Maker Pro
Maker Pro

LM358 not for audio

P

Phil Allison

Jan 1, 1970
0
** The LM358 is a poor choice for most audio circuits - particularly if you
want low distortion.

Why?

The LM358 and its quad version the LM324 are intended for single supply rail
operation and to use very little current. To save current, the IC has an
unusual output stage compared to most other op-amps. In audio circuits,
there is a BIG problem is with this oddball output circuit - see schem on
page 21 of this pdf:

http://www.ti.com/lit/ds/symlink/lm158-n.pdf

Qs 5, 6 and 13 form a complimentary, emitter follower, current booster
supplying the output current - BUT the bases of Q5 and Q13 are linked so
there is no forward bias on the stage.

Soooo - every time current flow into a load has to change between Q6 and
Q13 - the collector of Q12 has to *jump* a nearly 2 volt gap created by the
three B-E junctions involved. During the time this takes to happen, NO
output current flows. This results in "crossover distortion" at the output
which becomes worse and worse as the frequency rises.

On page 8 of the pdf, the makers allude to this fact:

" For AC applications, where the load is capacitively coupled to the output
of the amplifier, a resistor should be used, from the output of the
amplifier to ground to increase the class A bias current and prevent
crossover distortion. Where the load is directly coupled, as in DC
applications, there is no crossover distortion."

This is hardly an adequate explanation of the issue.

In the AC coupled case, one has to fit a resistor to ground of such a value
that Q13 is never required to conduct current - which could mean a very low
value resistor has to be used and is simply not practical.

In the DC case, the wording is quite misleading. Only if the load is
connected directly to ground and the negative supply is also the ground is
the comment true.

In the general case where the load is connected to a ground point that sits
between the supply rails - then Q13 has to supply current and serious
crossover distortion is the result.

BTW:

A low current, dual op-amp that IS suitable for audio is the TL062 and its
quad version the TL064. Supply current is less than 200uA per amplifier and
operation right down to +/- 2 volts is OK.

And definitely NO crossover distortion !!

http://www.ti.com/lit/ds/symlink/tl062.pdf



.... Phil
 
M

Michael Black

Jan 1, 1970
0
Interesting. Thanks for posting that. Looks like it's hard to overcome
the 2 volt step needed between Q5 and Q13 to crossover.
But there was a solution to some of this, I think it was a resistor at the
output to ground, causing the output stage to act differently.

Of course, if the device is being used for the low current aspect, that
loading isn't likely to be good for current consumption.

Michael
 
J

Jamie

Jan 1, 1970
0
Bill said:
Interesting. Thanks for posting that. Looks like it's hard to overcome
the 2 volt step needed between Q5 and Q13 to crossover.

-Bill
if you are using such an animal that allows 0 volts on the low side like
a 358/324 etc, try placing a load R on the output, this will force the
op-amp to attempt to maintain symmetrical output, at least that is what
i've done.

Last time I looked at the internals of that, I seem to remember the
output will go into a HI-Z state at the 0 volt level or some where near
that, which gives it an advantage in single rail applications but causes
issues else where.

Jamie
 
J

Jamie

Jan 1, 1970
0
Bill said:
Here's another audio amp example using a LM358. Looks like it's biased
to operate in the lower 50% region so the output signal never crosses
the mid point, and therefore no distortion. Am I reading that right?

http://www.techlib.com/electronics/graphics/audioa14.gif

-Bill
That is a generic complementary pair output that can suffer from a
few things.

As for the 358 not having the low side issue there, that is correct,
however, that circuit will have problems if you are expecting any amount
of output.. In your case, it may be just fine..

You will expect to lose off the top, ~ 1.4 volts. This is due to the
forward drop in the emitter following state of each rail or ~.7 volts.


On top of all this, the 2 diodes you have in the bias circuit for the
quiescent current is also going to remove a little from the output.

All in all, you'd be lucky to see a 3.0 PK /6.0 Pk-PK output that is
clean.


I noticed you are basically constructing a differential type of input ?
If so, I hope you were not looking for low phase errors between the inputs?

Jamie
 
Top