LM386 softer clipping

[Steben]

Hey Guys, would adding a diode pair and resistor in series between pin 1 and 5 on a 386 chip make for softer onset of clipping?

[antonis]

By maintainig gain cap between pin 1 and 8 or not..??

[Rob Strand]

You will need a series cap like a Big Muff because the DC pins 1 and 5 is different (and non zero).

Without the cap you would need an asymmetrical clipper to "take up" the DC offset.  Something like a zener might seem like it will work but it's likely to be conducting to some degree and act a little differently than expected because zener needs current to create the voltage drop.  So taking this route is likely to need more work.  For example a zener to the output then a resistor to ground to shift the voltage then that feeds the two diodes.  Since the diode clipper is only +/- 0.7V the DC voltage at the shifted point is going to be very sensitive to circuit conditions.   So instead of zener use a resistive divider to set the DC voltage at the divider output to be equal to the DC voltage on pin 5.    The only reason I mentioned the crazier versions is it might sound OK  .

[Vivek]

Which stage(s) clip inside the LM386 ?

In which order do they clip ?

Do we lose the LM386ness of the Opamp if we add external diodes (ie it becomes equivalent to any other Opamp if external diodes control the clipping)

[Steben]

You will need a series cap like a Big Muff because the DC pins 1 and 5 is different (and non zero).

Without the cap you would need an asymmetrical clipper to "take up" the DC offset.  Something like a zener might seem like it will work but it's likely to be conducting to some degree and act a little differently than expected because zener needs current to create the voltage drop.  So taking this route is likely to need more work.  For example a zener to the output then a resistor to ground to shift the voltage then that feeds the two diodes.  Since the diode clipper is only +/- 0.7V the DC voltage at the shifted point is going to be very sensitive to circuit conditions.   So instead of zener use a resistive divider to set the DC voltage at the divider output to be equal to the DC voltage on pin 5.    The only reason I mentioned the crazier versions is it might sound OK  .

Yes, the cap seems obvious now....
The "diodes" might be back to back zeners in series as well  Did not have fixed 0.7v in my mind.

[Steben]

Which stage(s) clip inside the LM386 ?

In which order do they clip ?

Do we lose the LM386ness of the Opamp if we add external diodes (ie it becomes equivalent to any other Opamp if external diodes control the clipping)

The idea would be to add some soft clipping with some resistance in series in order to have the chip itself still clipping. The treshold of "the diodes" shoudl be not much lower than the clipping of the chip itself.

[Steben]

By maintainig gain cap between pin 1 and 8 or not..??

Interesting. You mean higher gain is less feedback and softer onset?

[Vivek]

Please check this logic and inform me if there is an error :

Lm386 clips due to rail saturation

Which depends upon the rail voltage

Hence, to soften down the clipping, we will need external clipping diodes that are referenced to Vcc.


If I remember, Rob S. has posted a schematic of clipping level referenced to rail.

[antonis]

Maybe bootstrapping supply rails could result into softer "onset" clipping..

[r080]

Like Merlin's glassblower? It seems like it might be possible to try that without too much trouble.

http://www.valvewizard.co.uk/glassblower.html

[Steben]

Never knew blowing the glass ceiling might result in softer onset of clipping.
Can anyone explain? Is it about the base-emitter connection?


Fact: a 386 based amp draws some nice variable current that can be used to create power sag.

[r080]

I don't know enough about bootstrapping to explain how it might work, but my understanding is it is a form of positive feedback. Since negative feedback reduces distortion and tends to make the transition to clipping faster, maybe bootstrapping would have an opposite effect.

[anotherjim]

Some chip amps you can overlay an opamp style Rf/Rin gain network. Not sure about a blocking cap - the 386 inputs are ground referenced - it would probably be needed from output pin5.

[antonis]

I don't know enough about bootstrapping to explain how it might work..

From Douglas Self (as usual.. ) Small Signal Audio Design

[Rob Strand]

FYI, something that occurred to me about putting diodes in the feedback loop is the feedback loop could go unstable.   Fixed gain amplifiers are often only stable with that gain.   Most opamps people use are unity gain stable.    When you add clipping diodes the gain drops significantly during clipping that could cause the LM386 to oscillate.    For the LM386 it might be possible to add a feedback cap to help stabilize it under clipping.

If you see oscillation issues then try pulling the clipping diodes.

[teemuk]

Why not just limit the input signal?

[Steben]

Why not just limit the input signal?

Tricky one. I do see the 386 as main clipper. Sounds best with full gain. Any clipping in front around onset of 386 would need very low treshold.

[teemuk]

You can always attenuate the (clipping) limited signal to chip's input range.

[Rob Strand]

This might workout for a softening idea.

The LM386 as two gain settings high (x200) and low (x20).  High gain bridges pins 1 and 8.

So what if you configure the amp as high gain but you set-up the clipping network so it soft clips instead of hard clips.
In soft clipping to gain limits at the LM386 low gain setting (x20).   That means it's still within spec as far as stability is concerned.

The difference is how low gain is implemented.   Normal we open pins 1 and 8 for low gain which raises the resistance between the emitters inside the LM386 from 150 ohm to 1.35k + 150 = 1.5k.    So instead of opening pins 1 and 8 for low gain, we leave pins 1 and 8 shorted then decrease the gain by paralleling a resistance across the 15k feedback resistor.   For a clipping network all we need to do is add resistance in series with the clipper clipping levels at the lower gain.

However ...

If you look at the gain of the LM386 it's quoted as x200 and x20  but notice the ratio of the resistors is 15k/150 =100 and 15k/1.5k = 10.  There's a factor of 2 missing.   I have a feeling missing factor of 2 comes from the second 15k which is between the emitter of the first transistor and pin 7!!!.  So maybe a clipper isn't so simple because shorting out one 15k only reduces the gain by a factor of 2.    Taking that idea further perhaps leaving of the bypass cap increases the gain.   I need to look at this closer.  Seems like a little bit more going on.

[Vivek]

Yes, I want to investigate the pin 1 and 8 approach also !!!

Maybe for a compressor or sag or tone controls.


There are Spice models for LM386 for blackbox approach. I will have to input or find a discrete model.