10
\$\begingroup\$

I am building an audio amplifier with LM386 with a voltage supply of 5 V.

The input signal is 50-250 mV.

Can anybody tell me why there is a resistor and capacitor in series at the output?

As the value of this resistor is 10 ohms, will it draw some power from the speaker and decrease its loudness or I am pointing in wrong direction? Pardon me if you find this post irrelevant. I am just curious because I am integrating it in my design.

enter image description here

\$\endgroup\$
2
  • 3
    \$\begingroup\$ The series RC network you marked is called "Zobel Network" -- basically a filter to tame the unwanted resonances. I'll not dive into technical details here as there are plenty of tutorials and explanations online. \$\endgroup\$ Dec 3, 2021 at 7:23
  • 1
    \$\begingroup\$ The 10 ohm resistor will only draw current at high frequencies thanks to the 0.05uF C. Calculate where the impedance of the C equals 10 ohms to see the frequency where that starts to be important. \$\endgroup\$
    – user16324
    Dec 3, 2021 at 14:14

2 Answers 2

16
\$\begingroup\$

Can anybody tell me why there is a resistor and capacitor in series at the output?

It's called a Zobel network.

The LM386 (just like its predecessor the LM380) doesn't like to run into anything like a high impedance load on the output. If it does it will become pretty much unusable. So, if you look at the impedance that a normal speaker has, you'll see that at higher frequencies, the impedance rises.

This is due to the inductance of the speaker coil. Here's a nice picture that shows speaker impedance in blue and, the combined impedance of speaker and parallel Zobel network: -

enter image description here

Image taken from this site (theradioboard.com). But, it's only the higher frequencies where the LM386 becomes unstable so, at the naturally low mechanical resonance point of most speakers (shown with the peak in the blue line at the left), the LM386 is fine.

As the value of this resistor is 10 ohms, will it draw some power from the speaker and decrease its loudness or I am pointing in wrong direction?

Yes, it will consume power but, at low frequencies (say 500 Hz) the 0.05 μF capacitor has an impedance of 6366 Ω and nobody is going to be much worried by any current taken through the Zobel network. At 5 kHz, the capacitive reactance is clearly still a relatively large value (637 Ω) and still an unimportant factor.

Compare this with a relatively lower nominal impedance of a speaker at moderate/mid-range frequencies i.e. 8 Ω and, the small current in the Zobel network is fairly insignificant.

Given also that audio spectrums tend to follow a "pink-noise" profile, the amplitudes at the higher frequencies are significantly less than at lower frequencies. In summary, you'd be hard-pressed to come up with a wasteful power argument against the Zobel network.

\$\endgroup\$
3
  • \$\begingroup\$ Interesting that these amps don't like low load conditions. Is there anywhere to read up on it or some explanation what is the reason ? I think most opamps are indifferent to low load conditions. \$\endgroup\$
    – tobalt
    Dec 3, 2021 at 13:34
  • 1
    \$\begingroup\$ @tobalt they don't like high impedances I suspect because of the configuration of the output transistors to get them to work on low dc supply voltages. The upper transistor is a regular NPN emitter follower (as used in conventional push-pull stages) but, the lower transistor is an NPN common emitter and therefore has a much different gain characteristic to the upper transistor. I expect that this is the main reason for instability and, I expect it is the lower transistor that goes unstable. Op-amps are symmetrical in their output stage with the majority being push-pull emitter followers. \$\endgroup\$
    – Andy aka
    Dec 3, 2021 at 13:44
  • 2
    \$\begingroup\$ @tobalt electronics.stackexchange.com/a/331668/122656 The instability that Andy mentions has its source inside the I.C....the only access to it is via the output pin, One of the few options to calm instability with an external network is crude swamping. It is fortunate that instability is in the multi-MHz range - far above audio frequencies. \$\endgroup\$
    – glen_geek
    Dec 4, 2021 at 14:00
7
\$\begingroup\$

When closed loop amplifiers drive a capacitive load, it can cause resonant gain peaks or even instability/oscillations. You can find a lot about this with the terms "opamp capacitive drive", "capacitive drive compensation".

To "tame" this as Rohat points out, can be done in two basic ways: a) reduce the gain at the resonant frequency or b) damp the resonance by a resistive load.

The option (a) is the slightly more elegant approach but requires some calculation/sims and it often referred to as "in-the-loop-compensation". Option (b) either places a series resistor in line with the amplifier output (often undesired especially for high loads), or puts a large resistive load in parallel with the capacitive load. This is what is shown in your schematic. The capacitor in series prevents excessive DC current flow. It has to be large enough to make the RC load impedance look rather resistive at the resonant frequency.

However, in your circuit the 250µF coupling capacitor is not a capacitive load because it is in series with the speaker. I am not sure why such circuit would require the RC damping network, it can be probably to provide some insurance against unintended speaker resonances in the ultrasonic range.

\$\endgroup\$
3
  • \$\begingroup\$ Moreover, the OP schematic doesn't look a closed loop? \$\endgroup\$ Dec 3, 2021 at 8:56
  • 2
    \$\begingroup\$ @linuxfansaysReinstateMonica The LM386 has internal feedback if I understand the the schematic in the datasheet correctly. \$\endgroup\$
    – tobalt
    Dec 3, 2021 at 9:05
  • \$\begingroup\$ Correct. I noticed now while looking at the internals. I see the manufacturer shows that RC in every place, without explaining. I bet it serves simply to stabilize "a priori", even if the load is or will not be capacitive. \$\endgroup\$ Dec 3, 2021 at 9:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.