LM386 Output Impedance

[hgg]

Hi,

Does anybody knows what is the impedance that the LM386 presents at its output pin?

I want to know if by placing a capacitor from its output pin to ground, apart from
creating a current bank, I am also creating a low pass filter.

Thanks.

[fcb]

You needs a Zobel network on the LM386 when driving speakers). Don't rely on output impedance to form an LPF, you'll end up with a power oscillator!

Form your LPF at the input, not the output of the part.

[hgg]

Actually, I want to avoid making a low pass filter on the output, that is why I want to know
its impedance.

[David_AVD]

Maybe specify what you're trying to achieve and a diagram?

[hgg]

I've just told you.  I want to avoid creating a low pass filter when I place a capacitor on the
output of the LM386 and ground.

For example the op amp's output impedance is almost zero.
I cannot find the output impedance of the LM386 in any datasheet.

[David_AVD]

So, explain why you want to put a cap (to ground) on the output.   

[hgg]

I've also told you that as well...

I want to know if by placing a capacitor from its output pin to ground,
apart from creating a current bank, I am also creating a low pass filter.

[David_AVD]

I've also told you that as well...

I want to know if by placing a capacitor from its output pin to ground,
apart from creating a current bank, I am also creating a low pass filter.

So what does that really mean?   

Do you want to low pass filter the output or "create a current bank" ?

[Andy Watson]

I do not understand why people ask questions and yet are evasive when asked to clarify the context. However, since you want a specific answer: The output impedance is irrelevant because strapping a capacitor to ground will create a low pass filter. Whether this is significant to you requires more information.  But, note FCB's warning.

[hgg]

So what does that really mean?   

1) I want to create a "current bank".
    By placing a capacitor on the output of the LM386 you are creating a "current reserve"
    which will help on sudden current surges.

2) I want to avoid creating a low pass filter
    If the LM386 presents a respectable amount of resistance on its output, then if you
    ground that resistance with a cap you create a low pass filter which will decrease
    the bandwidth of the output. 

That is why I want to know the LM386 output impedance, in order to choose the maximum
capacitor for the bandwidth I want.

The output impedance is irrelevant because strapping a capacitor to ground will create a low pass filter

Why is that so? 
If I understand correctly, if the impedance is not zero then the cut-off frequency could be
pretty low and the bandwidth would be crippled.

F = 1 / (2 * pi * R * C)

[David_AVD]

1) I want to create a "current bank".
    By placing a capacitor on the output of the LM386 you are creating a "current reserve"
    which will help on sudden current surges.

Can you elaborate on the ultimate reason you want to do this?  I'm guessing it's not being used as an audio amplifier?

[hgg]

It will be used as an audio amplifer driving an 8 ohms speaker, but it will also output
frequencies greater than 20Khz.  (100Khz max).

But my initial question was much more simple than that.
Does the LM386 present an output impedance unlike an op amp?
Maybe this is not a known value and I am asking in vain.

[Andy Watson]

The output impedance is irrelevant because strapping a capacitor to ground will create a low pass filter

Why is that so? 
If I understand correctly, if the impedance is not zero then the cut-off frequency could be
pretty low and the bandwidth would be crippled.

F = 1 / (2 * pi * R * C)

Because you can always take the frequency to a high value at which the capacitor impedance will cut-off the bandwidth. The frequency at which the cut-off occurs may be outside the range of you application. However, I don't think adding capacitance across the output is going to have the effect that you expect, also, if the capacitance is large enough to have an effect it will (almost by definition) impinge on the frequency response.

[David_AVD]

To be honest, I think you're going about this the wrong way.

If the LM386 doesn't have enough output drive for the application, you'll need to use something else.  Adding a capacitor to the output is not going to help.

The LM386 data sheet will specify what impedance it will safely drive.  Is your load less that this?

[hgg]

Because you can always take the frequency to a high value at which the capacitor impedance will cut-off the bandwidth

Yes, I forgot.  ESR...   That is pretty low anyway, so it will not affect my application which needs a
bandwidth of 100Khz.  Thanks.

Adding a capacitor to the output is not going to help

I don't know if you have tried it, but it actually does help.
The load is an 8 ohms speaker.

[David_AVD]

I'm not sure if anyone can help you if you don't tell the whole story of what you're trying to achieve.

[hgg]

...Achieved...
https://www.eevblog.com/forum/projects/pulsing-light-amplifier/

Fine tuning now.

[hgg]

@Andy Watson

Because you can always take the frequency to a high value at which the capacitor impedance will cut-off the bandwidth.

Now that I think about it, I am not sure that the above is correct.




The ESR impedance presents itself in series with the capacitor,
so it will not act as low pass filter. It will only limit the current to
ground at high frequencies.

Anyway, what about the actual impedance of the LM386 output pin?
Is this a known value?

If not, can it be calculated somehow?

[w2aew]

Maybe I'm missing something on what you're trying to accomplish...

Using a large capacitor as a "current bank" is something you'd do on the power supply rails - to keep them from sagging during transient high-current demands from the circuit.  Putting a large capacitor on the output of the audio amplifier only serves to make the amplifier work harder (push/sink more current) to charge/discharge the capacitor in order to change the output voltage at the speaker. 

In other words - you WANT the voltage to vary at the 386 output, and you DON'T want the voltage to vary on the supply.  That's why you'd put the cap on the supply.  Putting it on the output is counter-productive, placing a heavy load on the amplifier output.

[Mark Hennessy]

Agreed. Slapping a capacitor across the output of an audio amplifier usually turns it into an oscillator. After which, the magic smoke often gets out. Bad, bad idea. Properly designed audio amplifiers have inductors in series with the output to reduce the risk of this, of course.

The open-loop output impedance of an op-amp is a complex affair. But in the case of the LM386, which is designed to drive a loudspeaker, the short answer will be "as low as possible". Perhaps just a couple of ohms. Why don't you measure it? Not with a Fluke, obviously, but by applying Thevenin's Theorem?

http://en.wikipedia.org/wiki/Th%C3%A9venin%27s_theorem

You'll probably have to close the loop to do this, in which case you'll have to calculate the open-loop Zout from your knowledge of the feedback factor at the test frequency.

The responses you got on this thread were pretty reasonable - it's not an easy answer to give based on the information you gave. This stuff is actually quite involved - to get started I'd recommend reading Douglas Self's power amplifier design book, assuming you're happy with the basics of transistors. PDF versions can be found online.

[hgg]

Hello w2aewm,

I used the capacitor because it was suggested in the LM386 datasheet.
http://www.ti.com/lit/ds/symlink/lm386.pdf
They are using a 47nF capacitor.

Now that you mention it, I did not take into account the 'sinking' part, and the capacitor
will indeed place an additional load on the output.  Why are they using it, then?

Maybe I'm missing something on what you're trying to accomplish...

I am trying to create a flat frequency response and apart from the opamp bandwidth, I am trying
to find out what other components might act as low pass filters in my circuit.  I liked the idea of
having small wireless light radios in the house by modulating the audio signal to the room LED
lighting.  I will make very small versions of the L2S that will read the modulated light and reproduce
the music.

I have already done that, but it seems that the circuit acts like the following equivalent equalizer
on the signal :




Sample music transferred : http://chirb.it/dnPd4C

If I solve this problem, the transferred audio would be of a very respectable quality.
I guess one thing is the opamp I used which has a low bandwidth but I am trying to find
other possible causes as well.

Mark, thank you for your suggestions.

[qno]

This could be the light sensitive transistor.
These are high impedance devices where capacitive load is influenncing the frequency charracteristic.

You could try FM or PWM modulation

[hgg]

Hi,  its not a light sensitive transistor.  I am using a the BPW34 photodiode which is supposed
to be very fast.  Its capacitance is 25-40pF @ 1Mhz.

[qno]

Same thing with the diode.

This is a high impedance device. Any capacitive load of the opamp it is connected to or PCBtrack or wire wiil form a low pass filter.
Try some preemphase at the transmitter.

[hgg]

Pre-emphasis.  Interesting. 
I was looking exactly for something like that .
Thank you!

[fcb]

This thread is bizarre.  Current banking...

So you are trying to build a small "wire-less" amplified speaker.

My suggestion would be use FM over your optical link, much much easier to get a flat response.  Use a 4046 to demodulate the FM carrier (say 40-60KHz) - Philips did a great 4046 app note many years back. Once you've got good audio, then worry about the amplification.

BTW the LM386 is shit.

[Hideki]

I don't think you understand WHY they put a capacitor there in the data sheet. It's a zobel network, and in most of the drawings you may notice that the capacitor is in series with a 10 ohm resistor.

The speaker may present about 8 ohms at lower frequencies (except for its resonance peak), but its impedance will rise at higher frequencies (it is a after all a big inductor) so the capacitor gradually shunts in the 10 ohms in parallel with the speaker to keep the total impedance seen by the LM386 about the same.

It has nothing what so ever to do with "current banking" - whatever that means. It does have something to do with stability.

[w2aew]

Hello w2aewm,

I used the capacitor because it was suggested in the LM386 datasheet.
http://www.ti.com/lit/ds/symlink/lm386.pdf
They are using a 47nF capacitor.

Now that you mention it, I did not take into account the 'sinking' part, and the capacitor
will indeed place an additional load on the output.  Why are they using it, then?

It's there to ensure stability (part of a Zobel network, as mentioned previously http://en.wikipedia.org/wiki/Zobel_network). It does not affect the frequency response.  47nF has a capacitive reactance of about 170 ohms at 20kHz.  This is not in the feedback path, and appears in parallel with your speaker (8 ohms?), so does not affect the frequency response to the speaker really at all, at least not as a low pass filter as you may have been expecting.

I suspect, as the others have stated, that the rolloff you're seeing the frequency response is due either to your sound-light conversion scheme, or your light-to-sound conversion.  Even though you're using a low capacitance photodiode to pickup the light, the circuit configuration will have a huge affect on the achievable BW (photoconductive mode vs. photovoltaic mode, etc.).   I highly suspect that your BW limitations lie in the S2L and L2S circuits...

[hgg]

FCB, I thought of using FM as well but I am a newbie and I though that I would learn much more,
if I try to find the cause of the problem.  If I fail, I will try FM.  (and mess up even more  )

Hideki, very good explanation!  Thank you.  When I was trying to figure out why they were using
the capacitor/resistor to ground, I couldn't find any info.  Then, one website was supposedly
explaining the reason.  http://www.learningaboutelectronics.com/Articles/How-to-connect-a-LM386-audio-amplifier-chip
It turns out that not only the explanation was wrong but the LM386 wiring was totally wrong as well...

w2aew, yes the problem is definitely in the S2L part, because when I used the 555 to test the
frequency response, the result was much better.  (but still there).  I need to make a LED pulser
that will have the same amount of light intensity up to 100Khz and then look at the L2S circuit
for similar problems.

Thank you very much for your help!

[TonyStewart]

I know this is an old thread but this audio amp is intended for lower power from a battery. Yet at max Vout [Vpp] vs V+ in ,  this output stages are based drive limited push-pull Darlington & https://www.wikiwand.com/en/Sziklai_pair  output are limited on the high side which protects it but also makes the Rout very high for a power amp at Vmax out.  My calculations show this rises from 4 to 6 ohms using the method of KVL at 50% load loss rising at maximum output power.  This means damping factor is low in this condition.

It would be better to use complementary Darlingtons driven from a Rail to Rail Op Amp with negative feedback as the excess gain lowers the output impedance reduces the error with effectively a lower regulated output impedance.  Don't be fussy about biasing the 2.4V drop between the bases to prevent shoot thru as the GBW of R2R OA will eliminate any crossover distortion and not be visible or audible.

" alt="" class="bbc_img" />