Unleashing The LM386

Unleashing the LM386
Sverre Holm, LA3ZA,

http://www.qsl.net/la3za/
Dliveien 1, NO 1383 Asker, Norway.
The LM386 must be one of the most popular audio output amplifiers among radio amateurs,
despite having been around for a long time. It can be obtained in both dual-in-line and surfacemount packages and outputs 325 mW in the standard version that runs from 4-12 Volts supply
voltage. Its voltage gain of 46 dB is in many cases too little, especially in direct conversion
receivers. When I built the Pixie 2 with the LM386 audio stage, it struck me how the sensitivity of
the receiver was limited by the audio gain. I asked myself if it would be possible to increase the
gain and add some filtering without in a simple way, and the result is a gain of more than 70 dB
and an audio bandwidth of a few hundred Hz by only adding 3 resistors, 1 capacitor and an
inductor.
Analysis of the LM386 Amplifier

The basic LM386 amplifier is used without any feedback between pins 1, 5, and 8. The standard
way for getting a larger gain is to connect a large capacitor (say 10 F) between pins 1 and 8. In
either case, the voltage gain equation is:
Av ,1 =
2 Z 15
150 + Z 18
Here Z1-5 and Z1-8 are the impedances between the respective pins. This equation describes the
feedback path from the output to the emitter of the input stage, where the factor 2 is due to the
differential input stage. Z1-5 must also include the built-in 15k resistor which is in parallel to
external circuitry, and likewise Z1-8 should include the built-in 1.35k resistor. Thus, without
external components, it has a gain of Av = 2 15k /(150 + 1350) = 20 or 26 dB and with a large
capacitor between pins 1 and 8 it has a gain of Av = 2 15k / 150 = 200 or 46 dB.
The application note of the LM386 suggests a bass boost by connecting 10k in series with 33nF
between pins 1 and 5 with pin 8 open, while a common set of values among radio amateurs is in
the order of 2.2k and 4.7nF. The effect is a roll-off at frequencies above 1-2 kHz. The effect can
be analysed with the gain equation above by inserting 2.2e3 + 1/(j2f4.7e-9) in parallel to the 15k
internal feedback resistor. I use Matlab for this kind of analysis.
Additional Gain
The LM386 data sheet says Gain control can also be achieved by capacitively coupling a resistor
from pin 1 to ground. The effect of a low value resistor here is to decrease feedback and
increase gain. JF1OZL has measured the gain with various resistors and by going as low as to a
3.3 ohms resistor, he got 74 dB, see http://www.intio.or.jp/jf10zl/LM386.htm. In this case, the
feedback consists of a division between the Z1-5 and the Z1-gnd impedances, indicating that the
gain is found from the equation of an inverting feedback amplifier:
Av , 2 =
A0
Z
15
Z 1 gnd
Z 1 gnd
1 + A0
Z 15
The approximation is in the case that the open-loop gain, Ao, is much larger than the closed-loop
gain, Av,2. This will give a gain of 15000/3.3 = 4546 = 73.2 dB which is close enough to JF1OZLs
G-QRP SPRAT Autumn 2003
measurement. Such a high gain requires careful circuit layout with attention to ground loops and
proper decoupling, or the amplifier will oscillate. I have measured a gain of over 80 dB in a welldecoupled circuit, but then the amplifier is at the verge of self-oscillation.
High Gain and Filtering

The two ideas above can be combined in order to get both a high gain and high frequency roll-off.
Further, if the circuit between pin 1 and ground is a series resonance circuit, the bandpass
characteristic can be made even sharper. An inductor of 1 mH will resonate with the 100 F
capacitor at about 500 Hz and is fine. The problem now is that the gain will drop so much at the
higher frequencies that it gets below the value of 9 which is the stability limit for the LM386. To
limit the attenuation at high frequencies, the inductor has to be paralleled with a resistor and a
value of 220 ohms or less seems to be adequate. A potentiometer in series with the inductor
makes the gain variable. The resulting schematic is shown in Fig. 1.
Figure 1. LM386 with enhanced gain and filtering
Measurements
Measurements were made for this circuit with an LM386-N1 with date code 99 from National
Semiconductors at a supply voltage of 9 Volts, no output load, and no input coupling capacitor. In
this circuit, the DC resistance of the 1 mH inductor of about 2 ohms, and the equivalent series
resistance of about 0.5 ohms of the 100 F capacitor play a role and must be accounted for. The
upper curve is with no series resistance except for those of the inductor and capacitor, the middle
curve is with a total resistance of 10 ohms, i.e. 10-2-0.5=7.5 ohms series resistance, and the
lower one is for the pot at its maximum value of 200 ohms.
Figure 2. Measurements with R = 2.5, 10, and 200 ohms
Discussion
I tried different LM386-N1s in this circuit and found that the peak gain of the highest curve would
vary. Another chip with date code 99 gave 2 dB more gain, while a third one with date code 92
had 7 dB lower gain. I also tried a surface mount LM386-M1 dated 93 which had 5 dB less gain.
In all cases, the gain at 1 kHz hardly changed at all. Also, the peak gain is sensitive to output
load. The peak value of the highest curve would fall by 8 dB with a 32 ohm load, while hardly
changing at all at 1 kHz. These results suggest that the gain, Av,2 depends on the open-loop gain,
and that the open-loop gain varies from batch to batch.
The muting of the amplifier by means of pin 7 (SPRAT no. 113, Winter 2002/3, pp18-20), is still
possible, but only if pin 7 is connected to Vcc. Grounding of pin 7 will only mute the amplifier in
the basic 26 and 46 dB circuits, while the amplifier of Fig. 1 will instead output a low-level low
frequency noise.
I also tried to find the gain equation for the circuit in Fig. 1 and my guess was Av = Av ,1 + Av , 2 .
However, this equation overestimates gain by something like 8 dB except for the peak of the
highest curve. Maybe some readers who are more into the inner workings of this amplifier can
come up with a better equation.
In summary, addition of a few components to the basic LM386 amplifier results in a response
which is fine for CW reception with a peak in the 500 Hz range. The amplifier also has enough
gain for a direct conversion receiver with a passive mixer like the Pixie 2, that it in many cases it
can drive a loudspeaker. Hopefully, the circuit can benefit other direct conversion receivers also.

Unleashing The LM386

Uploaded by

Copyright:

Available Formats

Unleashing The LM386

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Unleashing The LM386

Uploaded by

Copyright:

Available Formats

Unleashing the LM386

Sverre Holm, LA3ZA,

Analysis of the LM386 Amplifier

G-QRP SPRAT Autumn 2003

High Gain and Filtering

Figure 1. LM386 with enhanced gain and filtering

G-QRP SPRAT Autumn 2003

Figure 2. Measurements with R = 2.5, 10, and 200 ohms

G-QRP SPRAT Autumn 2003

You might also like