LM386 power amp IC question

[Kaz Kylheku]

I found an old LM386 chip in one of my piles of junk, and I'm thinking
of putting it into use.

All the schematics in the data sheet use an idiotic, large output
coupling capacitor between the chip and the speaker, which just
makes me cringe.

I'm thinking that since the IC references the output at halfway between
its power rails, and since I'm using a virtual reference in the rest of
my circuit at V/2, can't I just tie the speaker's return to the virtual
ground?

There is a chance that the two voltages may differ, causing some current
to flow at quiescence, and any such current should be kept well below
what the device can deliver.

Any other gotchas with this IC?

Thanks.

 

[Phil Allison]

"Kaz Kylheku"

I found an old LM386 chip in one of my piles of junk, and I'm thinking
of putting it into use.

All the schematics in the data sheet use an idiotic, large output
coupling capacitor between the chip and the speaker, which just
makes me cringe.

** No reason to cringe exists.

You must be an audiophool.

I'm thinking that since the IC references the output at halfway between
its power rails, and since I'm using a virtual reference in the rest of
my circuit at V/2, can't I just tie the speaker's return to the virtual
ground?

** ROTFL.

Jus use the electro cap.


..... Phil

 

[Jasen Betts]

On 2011-08-30, Kaz Kylheku <kaz@kylheku.com> wrote:

I found an old LM386 chip in one of my piles of junk, and I'm thinking
of putting it into use.

All the schematics in the data sheet use an idiotic, large output
coupling capacitor between the chip and the speaker, which just
makes me cringe.

I'm thinking that since the IC references the output at halfway between
its power rails, and since I'm using a virtual reference in the rest of
my circuit at V/2, can't I just tie the speaker's return to the virtual
ground?

what's going to happen if you feed the LM386 output current into the
"virtual ground"? a virtual ground is usually an (inverting) input
termial, not an output terminal.

you might be able to work it using a center tapped battery supply.

--
⚂⚃ 100% natural

--- Posted via news://freenews.netfront.net/ - Complaints to news@netfront.net ---

 

[Michael Black]

On Tue, 30 Aug 2011, Kaz Kylheku wrote:

I found an old LM386 chip in one of my piles of junk, and I'm thinking
of putting it into use.

All the schematics in the data sheet use an idiotic, large output
coupling capacitor between the chip and the speaker, which just
makes me cringe.

I'm thinking that since the IC references the output at halfway between
its power rails, and since I'm using a virtual reference in the rest of
my circuit at V/2, can't I just tie the speaker's return to the virtual
ground?

There is a chance that the two voltages may differ, causing some current
to flow at quiescence, and any such current should be kept well below
what the device can deliver.

Any other gotchas with this IC?

Thanks.

Dig further and find a second one. Then feed one 180degrees out of phase

with the first one, and connect your speaker across the outputs of the two
amplifiers. No more output capacitor, more power, and no Rube Goldberg.

Of course, it was never much of a "power amplifier", just enough for a lot
of things that needed a bit of power to a speaker. Nowadays, one can find
all kinds of more recent amplifier ICs in all kinds of devices, some
likely do have the out of phase second amplifier built in so there's no
coupling capacitor on the output.

One might find these on modem boards or even soundcards, though those may
still be 386s. "Computer speakers", I just brought home two sets from the
garbage this week, likely have them. Car radios will. Anything with a
speaker is a suspect for donating a more recent audio amplifier IC.

Michael

 

[Tim Wescott]

On Tue, 30 Aug 2011 02:14:49 +0000, Kaz Kylheku wrote:

I found an old LM386 chip in one of my piles of junk, and I'm thinking
of putting it into use.

All the schematics in the data sheet use an idiotic, large output
coupling capacitor between the chip and the speaker, which just makes me
cringe.

I'm thinking that since the IC references the output at halfway between
its power rails, and since I'm using a virtual reference in the rest of
my circuit at V/2, can't I just tie the speaker's return to the virtual
ground?

There is a chance that the two voltages may differ, causing some current
to flow at quiescence, and any such current should be kept well below
what the device can deliver.

Any other gotchas with this IC?

It's a crappy chip, but it'll make a noise on a speaker.

It used to be popular with the amateur radio crowd, so I bought a bunch
before I knew any better.

That "1/2 VCC" is going to be very approximate -- use the cap.

I've noticed that unless it is very well bypassed it'll tend to have a
high-frequency oscillation (several hundred kHz or a few MHz, I can't
remember which) on the audio. Some portion of the amplifier is obviously
saturating, because it makes the audio distort.

--
www.wescottdesign.com

 

On Tue, 30 Aug 2011 02:14:49 +0000 (UTC), Kaz Kylheku
<kaz@kylheku.com> wrote:

I found an old LM386 chip in one of my piles of junk, and I'm thinking
of putting it into use.

All the schematics in the data sheet use an idiotic, large output
coupling capacitor between the chip and the speaker, which just
makes me cringe.

I'm thinking that since the IC references the output at halfway between
its power rails, and since I'm using a virtual reference in the rest of
my circuit at V/2, can't I just tie the speaker's return to the virtual
ground?

There is a chance that the two voltages may differ, causing some current
to flow at quiescence, and any such current should be kept well below
what the device can deliver.

Any other gotchas with this IC?

Thanks.

I don't see why it wouldn't work - providing the virtual reference is
"stiff" enough to handle it. Presumably this reference is a power op
amp output, voltage follower, referenced to 1/2 Vcc? Stiff to the
tune of at least a 1/2 amp or so?

But there's no good reason not to use a coupling cap - all transistor
stereos did that at one time (60's and 70's).

I've used both varieties of integrated stereo amp (bipolar and single
supply) as their opposite number at one time or another.

 

[Rich Grise]

Michael Black wrote:

Dig further and find a second one. Then feed one 180degrees out of phase
with the first one,

No, don't phase-shift your signal. Just feed the inversion of the signal
to the other input. There should be no appreciable phase shift, just an
inversion of polarity.

With a sinusoid, a 180 degree phase shift may _look_ exactly the same
as a polarity inversion, but they are by no means the same. To illustrate
this, just use an asymmetrical signal on your input, like a pulse train.

Hope This Helps!
Rich

 

[Rich Grise]

Kaz Kylheku wrote:

I found an old LM386 chip in one of my piles of junk, and I'm thinking
of putting it into use.

All the schematics in the data sheet use an idiotic, large output
coupling capacitor between the chip and the speaker, which just
makes me cringe.

I'm thinking that since the IC references the output at halfway between
its power rails, and since I'm using a virtual reference in the rest of
my circuit at V/2, can't I just tie the speaker's return to the virtual
ground?

Just use two of them in a bridge amp.

Have Fun!
Rich

 

On Tue, 30 Aug 2011 02:14:49 +0000 (UTC), Kaz Kylheku
<kaz@kylheku.com> wrote:

Ruminating on this theme: A couple of equal resistors from Vcc to
ground with large electrolytics bypassing them should also make a
stiff virtual reference.

Stupid way to do things, and not net gain, but it should "work."

Familiar with Nelson Pass's super simple, single mosfet amps? He
calls it the Zen Amplifier. He uses a cap to keep DC off the speaker,
but several people have taken it further to use it push pull (bridged
amplifiers) to eliminate the cap or power voltage followers to
eliminate the cap.

But a LM386 is not high fidelity, so there's no point in killing
yourself to get rid of the cap.

 

[Michael Black]

On Tue, 30 Aug 2011, default wrote:

On Tue, 30 Aug 2011 02:14:49 +0000 (UTC), Kaz Kylheku
kaz@kylheku.com> wrote:

I found an old LM386 chip in one of my piles of junk, and I'm thinking
of putting it into use.

All the schematics in the data sheet use an idiotic, large output
coupling capacitor between the chip and the speaker, which just
makes me cringe.

I'm thinking that since the IC references the output at halfway between
its power rails, and since I'm using a virtual reference in the rest of
my circuit at V/2, can't I just tie the speaker's return to the virtual
ground?

There is a chance that the two voltages may differ, causing some current
to flow at quiescence, and any such current should be kept well below
what the device can deliver.

Any other gotchas with this IC?

Thanks.

I don't see why it wouldn't work - providing the virtual reference is
"stiff" enough to handle it. Presumably this reference is a power op
amp output, voltage follower, referenced to 1/2 Vcc? Stiff to the
tune of at least a 1/2 amp or so?

You mean like a 386 just sitting there in a static form, properly bypassed

for audio?

Michael

But there's no good reason not to use a coupling cap - all transistor
stereos did that at one time (60's and 70's).

I've used both varieties of integrated stereo amp (bipolar and single
supply) as their opposite number at one time or another.

 

[krw@att.bizzzzzzzzzzzz]

On Tue, 30 Aug 2011 11:08:08 -0500, Tim Wescott <tim@seemywebsite.com> wrote:

On Tue, 30 Aug 2011 02:14:49 +0000, Kaz Kylheku wrote:

I found an old LM386 chip in one of my piles of junk, and I'm thinking
of putting it into use.

All the schematics in the data sheet use an idiotic, large output
coupling capacitor between the chip and the speaker, which just makes me
cringe.

I'm thinking that since the IC references the output at halfway between
its power rails, and since I'm using a virtual reference in the rest of
my circuit at V/2, can't I just tie the speaker's return to the virtual
ground?

There is a chance that the two voltages may differ, causing some current
to flow at quiescence, and any such current should be kept well below
what the device can deliver.

Any other gotchas with this IC?

It's a crappy chip, but it'll make a noise on a speaker.

It used to be popular with the amateur radio crowd, so I bought a bunch
before I knew any better.

That "1/2 VCC" is going to be very approximate -- use the cap.

I've noticed that unless it is very well bypassed it'll tend to have a
high-frequency oscillation (several hundred kHz or a few MHz, I can't
remember which) on the audio. Some portion of the amplifier is obviously
saturating, because it makes the audio distort.

Put a small cap across the feedback resistor. Limit the amplifier's bandwidth
to 20kHz (2x what you care about), or so.

 

[David Eather]

On 30/08/2011 12:14 PM, Kaz Kylheku wrote:

I found an old LM386 chip in one of my piles of junk, and I'm thinking
of putting it into use.

All the schematics in the data sheet use an idiotic, large output
coupling capacitor between the chip and the speaker, which just
makes me cringe.

I'm thinking that since the IC references the output at halfway between
its power rails, and since I'm using a virtual reference in the rest of
my circuit at V/2, can't I just tie the speaker's return to the virtual
ground?

There is a chance that the two voltages may differ, causing some current
to flow at quiescence, and any such current should be kept well below
what the device can deliver.

Any other gotchas with this IC?

Thanks.

The LM386 was/is a great chip for what it was designed for which was low
cost audio at *modest* output levels. It is not HiFi. Using the cap will
not degrade the sound in anyway if the cap is of suitable size. The cap
will in fact provide some protection to the speaker if things go pear
shaped.

 

On Tue, 30 Aug 2011 15:00:05 -0400, Michael Black <et472@ncf.ca>
wrote:

On Tue, 30 Aug 2011, default wrote:

On Tue, 30 Aug 2011 02:14:49 +0000 (UTC), Kaz Kylheku
kaz@kylheku.com> wrote:

I found an old LM386 chip in one of my piles of junk, and I'm thinking
of putting it into use.

All the schematics in the data sheet use an idiotic, large output
coupling capacitor between the chip and the speaker, which just
makes me cringe.

I'm thinking that since the IC references the output at halfway between
its power rails, and since I'm using a virtual reference in the rest of
my circuit at V/2, can't I just tie the speaker's return to the virtual
ground?

There is a chance that the two voltages may differ, causing some current
to flow at quiescence, and any such current should be kept well below
what the device can deliver.

Any other gotchas with this IC?

Thanks.

I don't see why it wouldn't work - providing the virtual reference is
"stiff" enough to handle it. Presumably this reference is a power op
amp output, voltage follower, referenced to 1/2 Vcc? Stiff to the
tune of at least a 1/2 amp or so?

You mean like a 386 just sitting there in a static form, properly bypassed
for audio?

Michael

I thought we are talking about coupling to the speaker - the purpose
of the large cap in series with the speaker lead.

Virtual references do work - you use a voltage follower instead of
just a divider because it has the ability to respond to a change in
current without changing the voltage. But in something like an audio
speaker signal it has to be able to handle the few milliamps the 386
can throw at it. Most op amps have low current outputs and would
probably not work well.

Your bridged amplifier suggestion is equally valid - lots of car audio
equipment uses that as a way to drive more power into a load from the
limited 12 volts they have to work with.

But there's no good reason not to use a coupling cap - all transistor
stereos did that at one time (60's and 70's).

I've used both varieties of integrated stereo amp (bipolar and single
supply) as their opposite number at one time or another.

 

[krw@att.bizzzzzzzzzzzz]

On Wed, 31 Aug 2011 18:20:50 +0000 (UTC), Kaz Kylheku <kaz@kylheku.com> wrote:

On 2011-08-30, Michael Black <et472@ncf.ca> wrote:
On Tue, 30 Aug 2011, Kaz Kylheku wrote:

I found an old LM386 chip in one of my piles of junk, and I'm thinking
of putting it into use.

All the schematics in the data sheet use an idiotic, large output
coupling capacitor between the chip and the speaker, which just
makes me cringe.

I'm thinking that since the IC references the output at halfway between
its power rails, and since I'm using a virtual reference in the rest of
my circuit at V/2, can't I just tie the speaker's return to the virtual
ground?

There is a chance that the two voltages may differ, causing some current
to flow at quiescence, and any such current should be kept well below
what the device can deliver.

Any other gotchas with this IC?

Thanks.

Dig further and find a second one. Then feed one 180degrees out of phase
with the first one, and connect your speaker across the outputs of the two
amplifiers.

Ah yes. Some audio power IC's do this bridging trick internally. Thanks
for the suggestion.

The ones I've used simply have an inverting stage driving the other leg.
Nothing fancy at all.

Actually, I already have a capacitance there: the two capacitors (with
balancing resistors) forming a divider which establish the reference
voltage.

Be careful of any noise on this reference. We've generally used an active
filter to drive the reference.

So if the speaker is based off the reference voltage, there is
AC coupling to ground already.

But your driving impedance into the reference is very low.

I just bumped these caps up to 100uF each, which makes 200uF (since we
can consider them to be parallel, looking from the p.o.v. of AC at the
middle reference.)

What's your filter corner?

This is good: use the caps we already have, avoding the introduction
of redundant ones.

Also it occurs to me that if I had a second LM386, I could use it
simply to generate the middle reference voltage.

Yes, but you'd be better off driving the load differentially.

This is similar to the out-of-phase trick except that the voltage
of the second amp does not move opposite to the signal. The advantage
is that this reference voltage can be used by the entire circuit,
allowing the capacitor divider to be removed.

You still have a pretty low impedance driving that reference. That can cause
all sorts of grief up the line. You have feedback paths all over the place.

 

[Kaz Kylheku]

On 2011-08-30, Michael Black <et472@ncf.ca> wrote:

On Tue, 30 Aug 2011, Kaz Kylheku wrote:

I found an old LM386 chip in one of my piles of junk, and I'm thinking
of putting it into use.

All the schematics in the data sheet use an idiotic, large output
coupling capacitor between the chip and the speaker, which just
makes me cringe.

I'm thinking that since the IC references the output at halfway between
its power rails, and since I'm using a virtual reference in the rest of
my circuit at V/2, can't I just tie the speaker's return to the virtual
ground?

There is a chance that the two voltages may differ, causing some current
to flow at quiescence, and any such current should be kept well below
what the device can deliver.

Any other gotchas with this IC?

Thanks.

Dig further and find a second one. Then feed one 180degrees out of phase
with the first one, and connect your speaker across the outputs of the two
amplifiers.

Ah yes. Some audio power IC's do this bridging trick internally. Thanks
for the suggestion.

Actually, I already have a capacitance there: the two capacitors (with
balancing resistors) forming a divider which establish the reference
voltage.

So if the speaker is based off the reference voltage, there is
AC coupling to ground already.

I just bumped these caps up to 100uF each, which makes 200uF (since we
can consider them to be parallel, looking from the p.o.v. of AC at the
middle reference.)

This is good: use the caps we already have, avoding the introduction
of redundant ones.

Also it occurs to me that if I had a second LM386, I could use it
simply to generate the middle reference voltage.

This is similar to the out-of-phase trick except that the voltage
of the second amp does not move opposite to the signal. The advantage
is that this reference voltage can be used by the entire circuit,
allowing the capacitor divider to be removed.

 

[Kaz Kylheku]

On 2011-08-30, Tim Wescott <tim@seemywebsite.com> wrote:

On Tue, 30 Aug 2011 02:14:49 +0000, Kaz Kylheku wrote:

I found an old LM386 chip in one of my piles of junk, and I'm thinking
of putting it into use.

All the schematics in the data sheet use an idiotic, large output
coupling capacitor between the chip and the speaker, which just makes me
cringe.

I'm thinking that since the IC references the output at halfway between
its power rails, and since I'm using a virtual reference in the rest of
my circuit at V/2, can't I just tie the speaker's return to the virtual
ground?

There is a chance that the two voltages may differ, causing some current
to flow at quiescence, and any such current should be kept well below
what the device can deliver.

Any other gotchas with this IC?

It's a crappy chip, but it'll make a noise on a speaker.

Yes it is, but this is not a hi-fi application. I've looked at lots of
datasheets for possible alternatives, such as the RC4580.

Next project, maybe.

It used to be popular with the amateur radio crowd, so I bought a bunch
before I knew any better.

That "1/2 VCC" is going to be very approximate -- use the cap.

Right. Since the 1/2 VCC has caps to either power rail, that is
more or less taken care of.

I've noticed that unless it is very well bypassed it'll tend to have a
high-frequency oscillation (several hundred kHz or a few MHz, I can't
remember which) on the audio.

I will watch for that. I do have the 0.05 uF cap + 10 ohm resistor at the
output to pull high frequencies down, right from the datasheet example.

 

[Kaz Kylheku]

On 2011-08-30, Rich Grise <richg@example.net.invalid> wrote:

Michael Black wrote:

Dig further and find a second one. Then feed one 180degrees out of phase
with the first one,

No, don't phase-shift your signal. Just feed the inversion of the signal
to the other input. There should be no appreciable phase shift, just an
inversion of polarity.

Of course what Michael is talking about is a polarity reversal.

This is often called "180 degrees out of phase" by electronic
hobbyists. It's a kind of jargon.

I know, it makes me cringe too, but as long as everyone understands
that what is being discussed is a polarity reversal, it isn't an issue.

 

[Kaz Kylheku]

On 2011-08-30, David Eather <eather@tpg.com.au> wrote:

The LM386 was/is a great chip for what it was designed for which was low
cost audio at *modest* output levels. It is not HiFi. Using the cap will
not degrade the sound in anyway if the cap is of suitable size. The cap
will in fact provide some protection to the speaker if things go pear
shaped.

I think that this protection is taken care of by the voltage divider
with bypass caps.

The divider has enough resistance such that any DC flowing through the
speaker to the V/2 reference is going to be quite limited.

(Of course there could be a fault in that V/2 network somewhere,
and the speaker is exposed to that.)

Ah well, I have some airline headphones here that can be sacrificed for
a smoke test, and before that, I can measure DC current at the output
terminals.

 

[Phil Allison]

"Kaz Kylheku"

Actually, I already have a capacitance there: the two capacitors (with
balancing resistors) forming a divider which establish the reference
voltage.

So if the speaker is based off the reference voltage, there is
AC coupling to ground already.

I just bumped these caps up to 100uF each, which makes 200uF (since we
can consider them to be parallel, looking from the p.o.v. of AC at the
middle reference.)

This is good: use the caps we already have, avoding the introduction
of redundant ones.

** Our smug friend here is about to discover what the term "motor boating "
refers to .


..... Phil

 

[David Eather]

On 1/09/2011 4:39 AM, Kaz Kylheku wrote:

On 2011-08-30, David Eather<eather@tpg.com.au> wrote:
The LM386 was/is a great chip for what it was designed for which was low
cost audio at *modest* output levels. It is not HiFi. Using the cap will
not degrade the sound in anyway if the cap is of suitable size. The cap
will in fact provide some protection to the speaker if things go pear
shaped.

I think that this protection is taken care of by the voltage divider
with bypass caps.

The divider has enough resistance such that any DC flowing through the
speaker to the V/2 reference is going to be quite limited.

(Of course there could be a fault in that V/2 network somewhere,
and the speaker is exposed to that.)

Ah well, I have some airline headphones here that can be sacrificed for
a smoke test, and before that, I can measure DC current at the output
terminals.

Consider:

Your weird biasing arrangement capacitively couples to the speaker - so
you do not get the benefit you wanted by avoiding the standard circuit
configuration - so on that front it is an outright failure.

In addition to this you introduce a whole bunch of problematic biasing
arrangements which solve *no* problems and introduce others - so that is
a outright failure.

Your new arrangement needs more board space and more components to work
and at *best* it will be only as good as the standard configuration - so
that is a bust too.

Just use the standard configuration and use the extra power supply cap
from the other configuration as the speaker coupling cap.