Audio Amp Crossover Distortion?

On Tue, 17 Jan 2012 17:57:06 -0500, Jamie
jamie_ka1lpa_not_valid_after_ka1lpa__at_charter.net> wrote:


Jim Thompson wrote:


On Mon, 16 Jan 2012 19:59:06 -0500, Jamie
jamie_ka1lpa_not_valid_after_ka1lpa__at_charter.net> wrote:



Jim Thompson wrote:



On Mon, 16 Jan 2012 19:07:38 -0500, Jamie
jamie_ka1lpa_not_valid_after_ka1lpa__at_charter.net> wrote:




Jim Thompson wrote:




On Mon, 16 Jan 2012 09:34:51 -0500, Jamie
jamie_ka1lpa_not_valid_after_ka1lpa__at_charter.net> wrote:





Bill Bowden wrote:





On Jan 15, 11:40 am, Jamie

jamie_ka1lpa_not_valid_after_ka1l...@charter.net> wrote:




I don 't understand why you don't see it in the sim?

I see all kinds of problems there.

THe outputs are current modes and the beta on the outputs are

most



likely are not going to match well.

On top of that, LTspice shows the upper output (PNP) going into
discontinue state at the cross over. This is going to give you a

period



of what I call a flat liner and 99% sure this is where your cross

over



error is coming from.

Plot the current on R12.

Jamie

Yes, I did view the current through R12 which looks normal. The thing
operates class AB, so only one transistor is on at a time, so a 50%
discontinuous current is normal. I did improve the distortion using a
1458 op-amp in place of the 358. Looks much better now. The problem
now is I only get 1.5 volts peak into 8 ohms with an 8 volt supply and
I was trying for 3 volts or more. The HFE figure for the 2N2219A is
minimum 40 at 500mA or 7.5mA at 300mA. The 120 ohm resistor draws .
7/120 = about 6 mA so the op-amp must deliver 13.5 mA and the spec
sheet says only 10 mA short circuit. So, apparently, it needs higher
gain transistors or an op-amp with lower output impedance, or both.
Any ideas?

-Bill

That op-amp does not pull the reals, the 358 will do that effect on the
load side and there by give you more v to bias the transistors.

I don't think you have a current demand problem, you may have a rail
to rail problem how ever.

The 1458, as old as it is, still has a lot of usages. The las time I
looked, that op-amp (dual) only provides ~ Vcc-1.5 and Vee-1.5. Here you
have lost 3 volts to start with. This now gives you 5 volts to play with.

Of course, you really don't want to saturate the amp, so lets assume
you have only 4.5V to work with..

split that in half, since you looks apparrent you are operating in
Class A state on the output side of the op-amp and you get ~ 2.25 volts
Peak to play with.

Now., let us not forget, the minimum required for each of those
transistors to start working. ~ 0.7 and then times this by 2 and you get
1.4. Remove that value off the top and you are now getting closer to
where the problem is.

That configuration you're using in the first place is fighting against
you. As one side is conducting the other side is still conducting, just
about all the way through. This is going to remove a good chunk of your
output.

Have you considered a config like the following or something in this line?



8Volts-----------------------------+
+ |
|
|
+
|
+-----------------+|
___ | |\
+--+|___|-+----------------++ |
| | | +
+ | | |
|| | |\+ | | 1Ku
-||+--++-------------+|-\ | | ||
|| | >+-------+----------+---+||+-----+
+-+-|+/ | || |
4Volts |/+ | +
| | |
| | .-.
| | | |
+ + | |8
| |/ '-'
+-----------------+| |
|> GND
+
|
|
===
GND


Jamie



What sets the quiescent current thru the PNP-NPN path?

...Jim Thompson

The quiescent I of the op-amp and load.

In Ltspice it was ~ 70 ma with that basic circuit in both
the PNP and NPN. I was using 700..800 ma trannies.

Putting an R between Vcc and B of the PNP will drop that down
a bit and bring it closer to being symmetrical. Also, you can
use a R on the op output to the collector bridge to lower the
Quies.

I Didn't save the spice I used but I am sure I can put it together
again and post it if you wish, I just used what was in the stock lib.


Jamie



Try it in the real world. Have lots of spare transistors on hand

...Jim Thompson

Actually Jim, I have done that in real world cases. Which is why I
suggested it.. It does work when you use the correct set of components
and is a basic of many designs that i've seen over the years. A very
particle way to having current outputs

Old time 741 with Emitter outputs and related type power amp Ic's
work very nicely with a config like that. Of course in many cases, you
need to tailor the bias a little with some bypass R's on the Rail to the
base of each side if the Qu current is too high.

I had a load of LM380's years ago that I put to some good use making
little half bridge servos with a circuit of that type.

The only problem with that design is, if the driving op-amp happens
to circuit for some reason. It will take out the outputs. A current
limiting R on the op-amp output to the collector node is a good
practice, something I did not use here in this example. And don't forget
the bypass bias R's incase the QU Is too high in the chip that is used.


I know your an old pro and I am sure you have had your share of
smoking some silicon. :)

Jamie



Larkin will bless your work. Me? I do not recommend it... do
yourself a favor and do some Algebra. On a job interview I'd toss
your ass in a blink

...Jim Thompson

Jim, I don't need to, I've assembled circuits like that, that are still
operating today.


OK. Show me your math. What's the quiescent current thru the PNP-NPN
path?


You can toss my ass all you want, all it will get you is a bad back..

P.S.
I never used a simulator all these years until recently, my math
skills have served me well.

Jamie



Those Youtube videos Bitrex found of IIT lectures are based on chips I
designed 50 years ago... all math... no simulator existed until about
25 years into my career. Also have some of my chips used as class
examples at Rochester Institute of Technology.

Calculate that quiescent current and get back to me with the result.
(Show all your work

...Jim Thompson
I suppose I could, but I don't know where that would lead me to?

On Wed, 18 Jan 2012 01:18:41 +0000 (UTC), Kaz Kylheku <kaz_at_kylheku.com> wrote:


On 2012-01-18, krw_at_att.bizzzzzzzzzzzz <krw_at_att.bizzzzzzzzzzzz> wrote:

On Tue, 17 Jan 2012 16:23:22 -0700, Jim Thompson

Those Youtube videos Bitrex found of IIT lectures are based on chips I
designed 50 years ago... all math... no simulator existed until about
25 years into my career. Also have some of my chips used as class
examples at Rochester Institute of Technology.

25 years? Circuit simulators existed more than 40 years ago.

You're assuming Jim's career launched 50 years ago, when upon walking into a
firm off the street without a clue, he got a job designing integrated circuits.
Then, 25 years later, simulators appeared.


Well, he's just about to turn 19 (76 in human years).


Indeed, it is quite obvious that a little math around here wouldn't be such a
bad thing.


You might try it. Unless he started his career before age 11 (76-40-25), I
doubt his statement. Note that this assumes I used the very first circuit
simulators, which I highly doubt.
The only thing I ever did was tried out Verilog and I absolutely hated it..

krw_at_att.bizzzzzzzzzzzz wrote:

On Wed, 18 Jan 2012 01:18:41 +0000 (UTC), Kaz Kylheku <kaz_at_kylheku.com> wrote:


On 2012-01-18, krw_at_att.bizzzzzzzzzzzz <krw_at_att.bizzzzzzzzzzzz> wrote:

On Tue, 17 Jan 2012 16:23:22 -0700, Jim Thompson

Those Youtube videos Bitrex found of IIT lectures are based on chips I
designed 50 years ago... all math... no simulator existed until about
25 years into my career. Also have some of my chips used as class
examples at Rochester Institute of Technology.

25 years? Circuit simulators existed more than 40 years ago.

You're assuming Jim's career launched 50 years ago, when upon walking into a
firm off the street without a clue, he got a job designing integrated circuits.
Then, 25 years later, simulators appeared.


Well, he's just about to turn 19 (76 in human years).


Indeed, it is quite obvious that a little math around here wouldn't be such a
bad thing.


You might try it. Unless he started his career before age 11 (76-40-25), I
doubt his statement. Note that this assumes I used the very first circuit
simulators, which I highly doubt.
The only thing I ever did was tried out Verilog and I absolutely hated it..

It may have been greatly improved since then but it just didn't turn
me on.

Being that I also do C, Pascal, Asm etc, you would think that would
just fit in but it didn't.

On Tue, 17 Jan 2012 22:00:36 -0500, Jamie
jamie_ka1lpa_not_valid_after_ka1lpa__at_charter.net> wrote:


krw_at_att.bizzzzzzzzzzzz wrote:


On Wed, 18 Jan 2012 01:18:41 +0000 (UTC), Kaz Kylheku <kaz_at_kylheku.com> wrote:



On 2012-01-18, krw_at_att.bizzzzzzzzzzzz <krw_at_att.bizzzzzzzzzzzz> wrote:


On Tue, 17 Jan 2012 16:23:22 -0700, Jim Thompson


Those Youtube videos Bitrex found of IIT lectures are based on chips I
designed 50 years ago... all math... no simulator existed until about
25 years into my career. Also have some of my chips used as class
examples at Rochester Institute of Technology.

25 years? Circuit simulators existed more than 40 years ago.

You're assuming Jim's career launched 50 years ago, when upon walking into a
firm off the street without a clue, he got a job designing integrated circuits.
Then, 25 years later, simulators appeared.


Well, he's just about to turn 19 (76 in human years).



Indeed, it is quite obvious that a little math around here wouldn't be such a
bad thing.


You might try it. Unless he started his career before age 11 (76-40-25), I
doubt his statement. Note that this assumes I used the very first circuit
simulators, which I highly doubt.

The only thing I ever did was tried out Verilog and I absolutely hated it..


Huh? In English, please.


It may have been greatly improved since then but it just didn't turn
me on.


I don't like C, so have never bothered with Verilog. VHDL (and likewise,
PL/I) are much more my style.


Being that I also do C, Pascal, Asm etc, you would think that would
just fit in but it didn't.


I still don't understand the relevance.

krw_at_att.bizzzzzzzzzzzz wrote:

On Wed, 18 Jan 2012 01:18:41 +0000 (UTC), Kaz Kylheku <kaz_at_kylheku.com> wrote:


On 2012-01-18, krw_at_att.bizzzzzzzzzzzz <krw_at_att.bizzzzzzzzzzzz> wrote:

On Tue, 17 Jan 2012 16:23:22 -0700, Jim Thompson

Those Youtube videos Bitrex found of IIT lectures are based on chips I
designed 50 years ago... all math... no simulator existed until about
25 years into my career. Also have some of my chips used as class
examples at Rochester Institute of Technology.

25 years? Circuit simulators existed more than 40 years ago.

You're assuming Jim's career launched 50 years ago, when upon walking into a
firm off the street without a clue, he got a job designing integrated circuits.
Then, 25 years later, simulators appeared.


Well, he's just about to turn 19 (76 in human years).

Indeed, it is quite obvious that a little math around here wouldn't be such a
bad thing.


You might try it. Unless he started his career before age 11 (76-40-25), I
doubt his statement. Note that this assumes I used the very first circuit
simulators, which I highly doubt.
The only thing I ever did was tried out Verilog and I absolutely hated it..

It may have been greatly improved since then but it just didn't turn
me on.

Being that I also do C, Pascal, Asm etc, you would think that would
just fit in but it didn't.

Jamie

Jim Thompson wrote:

[snip]

Those Youtube videos Bitrex found of IIT lectures are based on chips I
designed 50 years ago... all math... no simulator existed until about
25 years into my career. Also have some of my chips used as class
examples at Rochester Institute of Technology.

Calculate that quiescent current and get back to me with the result.
(Show all your work

...Jim Thompson
I suppose I could, but I don't know where that would lead me to?

In any case, the LT part# I used in the sim was a low quiescent type
to start with. I am sure as simple as that circuit is, the sim can't be
that far off. I suppose if you select an opamp that has more than one
diode drop from the rails it could cause the outputs to have a high
Qu level. But why would you want to use a op-amp that bad? Infact I
can't think of one that has more than 1 diode drop on the rails at the
moment.

And Like I said, proven circuits that are still operating today tells
me alot..

Back in the old days my HP calculator, slide rule and lots of paper
was my friend. With Polaroid's of scope shots stuffed in scrap books.

Btw

It just came to me, this type of circuit is used to drive a optics
focus coil current mode amplifier on a 2 Mev irradiation unit. The lower
NPN is used to suppress the reactive energy when throttling back
instead of using a snubber or diode, this keeps it symmetrical when we
turn off the mag field to avoid unexpected beam steering. That uses a
large power
IC audio amplifier with the base of the transistors in the rails of the
IC with some by pass R to set the Qu I.

Jamie

krw_at_att.bizzzzzzzzzzzz wrote:

On Tue, 17 Jan 2012 22:00:36 -0500, Jamie
jamie_ka1lpa_not_valid_after_ka1lpa__at_charter.net> wrote:


krw_at_att.bizzzzzzzzzzzz wrote:


On Wed, 18 Jan 2012 01:18:41 +0000 (UTC), Kaz Kylheku <kaz_at_kylheku.com> wrote:



On 2012-01-18, krw_at_att.bizzzzzzzzzzzz <krw_at_att.bizzzzzzzzzzzz> wrote:


On Tue, 17 Jan 2012 16:23:22 -0700, Jim Thompson


Those Youtube videos Bitrex found of IIT lectures are based on chips I
designed 50 years ago... all math... no simulator existed until about
25 years into my career. Also have some of my chips used as class
examples at Rochester Institute of Technology.

25 years? Circuit simulators existed more than 40 years ago.

You're assuming Jim's career launched 50 years ago, when upon walking into a
firm off the street without a clue, he got a job designing integrated circuits.
Then, 25 years later, simulators appeared.


Well, he's just about to turn 19 (76 in human years).



Indeed, it is quite obvious that a little math around here wouldn't be such a
bad thing.


You might try it. Unless he started his career before age 11 (76-40-25), I
doubt his statement. Note that this assumes I used the very first circuit
simulators, which I highly doubt.

The only thing I ever did was tried out Verilog and I absolutely hated it..


Huh? In English, please.


It may have been greatly improved since then but it just didn't turn
me on.


I don't like C, so have never bothered with Verilog. VHDL (and likewise,
PL/I) are much more my style.


Being that I also do C, Pascal, Asm etc, you would think that would
just fit in but it didn't.


I still don't understand the relevance.

Ok, if you say so.

On Mon, 16 Jan 2012 18:11:03 -0800 (PST), George Herold
gherold_at_teachspin.com> wrote:

On Jan 16, 9:34 am, Jamie
jamie_ka1lpa_not_valid_after_ka1l...@charter.net> wrote:
Bill Bowden wrote:
On Jan 15, 11:40 am, Jamie

jamie_ka1lpa_not_valid_after_ka1l...@charter.net> wrote:

I don 't understand why you don't see it in the sim?

I see all kinds of problems there.

THe outputs are current modes and the beta on the outputs are
most
likely are not going to match well.

On top of that, LTspice shows the upper output (PNP) going into
discontinue state at the cross over. This is going to give you a
period
of what I call a flat liner and 99% sure this is where your cross
over
error is coming from.

Plot the current on R12.

Jamie

Yes, I did view the current through R12 which looks normal. The thing
operates class AB, so only one transistor is on at a time, so a 50%
discontinuous current is normal. I did improve the distortion using a
1458 op-amp in place of the 358. Looks much better now. The problem
now is I only get 1.5 volts peak into 8 ohms with an 8 volt supply and
I was trying for 3 volts or more. The HFE figure for the 2N2219A is
minimum 40 at 500mA or 7.5mA at 300mA. The 120 ohm resistor draws .
7/120 = about 6 mA so the op-amp must deliver 13.5 mA and the spec
sheet says only 10 mA short circuit. So, apparently, it needs higher
gain transistors or an op-amp with lower output impedance, or both.
Any ideas?

-Bill

That op-amp does not pull the reals, the 358 will do that effect on the
load side and there by give you more v to bias the transistors.

I don't think you have a current demand problem, you may have a rail
to rail problem how ever.

The 1458, as old as it is, still has a lot of usages. The las time I
looked, that op-amp (dual) only provides ~ Vcc-1.5 and Vee-1.5. Here you
have lost 3 volts to start with. This now gives you 5 volts to play with.

Of course, you really don't want to saturate the amp, so lets assume
you have only 4.5V to work with..

split that in half, since you looks apparrent you are operating in
Class A state on the output side of the op-amp and you get ~ 2.25 volts
Peak to play with.

Now., let us not forget, the minimum required for each of those
transistors to start working. ~ 0.7 and then times this by 2 and you get
1.4. Remove that value off the top and you are now getting closer to
where the problem is.

That configuration you're using in the first place is fighting against
you. As one side is conducting the other side is still conducting, just
about all the way through. This is going to remove a good chunk of your
output.

Have you considered a config like the following or something in this line?

8Volts-----------------------------+
+ |
|
|
+
|
+-----------------+|
___ | |\
+--+|___|-+----------------++ |
| | | +
+ | | |
|| | |\+ | | 1Ku
-||+--++-------------+|-\ | | ||
|| | >+-------+----------+---+||+-----+
+-+-|+/ | || |
4Volts |/+ | +
| | |
| | .-.
| | | |
+ + | |8
| |/ '-'
+-----------------+| |
|> GND
+
|
|
===
GND

Jamie- Hide quoted text -

- Show quoted text -

That's weird looking too! There's a JL circuit with tran's in the
power leads,
but I think it's push-pull. You can get a lot closer to the supply
rail that way...
At least that's my understanding. I've yet to try a circuit with
power gain in the supply leads.

George H.

I did that sort of thing as an LDO, ~1970... it's in a hybrid I made
for the TOW missile. You need base-emitter resistors such that the
"crowbar" current is zero. Disconnect OpAmp output from as shown,
then connect it via a resistive divider between rails (it'd be a
single resistor to ground if split supplies were used. It's then a
CLASS-B amplifier.