Maker Pro
Maker Pro

LM386 (audio amp) oscillates

B

Bill Bowden

Jan 1, 1970
0
I have a LM386 that oscillates when the volume is turned up too far.
It's operating with the default gain of 20 and there is a RC on the
output as recommended in the app notes (10 ohms in series with .05uF).
The oscillations will stop if I add a .05 cap across the input pot,
but that reduces the signal and rolls off the high end. There is a
1000uF cap across the power terminals

Seems to be set up right, but makes too much noise when the volume is
cranked up.

Any ideas?

-Bill
 
J

John Popelish

Jan 1, 1970
0
Bill said:
I have a LM386 that oscillates when the volume is turned up too far.
It's operating with the default gain of 20 and there is a RC on the
output as recommended in the app notes (10 ohms in series with .05uF).
The oscillations will stop if I add a .05 cap across the input pot,
but that reduces the signal and rolls off the high end. There is a
1000uF cap across the power terminals

Seems to be set up right, but makes too much noise when the volume is
cranked up.

Any ideas?

Have you got a capacitor between ground and the bypass pin 7?

Have you got a bypass capacitor between ground and the
supply, pins 4 to 6?

Does the speaker ground end tie back to the ground pin, or
to somewhere else on a ground bus?

You might try replacing the .05 uF output filter capacitor
with a pair of .022 uF or .027 uF or .o33 uF caps in series
across the supply pins of the chip, with the 10 ohm resistor
connected to their common node.
 
E

Eeyore

Jan 1, 1970
0
Bill said:
I have a LM386 that oscillates when the volume is turned up too far.

This seems to be the most common complaint on the sci.electronics groups.

Why are you using that POS ? It was dreadful when it was introduced.

Graham
 
A

Anthony Fremont

Jan 1, 1970
0
Eeyore said:
This seems to be the most common complaint on the sci.electronics
groups.

Why are you using that POS ? It was dreadful when it was introduced.

I have a cheapy radio shack amplifed speaker that uses one. It has a huge
amount of gain and never motor-boats or oscillates. It has enough gain that
you can attach it directly to the output of a NE602 mixer and hear audio.
That's allot of gain. :)
 
E

Eeyore

Jan 1, 1970
0

I've never needed to nor do I wish to. It's rubbish and I have no use for it. My
preference going back to days of old was TI's SN76023. You could make one of
those perform surprisingly well !

I'm sure it would present no problems to an expert in audio amplifiers like
myself though. After all, my 2kW amplifiers work beautifully and have
class-leading performance figures and need I add, totally unconditional
stability.

Graham
 
B

Bill Bowden

Jan 1, 1970
0
Have you got a capacitor between ground and the bypass pin 7?

No. The app notes show pin 7 open in the 26dB setup.But it looks like
a good idea to use a 10uF- 50uF anyway. I will add a cap and report
results.
Have you got a bypass capacitor between ground and the
supply, pins 4 to 6?
Yes, 1000uF
Does the speaker ground end tie back to the ground pin, or
to somewhere else on a ground bus?

It's close to the chip, and one side of the board is all ground plane,
so it's pretty much zero ohms wherever you go.
You might try replacing the .05 uF output filter capacitor
with a pair of .022 uF or .027 uF or .o33 uF caps in series
across the supply pins of the chip, with the 10 ohm resistor
connected to their common node.

-Bill
 
J

Jamie

Jan 1, 1970
0
Bill said:
I have a LM386 that oscillates when the volume is turned up too far.
It's operating with the default gain of 20 and there is a RC on the
output as recommended in the app notes (10 ohms in series with .05uF).
The oscillations will stop if I add a .05 cap across the input pot,
but that reduces the signal and rolls off the high end. There is a
1000uF cap across the power terminals

Seems to be set up right, but makes too much noise when the volume is
cranked up.

Any ideas?

-Bill
try putting a non inductive cap across the rails as close to the
leads as possible., you can still keep the 1000 there.

.1 uf ceramic might do it.
 
J

John Popelish

Jan 1, 1970
0
Jamie said:
try putting a non inductive cap across the rails as close to the
leads as possible., you can still keep the 1000 there.

.1 uf ceramic might do it.

I really like the Panasonic V series stacked mylar film
capacitors for low ESR hefty bypass use. The 1 uF at 50
volts is about a quarter inch cubed. But if space is very
limited, a 10 uF 25 volt 1206 X5R SMT capacitor will do a
nice job with something like a milliohm of ESR.
 
J

Jamie

Jan 1, 1970
0
John said:
I really like the Panasonic V series stacked mylar film capacitors for
low ESR hefty bypass use. The 1 uF at 50 volts is about a quarter inch
cubed. But if space is very limited, a 10 uF 25 volt 1206 X5R SMT
capacitor will do a nice job with something like a milliohm of ESR.
I have a few mica smt caps that work very nicely, they are expensive
how ever because they are hard to make..
When I was working at Semco as an consultant, I was part of a project
where we automated a robotics arm that had visual to the computer.

The arm used a mini vacuums finger to pick up the mica sheets to be
stacked.
This process would pick a cut piece from a pile, visually inspect it
for defeats first, put it under a pressure plate where High voltage was
applied for leakage test. If all of that passed, it was then pick up
and place in a contact frame where the cap was being assembled.
Very thin silver foil was laid between for the plate's from a mini
spool entering from the side's of the frame..

When the stack was complete, a push finger comes down along with a
button lift finger that pushes it out of the contact frame and is
compressed between these 2 fingers as it is quickly hit with a HOT
iron (very small) to bond the foils..
from there another HV test is done along with capacitance test which
will very the compression tension of the fingers to make adjustments on
the fly.
the next process then passes this over to two other holding fingers
where the encapsoluation is applied and harden with UV (Very Strong) UV.
then another test is done, then they are place in little strip frames
for later encaps and printing ..

All of that above it done using robotics arm with attached video camera's
into a PC using DELPHI has the graphics and guidance for the movements.
Things were color coded on the table to make it easy to track.
this may sound like a lot of steps how ever, the process actually was
very fast once you got it going. the operators only needed to dump small
piles of pre-cut mica on a pick up area. From there, the robotics did the
rest.
Before that time, the Japanese was the only one's making then and
their yield was like 5 out of 10 good, and that was done by hand as for
as the stacking.
We also suffered the same problem, but after the automation, it was
like 99.99% good other than an unexpected machine failure/Jam up.
 
B

Bill Bowden

Jan 1, 1970
0
No. The app notes show pin 7 open in the 26dB setup.But it looks like
a good idea to use a 10uF- 50uF anyway. I will add a cap and report
results.

Well, I had high hopes for the bypass cap on pin 7, but unfortunately
it didn't help. I tried various caps (0.4uF, 1uF,22uF) from pin 7 to
ground with no improvement. Also tried small caps across the chip
power pins with no change.

-Bill
 
E

Eeyore

Jan 1, 1970
0
Bill said:
Well, I had high hopes for the bypass cap on pin 7, but unfortunately
it didn't help. I tried various caps (0.4uF, 1uF,22uF) from pin 7 to
ground with no improvement.

What dielectrics ?

Also tried small caps

How small ?

across the chip power pins with no change.

What frequency is your oscillation ?

What value are all your components ? Is your circuit DS006976-3 from the
datasheet with pins 1,7 and 8 open ? Have you tweaked any values ?

And indeed what value is your input pot ?

I'm suspecting you're trying to get to high an input impedance and you have some
capacitive coupling from the output to the input.

What's the output Z of the signal source too ?

What's the supply voltage and load impedance ?

Are you absolutely stuck with using this chip ?


Graham
 
C

christopher

Jan 1, 1970
0
I have a LM386 that oscillates when the volume is turned up too far.
It's operating with the default gain of 20 and there is a RC on the
output as recommended in the app notes (10 ohms in series with .05uF).
The oscillations will stop if I add a .05 cap across the input pot,
but that reduces the signal and rolls off the high end. There is a
1000uF cap across the power terminals

Seems to be set up right, but makes too much noise when the volume is
cranked up.

Any ideas?

*************

"One mans ringing may be another mans oscillation."

I solved my own ringing problem on the output of the LM386 using a
ferrite bead. Why do I use this chip, I can walk into Radio Shack and
find it for quick projects.


View this ferrite on page 6 of this PDF. Many good pointers on the
page.

http://downloads.solarbotics.com/PDF/LM386.pdf

good luck,

* * *
Christopher

Temecula CA.USA
http://www.oldtemecula.com
 
B

Bill Bowden

Jan 1, 1970
0
What dielectrics ?


How small ?


What frequency is your oscillation ?

What value are all your components ? Is your circuit DS006976-3 from the
datasheet with pins 1,7 and 8 open ? Have you tweaked any values ?

And indeed what value is your input pot ?

I'm suspecting you're trying to get to high an input impedance and you have some
capacitive coupling from the output to the input.

What's the output Z of the signal source too ?

What's the supply voltage and load impedance ?

Are you absolutely stuck with using this chip ?

Graham-

There is diode/fet detector circuit that drives a 10K pot that feeds
the 386. The problem is not really oscillations but a loud roaring
noise that seems to be related to the diode. I'm going to add a 100K
in series with the 2 meg and a capacitor from the junction to ground.
That should reduce feedback to the gate of the jfet. Something like
this without the extra restistor or cap shown


+6
|
| +6
[2 Meg] |
| |
| |---+
[diode]-+--g>| jfet
| |---+
| |------||-----+
[270 pF] | |
| [R] [10k pot]----> LM386
| | |
| | |
GND GND GND


-Bill
 
Top