driver. But
other arrangements (such as combining the VAS and output
driver using a signal splitting BJT) would work for me, in
learning.

I said "BJT" and "discrete" but I'm also open to the idea of
using BJT pairs, such as the BCV61 and BCV62. In the case of
current mirrors, that may make sense. But not high-priced,
elite and/or hard to get, or obsolete. And no FETs. This is
to be about learning to design with BJTs.

SMT vs through-hole isn't an issue for learning about a
design, I suppose. If I need to build up some section and
test it with a signal, I'll probably want to do it quickly
and without having to buy services every step of the way. So
I may 'dead bug' SMT parts to get there. (The basic idea
here is to learn, not to make something tiny.)

Although I have some other applications, right now I'd like
the target use to be as a computer speaker system (not unlike
those dirt cheap, sadly almost all of them 10% THD, systems
sold today into this market. Except that I'd like to work
through the design on my own, from start to end.

Given what I understand right now from a very short search on
the topic, the input should be taken as a maximum of 1.0Vrms
and the input's load should appear to be something like 10k
ohms. If someone knows different from that, I'll accept the
criticism and change that spec.

I'd like to consider a tone control and a volume control to
be included.

Output is to be into a small 8 ohm speaker. With that
maximum 1.0Vrms at the input and the volume control set to
maximum the wattage into 8 ohms should be around 10 watts.
Since human hearing won't tell much difference between 8
watts and 12 watts, this is a bit of a sloppy spec and I'm
open to anything in the area of 5-20 watts... though I'm
really wanting to keep the rail voltages down to something
modest and the BJTs not having to tolerate hugish Vce.

Now that I say this, an odd idea comes to mind because the
CFL light bulbs include two TO-220 BJTs that can handle quite
a high Vce on them. I could cannibalize those. But to be
honest, I'm still not needing high watt outputs. So there's
no reason to think about scavenging such parts.

I would like to design it to work into 4 ohms as a margin
bound and not as a design goal, but even 5.6 ohms would be
acceptable.

I'm not looking for this to be done quickly, either. If it
takes months of only occasional back-and-forth, I'm fine with
that. Also, I expect to do my work and don't expect someone
else to hand-hold me from complete ignorante to complete
enlightenment. I just need someone to slap my face when
I say something terribly stupid and/or point in a truely
useful direction when I need it. Or else someone who is
wanting to explore this with me and willing to work for it.

Is anyone here willing to consider a sincere discussion?

Jon
Back when transistors were young, and transistor manuals (GE, RCA)
were published, there were tons of such circuits around. They all
pretty much converged to a few forms, and haven't changed much since.

I remember reading in popular electronics about some audio
amps that I couldn't even come close to following at the
time. The series of them with the name 'tiger' in them.

I'm not so much interested in _circuits_, per se, as I am in
learning about topologies, various ideas built upon them, and
then the specific details of designing towards a specific
implementation. For example, I enjoyed learning about
bootstrapping as a general idea _and_ as it applies to a
couple of specific areas. Having both theory _and_ specific
practice helps firm up the ideas better.

I could post some circuits from the old manuals, it that wouldn't
spoil what you want to do.

It may serve as a point of discussion. Would you be willing
to discuss their details and the broader theories as applied?

10 watts into computer speakers sounds like a lot. Most AM radios
didn't make one watt. You might experiment first to see how much power
you really need.

Oh, I figure one watt is enough, too. As a practical matter
and as a consumer using a device and not as a designer trying
to learn something. That's what... 3Vrms? Into 8 ohms? A
voltage gain of 3, given 1Vrms input? I'm wanting to learn
some things, not place one BJT (okay, not really, but it
almost seems like that) down as an emitter follower and then
calling it good.

Up front, I thought I'd like to deal with perhaps something
on the order of about 10Vrms into 8 ohms. I figured that is
enough 'bad' that I'd have to cope with some interesting
corners along the way; but not enough 'bad' that I'd have to
deal with too much all at once.

For example, at around 10 watts or so, it's enough that I may
need to seriously consider avoiding class-A operation of the
output stage and move to class-B, instead. But it is low
enough that there is some room to discuss each, as well as
class-AB biasing, too. More power and I'm almost certain I'm
pushed into class-B. Less power and.. well, who cares that
much? At one watt or so, just class-A and be done with it? I
won't learn the reasoning behind trade-offs that way.

There's more. I just figured at about 10 watts I'm likely to
learn some things but not be forced to learn so much that I'm
overwhelmed.

I'm open to specific advice about all this, of course.

Jon

I like 10 watts as a starting size - at this size you have to start
doing things the way the big amps do, but it is not so big as to be
outrageously expensive, for example you still use a relatively small
power supply, heatsinks, and inexpensive transistors,

You appear to confirm my instincts.

and at the end you
can use it with your PC and really blow those 320 watt PFPO (peak
fantasy power output)speakers away.

Well, mostly I'm just trying to learn... not impress others
about the results.

I like your hesitation on class A. You want an amp with some power
output and class A is very inefficient, never more than 25% and often
way less. This would add greatly to the cost - a 40 watt power supply,
heatsinks capable of getting rid of the same as heat while keeping the
transistor junction temperature low, and beefier transistors. You also
get to put up with a shorter service life from all that heat. The "big
thing" with class A is there is no crossover distortion, which can lead
to better overall distortion figures, but the cost is huge - a kit for 2
x 20 watt class A sells for $600.

Egads. My instincts said class-A would add a lot to weight
and cost, but no idea a mere 20W kit could sell for $300!

My particular bias for an amp this size is to go class AB with a split
power supply. The majority of quality audio amps follow this topology
and this is, I think, I great reason to go down this design path (what
you learn is applicable in the most number of situations). I should hunt
down a schematics of what I'm seeing in the distance (which can/will
change as decisions are made) - some of the justifications will have to
wait

I'm fine with taking things as they come.

As far as the class, I guessed that at 10 watts class-A would
be too power-hungry and probably not worth its weight but
that class-AB might be okay.

I have to warn you, though, that I'm not focused upon some
20ppm THD. I'd like to learn, not design something whose
distortion (or noise, for that matter) is around a bit on a
16-bit DAC or less. I figure winding up close to class-B
operation in the end. But I'd like to take the walk along
the way, so to speak.

The first step is to think about the output. The basic equations are

(1).....Vout = sqrt(2*P*R)

With R as 8 ohms for a common speaker and 10 watts that is 12.7 volts -
actually +/- 12.7 volts with a split power supply.

If you don't mind, I'd like to discuss this more closely. Not
just have it tossed out. So, P=V*I; or P=Vrms^2/R with AC.
Using Vpeak=SQRT(2)*Vrms, I get your Vpeak=SQRT(2*P*R)
equation. Which suggests the +/-12.7V swing. Which further
suggests, taking Vce drops and any small amounts emitter
resistor drops into account, something along the lines of +/-
14-15V rails?

Or should the rails be cut a lot closer to the edge here to
improve efficiency. What bothers me is saturation as Vce on
the final output BJTs goes well below 1V each and beta goes
away, as well, rapidly soaking up remaining drive compliance.

(2).....Imax = sqrt(2*P/R)

This comes out to 1.6 amps. You should probably also consider the case
when R speaker = 4 ohms when initially selecting a transistor for the
output 2.2 amps - remember this is max output current. The power supply
voltage will have to be somewhat higher than Vout to take into account
circuit drive requirements, ripple on the power supply and transformer
regulation etc.

Okay. I missed reading this when writing the above. Rather
than correct myself, I'll leave my thinking in place.

So yes, the rails will need to be a bit higher. Agreed. On
this subject, I'm curious about the need to _isolate_, just a
little, the rails used by the input stage vs the output stage
rails. I'm thinking an RC (or LC for another pole?) for
isolation. But I honestly don't know if that's helpful, or
not.

Are you OK with connecting mains to a transformer? or would you rather
use an AC plug pack (10 watts is about the biggest amp a plugpack can be
used for)? The "cost" for using an AC plug pack is you will need larger
filter capacitors.

I'd much prefer to __avoid__ using someone else's "pack" for
the supply. All discrete parts should be on the table, so to
speak, in plain view. And I don't imagine _any_ conceptual
difficulties for this portion of the design. I'm reasonably
familiar with transformers, rectifiers, ripple calculations,
and how to consider peak charging currents vs averge load
currents as they relate to the phase angles available for
charging the caps. So on this part, I may need less help
than elsewhere. In other words, I'm somewhat comfortable
here.

I should also ask if you have a multi meter, oscilloscope (not necessary
but useful)and how is your soldering? But it would be wise to keep this
whole thing as a paper exercise before you commit to anything.

I have a 6 1/2 digit HP multimeter, a Tek DMM916 true RMS
handheld, two oscilloscopes (TEK 2245 with voltmeter option
and an HP 54645D), three triple-output power supplies with
two of them GPIB drivable, the usual not-too-expensive signal
generator, and a fair bunch of other stuff on the shelves.
Lots of probes, clips, and so on. For soldering, I'm limited
to a Weller WTCPT and some 0.4mm round, 0.8mm spade, and
somewhat wider spade tips in the 1.5mm area. I have tubs and
jars of various types of fluxes, as well, and wire wrap tools
and wire wrap wire, as well. I also have a room set aside
for this kind of stuff, when I get time to play.

Jon

 

[David Eather]

Phil Allison wrote:

"David Eather"

** Learn to trimv- wanker!!

I like your hesitation on class A. You want an amp with some power output
and class A is very inefficient, never more than 25% and often way less.

** Class A amplifiers are up to 50 % efficient.

Yes, if you use a matching transformer, not really an option here

This would add greatly to the cost - a 40 watt power supply,

** 20 watts is all that is needed.

IBID

heatsinks capable of getting rid of the same as heat while keeping the
transistor junction temperature low,

** Junction temps can settle at 125 C with no problems.

Most transistors are spec'ed at 125 C for 1000 or 2000 hours. Most
people would like there devices to run a bit longer than this. At least
one semiconductor manufacturer believes that for each lowering of the
junction temperature by 10 C doubles the life of transistor

and beefier transistors.

** Nonsense.

needed for lower Th(jc) and Th(ch)

You also get to put up with a shorter service life from all that heat.

** Bull.

I think just about everyone will disagree with you on that point. heat
dries out electrolytics and heat stress cycling is a major failure mode
of semiconductors

The "big thing" with class A is there is no crossover distortion, which
can lead to better overall distortion figures, but the cost is huge - a
kit for 2 x 20 watt class A sells for $600.

** Irrelevant what some unspecified kit sells for.
No, it is an example of how costly class A can become. I don't think

anyone can realistically find a way to make a 2 x 20 watt class AB amp
kit sell for anything like that price (short of gold plated everything
and mil spec components). The kit price I quoted was the magazine
"Silicon Chip" May-September 2007 design from Altronics
http://www.altronics.com.au/index.asp?area=item&id=K5125

A 20 watt class A power stage only needs bigger heatsinks compared to usual
low bias, class AB operation.

That was my part of my point. Thanks for making it so exactly.

My particular bias for an amp this size is to go class AB with a split
power supply. The majority of quality audio amps follow this topology

** Bollocks.

Well, somewhat recently the trend has gone to mono block IC amps so I'll
take that on the chin

Only some very high powered hi-fi and pro-audio amps use additional DC
rails.

So what. Even you say "some". Class G is not yet very common.

The power supply voltage will have to be somewhat higher than Vout to take
into account circuit drive requirements, ripple on the power supply and
transformer regulation etc.


** There is no variation in the DC current draw from a PSU with class A
amplifiers - in fact, this is the very definition of class A operation of
an audio amp. So the DC rails will remain steady from no drive to full
output.

True! But you may still need extra voltage for the circuit and you also
need extra voltage for the power supply ripple (unless you use a
*regulated* power supply) - which neither you or I specified.

Also, the heatsink will cool considerably when the amp is operated at full
sine wave power.

Not relevant at the moment.

Why is SOOO much bollocks posted about something so very simple ??



..... Phil

Phil,

Your welcome to take the lead and go through an amplifier design. Just
say the word.

 

[Bob Masta]

On Tue, 26 Jan 2010 12:57:13 -0800, Jon Kirwan
<jonk@infinitefactors.org> wrote:

I'd like to take a crack at thinking through a design of an
audio amplifier made up of discrete BJTs and other discrete
parts as an educational process.

snip

When I was first getting interested in power amp
design (back in the '70s) I started collecting
schematics for all the power amps I could get my
hands on, to compare them. I noticed that the
schematics for simple bipolar op-amp ICs were
remarkably similar to those for big discrete power
amps. If you have an old National Linear Databook
(or don't mind a lot of rooting around on the Web
for individual datasheets), you might take a look.

You can build a pretty decent amp with only a
handful of transistors. The same basic circuit
can be used for a wide range of output powers,
just by changing the power supply voltages and the
output device ratings.

Best regards,


Bob Masta

DAQARTA v5.00
Data AcQuisition And Real-Time Analysis
www.daqarta.com
Scope, Spectrum, Spectrogram, Sound Level Meter
Frequency Counter, FREE Signal Generator
Pitch Track, Pitch-to-MIDI
DaqMusic - FREE MUSIC, Forever!
(Some assembly required)
Science (and fun!) with your sound card!

 

[Phil Allison]

"David Eather = a LIAR and a Wanker "

** Learn to trim - wanker!!

I like your hesitation on class A. You want an amp with some power
output and class A is very inefficient, never more than 25% and often
way less.

** Class A amplifiers are up to 50 % efficient.

Yes, if you use a matching transformer,

** 100%, absolute BULLSHIT !!!!!!!

YOU KNOW NOTHING MORON !!!!!!!

This would add greatly to the cost - a 40 watt power supply,

** 20 watts is all that is needed.

IBID

** Fuck you - you damn imbecile.

heatsinks capable of getting rid of the same as heat while keeping the
transistor junction temperature low,

** Junction temps can settle at 125 C with no problems.

Most transistors are spec'ed at 125 C for 1000 or 2000 hours.

** More, 100% absolute BULLSHIT !!!!!!!

This RATBAG just keeps piling on the LIES !!!!!

people would like there devices to run a bit longer than this. At least
one semiconductor manufacturer believes that for each lowering of the
junction temperature by 10 C doubles the life of transistor

** MORE 100% absolute BULLSHIT !!!!!!!

This RATBAG just keeps piling on the damn LIES !!

You also get to put up with a shorter service life from all that heat.

** Bull.

I think just about everyone will disagree with you on that point.

** What some fuckwit like you choses to "think" is of zero consequence.

heat dries out electrolytics

** ROTFLMAO !!!

Shame BJTs are not electrolytic.

You FUCKING MORON !!!!!!!!

and heat stress cycling is a major failure mode of semiconductors

** No heat cycling occurs with class A.

Huge amounts occur with all class B and AB designs.

You FUCKING KNOW NOTHING MORON !!!!!!!!

The "big thing" with class A is there is no crossover distortion, which
can lead to better overall distortion figures, but the cost is huge - a
kit for 2 x 20 watt class A sells for $600.

** Irrelevant what some unspecified kit sells for.

No, it is an example of how costly class A can become.

** Absolute fucking CRAP !!!!!!!!

YOU are an example of how STUPID people can become.

An extreme example of an damn LIAR too.

A 20 watt class A power stage ONLY needs bigger heatsinks compared to
usual low bias, class AB operation.

** No big expense there.

My particular bias for an amp this size is to go class AB with a split
power supply. The majority of quality audio amps follow this topology

** Bollocks.

Well, somewhat recently the trend has gone to mono block IC amps so I'll
take that on the chin

** Go drop dead - you pig ignorant LYING WANKER !

Only some very high powered hi-fi and pro-audio amps use additional DC
rails.


So what.

** Proves you WRONG - you stinking LIAR !!!!!!

The power supply voltage will have to be somewhat higher than Vout to
take into account circuit drive requirements, ripple on the power supply
and transformer regulation etc.


** There is no variation in the DC current draw from a PSU with class A
amplifiers - in fact, this is the very definition of class A operation
of an audio amp. So the DC rails will remain steady from no drive to
full output.

True!

** Proves you WRONG AGAIN - you stinking LIAR !!!!!!

Also, the heatsink will cool considerably when the amp is operated at
full
sine wave power.

Not relevant at the moment.

** Course it is - you DAMN LIAR !!!!!!!!!

GO DROP FUCKING DEAD

- YOU BULLSHITTING IMBECILE !!!!!!!!!




..... Phil

 

[Phil Allison]

"Jon Kirwan is a FUCKWIT TROLL "


DO NOT FEED THE TROLL !!!!!!!!!!!!!!!

DO NOT FEED THE TROLL !!!!!!!!!!!!!!!


TROLLS are DESTROYERS of all NEWSGROUPS.
-------------------------------------------------------------



.... Phil

 

[amdx]

"Phil Allison" <phil_a@tpg.com.au> wrote in message
news:7sau8iFahqU1@mid.individual.net...

"Jon Kirwan is a FUCKWIT TROLL "


DO NOT FEED THE TROLL !!!!!!!!!!!!!!!

DO NOT FEED THE TROLL !!!!!!!!!!!!!!!


TROLLS are DESTROYERS of all NEWSGROUPS.
-------------------------------------------------------------



... Phil

Well Jon,
When you said,
"I just need someone to slap my face......"
Phil was the first one I thought of.
Mike

 

[pimpom]

Jon Kirwan wrote:

On Tue, 26 Jan 2010 16:15:45 -0800, John Larkin
jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Tue, 26 Jan 2010 12:57:13 -0800, Jon Kirwan
jonk@infinitefactors.org> wrote:

I'd like to take a crack at thinking through a design of an
audio amplifier made up of discrete BJTs and other discrete
parts as an educational process.

.......<snip>.........

Back when transistors were young, and transistor manuals (GE,
RCA)
were published, there were tons of such circuits around. They
all
pretty much converged to a few forms, and haven't changed much
since.

I remember reading in popular electronics about some audio
amps that I couldn't even come close to following at the
time. The series of them with the name 'tiger' in them.

That was probably an adaptation of RCA's 70-watt power amp,
published in their transistor manual of the mid-60s. A true
70-watt continuous output was pretty hefty then. Still is, in
fact, for many applications. The design was state-of-the-art,
using their 4000 series transistors which were specifically made
for audio. THD was <0.25% at 70W (pretty good for the time).

IIRC, Popular Electronics published it virtually unchanged (I
don't think they credited RCA with the design). I vaguely recall
their calling it 'tiger' something and the article title included
the word "indestructible". The latter term was because it
included current clamping for the output transistors as
short-circuit protection, a thermal fuse and a normal fuse. PE
claimed that they abused it with short-circuits and reactive
loads, and the worst they got was a blown fuse.

I remember that design fondly, if not perfectly, because I spent
a lot of time analysing it and others in the RCA manual.

 

[Jon Kirwan]

On Wed, 27 Jan 2010 13:23:28 GMT, Bob Masta wrote:

On Tue, 26 Jan 2010 12:57:13 -0800, Jon Kirwan
jonk@infinitefactors.org> wrote:

I'd like to take a crack at thinking through a design of an
audio amplifier made up of discrete BJTs and other discrete
parts as an educational process.

snip

When I was first getting interested in power amp
design (back in the '70s) I started collecting
schematics for all the power amps I could get my
hands on, to compare them. I noticed that the
schematics for simple bipolar op-amp ICs were
remarkably similar to those for big discrete power
amps. If you have an old National Linear Databook
(or don't mind a lot of rooting around on the Web
for individual datasheets), you might take a look.

You can build a pretty decent amp with only a
handful of transistors. The same basic circuit
can be used for a wide range of output powers,
just by changing the power supply voltages and the
output device ratings.

Best regards,

Thanks, Bob. Audio amplifiers, especially ones delivering
_some_ power, seem to offer such an excellent way to learn.
The basic idea, at a behavioral level, is fairly simple. An
implementation requires some knowledge and thought in the
end. So the destination is arrived at by taking a great path
to walk, with such wonderful vistas to see, I think. Much of
interest is along the way of getting there.

I may have an old National databook on linear parts
somewhere. I keep a lot, but I also have several thousand
books in my library which covers all of the walls in one of
the rooms. I'm at a point now where to get room for more
books, others must be boxed and stored or simply destroyed
and pulped. So it's a _maybe_.

One of the nice things (to me) about this kind of a path,
too, is that what I learn can be used for lots of things. An
audio amplifier is, in effect, not that much different from
an op amp. There is the usual basic idea of open loop gain
and closed loop gain with negative feedback, phase margins,
problems to solve over a frequency range spanning many
decades, and so on.

A completely separate project I'd like to play with, which
this learning will help prepare me for, is designing a pin
driver. I'd like to sink or source a programmable current
spanning decades from perhaps 100nA to perhaps 100uA while
reading the voltage at the node, as well as being able to
program a low impedance voltages spanning from -15V to +15V
there and read the current, or read a voltage at the same
node while presenting a fairly high impedence to it. I
imagine what I learn here will aid me there. And I'd like to
do this at some speed, as well. I may then start with a BJT
tester, for example, making up only three of these to start
and tying them into a micro for playing. Expanding that for
other purposes, later. It would be fun.

Jon

 

[Michael A. Terrell]

pimpom wrote:

Jon Kirwan wrote:
On Tue, 26 Jan 2010 16:15:45 -0800, John Larkin
jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Tue, 26 Jan 2010 12:57:13 -0800, Jon Kirwan
jonk@infinitefactors.org> wrote:

I'd like to take a crack at thinking through a design of an
audio amplifier made up of discrete BJTs and other discrete
parts as an educational process.

......<snip>.........


Back when transistors were young, and transistor manuals (GE,
RCA)
were published, there were tons of such circuits around. They
all
pretty much converged to a few forms, and haven't changed much
since.

I remember reading in popular electronics about some audio
amps that I couldn't even come close to following at the
time. The series of them with the name 'tiger' in them.


That was probably an adaptation of RCA's 70-watt power amp,
published in their transistor manual of the mid-60s. A true
70-watt continuous output was pretty hefty then. Still is, in
fact, for many applications. The design was state-of-the-art,
using their 4000 series transistors which were specifically made
for audio. THD was <0.25% at 70W (pretty good for the time).

That was the RCA 404xx series of their house numbered transistors. I
think they used:
1 40406
1 40407
1 40408
1 40409
1 40410
2 40411

and a couple 1N series metal cased diodes for temperature sensing.

It's been 40 years since I built that PE Tiger amp and the preamp
that went with it.

IIRC, Popular Electronics published it virtually unchanged (I
don't think they credited RCA with the design). I vaguely recall
their calling it 'tiger' something and the article title included
the word "indestructible". The latter term was because it
included current clamping for the output transistors as
short-circuit protection, a thermal fuse and a normal fuse. PE
claimed that they abused it with short-circuits and reactive
loads, and the worst they got was a blown fuse.

I remember that design fondly, if not perfectly, because I spent
a lot of time analysing it and others in the RCA manual.

--
Greed is the root of all eBay.

 

[Jon Kirwan]

On Wed, 27 Jan 2010 23:54:15 +0530, "pimpom"
<pimpom@invalid.invalid> wrote:

Jon Kirwan wrote:
On Tue, 26 Jan 2010 16:15:45 -0800, John Larkin
jjlarkin@highNOTlandTHIStechnologyPART.com> wrote:

On Tue, 26 Jan 2010 12:57:13 -0800, Jon Kirwan
jonk@infinitefactors.org> wrote:

I'd like to take a crack at thinking through a design of an
audio amplifier made up of discrete BJTs and other discrete
parts as an educational process.

......<snip>.........


Back when transistors were young, and transistor manuals (GE,
RCA)
were published, there were tons of such circuits around. They
all
pretty much converged to a few forms, and haven't changed much
since.

I remember reading in popular electronics about some audio
amps that I couldn't even come close to following at the
time. The series of them with the name 'tiger' in them.

That was probably an adaptation of RCA's 70-watt power amp,
published in their transistor manual of the mid-60s. A true
70-watt continuous output was pretty hefty then. Still is, in
fact, for many applications. The design was state-of-the-art,
using their 4000 series transistors which were specifically made
for audio. THD was <0.25% at 70W (pretty good for the time).

IIRC, Popular Electronics published it virtually unchanged (I
don't think they credited RCA with the design). I vaguely recall
their calling it 'tiger' something and the article title included
the word "indestructible". The latter term was because it
included current clamping for the output transistors as
short-circuit protection, a thermal fuse and a normal fuse. PE
claimed that they abused it with short-circuits and reactive
loads, and the worst they got was a blown fuse.

I remember that design fondly, if not perfectly, because I spent
a lot of time analysing it and others in the RCA manual.

Thanks for that bit. I remember looking at the first article
and wondering about trying my hand at building it. There was
no way I was prepared to understand it, though. The largest
problem I faced at the time, besides my own limitations in
education, was funds. I couldn't afford to even buy the
boards they offered, let alone the parts. So it was a non-
starter for me. I got my parts by scavenging TV sets and
radios others threw away. One of my larger hauls was when a
tornado knocked down a bowling alley and I called up the
owner and received permission to walk through the mess and
extract parts. I _never_ paid for anything. (Dad had died
when I was 7 and I literally had to work the fields picking
vegetables to earn enough to survive.)

Now, I might go back. But to be honest, I'd much prefer
being able to ask questions as they arise and work on
refining as I go. I learn more from a "movie in progress"
than studying a "snapshot," I guess.

Jon