Audio Amplifiers

Modest power audio amplifiers for driving small speakers or other light loads can be constructed in a number of ways. The first choice is usually an integrated circuit designed for the purpose such as the LM386 or newer class D switching types that often accept digital data instead of simple audio voltage. Discrete designs can also be built with readily available transistors or op-amps and many designs are featured in manufacturers' application notes. Older designs employed audio interstage and output transformers but the cost and size of these parts has made them all but disappear. Here are a few easy-to-build analog audio amplifier circuits for a variety of hobby applications:

Simple LM386 Audio Amplifier

This simple amplifier shows the LM386 in a high-gain configuration (A = 200). For a maximum gain of only 20, leave out the 10 uF connected from pin 1 to pin 8. Maximum gains between 20 and 200 may be realized by adding a selected resistor in series with the same 10 uF capacitor. The 10k potentiometer will give the amplifier a variable gain from zero up to that maximum.

High Gain and Fidelity Audio Amplifier

Here's a general-purpose 2 watt audio amplifier with excellent overall performance. It is easily configured to serve as an audio output stage for lower power applications or as a more powerful amplifier for room-filling volume.

The output stage uses a unique technique to stabilize the quiescent current without the traditional temperature-compensating diodes. Each half of IC2 acts as a voltage-follower through its corresponding power transistor, holding the differential voltage across the two 1 ohm emitter resistors to a precise value set by the selected 220 ohm across the inputs. With a 15 volt supply, one can calculate that there's about 16 mV across the 220 ohm in the schematic and the op-amps force that voltage to appear across the two 1 ohm resistors, giving a bias current of about 8 mA (modified slightly by op-amp offset voltages). The quiescent current is easily changed by adjusting that one resistor value. Dropping to zero ohms gives excellent results on one prototype, hitting the lowest distortion level I've seen. In this way the quiescent current can be dropped to nearly zero with still excellent fidelity - perfect for small battery projects. When the audio stream stops current drops to about 1.5 mA on that prototype.

The 1N4148 diodes reduce the op-amp output swing when the related transistor turns off; the op-amp only has to swing a diode drop to maintain control over its negative input. When the diode is conducting that half of the circuit is essentially off. This little trick combined with the full gain and bandwidth of the op-amp being used to simply follow the input voltage results in very low distortion, even at full output and high frequency. Ironically, a germanium diode works well here, reducing the range the op-amp must swing even more (as does a Schottky type).

Which op-amps to use

The first two op-amps should be suitable for audio, like the MC33182, LM833 or many, many others. If you see the word "audio" or "distortion" anywhere in the manufacturer's data sheet, it's probably a good choice. : ) Remember that many modern op-amps have low power supply voltage ratings. The gain of those first two stages is set to only 11 each so that even slower op-amps will work fairly well but look for at least 1 MHz GBW. Some perfectly fine op-amps have less output voltage swing than is optimal for this circuit because the output stage has no voltage gain. Less output swing from IC1b means less maximum power but that's usually not a big issue. Don't bother trying the LM358 or LM324 for the first two stages; that part has a nasty cross-over problem when providing voltage gain.

The output op-amp is an LM358. Originally I used faster op-amps here with good results into an 8 ohm dummy load but instability showed up with some speakers due to their reactance. However, adding a snubber network like that typically used with an IC amplifier (like the LM386) should do the trick but some experimentation might be required. Try a 10 ohm in series with a good .05 uF ceramic capacitor to ground. With the LM358 a little distortion is visible as tiny wiggles near zero-crossing at several kHz but the resulting harmonics are well above the human hearing range. Those can be eliminated by a 10 uH inductor in parallel with a 47 ohm resistor in series with the output (also not a bad idea in general with such amplifiers). Something like this:

The 10 uH choke should have a low resistance compared to 8 ohms. Commercial molded snubbers typically with 1.5 uH and 47 ohms in parallel will work fine. The choke keeps supersonic frequencies from feeding back to the input causing oscillation.

I temporarily lowered the gain of the first stages by bridging the two 10k resistors with 1k resistors. Using a distortion analyzer I measured 0.16% total distortion at 15 to 18 VDC near maximum power. Not bad for an LM358! The distortion dropped to 0.1% while driving 16 ohms with an 18 volt supply. Substituting a 2N2219 and 2N2905 resulted in distortion just under 0.2% with 12 volts. Those are typical numbers for low-power amps like this. However:

Switching to an LS404 quad audio-grade op-amp and shorting the 220 ohm bias resistor resulted in an impressive .03% distortion at 1 kHz and 15 VDC while delivering 10 volts p-p to an 8 ohm resistor (first two amps have 1k feedback resistors for a total gain of four). That's pretty good distortion! I would add the snubber networks I mention above with this op-amp and make sure to check for stability with a scope. It's a little surprising that a little quiescent current doesn't help the distortion at all (actually hurts a little). Thus the current consumption is under 2 mA with no signal, another nice feature. Note: the LS404 op-amp runs out of swing at a p-p value 5 volts below the power supply voltage and that limits the total power available (to about 1.6 watts into 8 ohms with a 15 volt supply).

Which transistors to use

My prototype will operate from 9 volts to 21 volts without any transistor heat sinks using the TIP31 and TIP32. These complimentary TO220 transistors are rated to 2 watts in free air and they will be dissipating about 1.3 watt maximum at 21 volts with an 8 ohm speaker. That's technically okay but they will be quite hot to the touch. Little clip-on heat sinks might be a good idea in this case. For power supply voltages below 18 volts and an 8 ohm speaker, no heat sinks are needed. The maximum power my prototype delivers is approximated by:

Pmax = (0.183 * Vcc) - 1.3       (derived from empirical data)

You can use that equation to determine that my prototype delivers a respectable 350 mW into 8 ohms when powered by 9 volts. That's plenty of volume for little radio projects. At the other extreme, the equation predicts 2.5 watts into 8 ohms with a power supply voltage of 21 volts and that's just what I measured at the point of clipping. I used a 1 kHz sine wave for the tests.

Below Vcc = 12 volts the transistor dissipation drops below 350 mW so many small-signal types will work fine without a heat sink.

What problems will I have

This is a lot of gain in a small space and, to make matters worse, there's lots of current flowing in the output stage. Op-amps are quite good at rejecting feedback along the power supply rails and ground but stability can still be an issue. Bring the power and ground from the power supply to the circuit near the output transistors. Connect the input ground near the three 10 uF capacitors and 300k resistor ground connections. Also note the 1k and 10 uF filter at the input. The amp draws enough power that Vcc will move up and down a bit and letting that signal get back to the input will result in oscillation or, in my case, a mysterious drop in input impedance. That little RC filter prevents that feedback. You can also decide to just use less gain by dropping the 10k resistors or using only one stage of gain. A total gain around 15 or 20 is about right for most "line" outputs.

You can also have stability problems related to the op-amp and transistor choices, so an oscilloscope is a good idea to make sure the amplifier isn't misbehaving.  The LS404 seems completely trouble-free despite its bandwidth.

A regulated power supply isn't absolutely necessary but at least use a very large capacitor like the 2200 uF shown. Using a three-terminal regulator will add a degree of protection for the transistors in the event the output gets shorted to ground.

What good is it

There are IC audio amplifiers that work as well as this project but not typically, especially when comparing the distortion performance at full power. And this circuit uses ordinary parts that most experimenters have on hand. It's easy to change the quiescent current to fit the application and the circuit will work on a wide range of power supply voltages. Running the circuit on 9 volts makes for a nice general-purpose amplifier for small projects. Short the 220 ohm and the circuit practically turns off by itself!

I just hooked up a guitar, increased the gain of the first stage and this makes a perfect practice amp! With a good speaker and 18 volts it's surprisingly loud and the sound is quite clean. I'd add a volume control potentiometer across the input, with the wiper going to the 1 uF capacitor. The resistance of that potentiometer will set the input impedance of the amplifier. An audio-taper 10k pot would be a good choice.

The output section is the meat of this project and it would be fine to drive it with a different front-end amplifier. Just remember that the voltage swing needs to be nearly rail-to-rail to get maximum power since the output stage has no voltage gain.

Curiously Low Noise Amplifier

The Curiously Low Noise Amplifier takes advantage of the wonderful noise characteristics of the 2SK117 JFET that boasts a noise voltage below 1 nV/root-Hz and virtually no noise current. The noise voltage of the amplifier is only 1.4 nV/root-Hz at 1 kHz, increasing to only 2.7 nV/root-Hz at 10 Hz. The noise current is difficult to measure, so this simple utility amplifier can see the noise from a 50 ohm resistor and a 100k resistor, too. (The 1.4 nV input-referred noise will increase to about 1.7 nV with a 50 ohm resistor, instead of a short, and a 100k resistor will give an input-referred noise near 40 nV, with very little contribution from the amplifier.)

This amplifier is a "utility" amplifier with a gain of 100, that would typically be used in a lab setting to boost tiny signals for measurement or further processing. It isn't intended to drive a speaker or headphones directly. (It could drive the LM386 quite nicely.) The circuit is a simple discrete transistor feedback circuit with two gain stages and a unique class-A output buffer:

For even better performance, the bipolar stages could be replaced with a low noise op-amp. The input noise would drop a little, perhaps to 1 nV, as would the input capacitance, perhaps below 10 pf. Compensating the op-amp might be a bit of a challenge.

Computer Audio Booster

Here is a simple amplifier for boosting the audio level from low-power sound cards or other audio sources driving small speakers like toys or small transistor radios. The circuit will deliver about 2 watts as shown.  The parts are not critical and substitutions will usually work.  The two 2.2 ohm resistors may be replaced with one 3.9 ohm resistor in either emitter.


4-Transistor Amplifier for Small Speaker Applications


The circuit above shows a 4-transistor utility amplifier suitable for a variety of projects including receivers, intercoms, microphones, telephone pick-up coils, and general audio monitoring. The amplifier has a power isolation circuit and bandwidth limiting to reduce oscillations and "motorboating". The values are not particularly critical and modest deviations from the indicated values will not significantly degrade the performance.

Three cell battery packs giving about 4.5 volts are recommended for most transformerless audio amplifiers driving small 8 ohm speakers. The battery life will be considerably longer than a 9 volt rectangular battery and the cell resistance will remain lower over the life of the battery resulting in less distortion and stability problems.

The amplifier may be modified to work with a 9 volt battery if desired by moving the output transistors' bias point. Lowering the 33k resistor connected from the second transistor's base to ground to about 10k will move the voltage on the output electrolytic capacitor to about 1/2 the supply voltage. This bias change gives more signal swing before clipping occurs and this change is not necessary if the volume is adequate.

As before, the two 4.7 ohm resistors may be replaced with a single 10 ohm resistor in series with either emitter.

Op-Amp Audio Amplifier


The above circuit is a versatile audio amplifier employing a low cost LM358 op-amp. The differential inputs give the amplifier excellent immunity to common-mode signals which are a common cause of amplifier instability. The dotted ground connection represents the wiring in a typical project illustrating how the ground sensing input can be connected to the ground at the source of the audio instead of at the amplifier where high currents are present. If the source is a power supply referenced signal then one of the amplifier inputs is connected to the positive supply. For example, an NPN common-emitter preamplifier may be added for very high gain and by connecting the differential inputs across the collector resistor instead of from collector to ground, destabilizing feedback via the power supply is greatly reduced. By the way, the LM358 is a fairly poor audio amplifier and you may wish to switch to a better part for reduced distortion. Frankly, for a little bench amplifier, you'll never notice the distortion.

wpe12.jpg (9663 bytes)

My utility amplifier was built into an aluminum Bud box and eventually ended up bolted to the bottom of a shelf as shown. The well-behaved and ready-to-go amplifier is really handy.

Crystal Radio (and other purpose) Audio Amplifier

Here is a simple audio amplifier using a TL431 shunt regulator. The amplifier will provide room-filling volume from an ordinary crystal radio outfitted with a long-wire antenna and good ground. The circuitry of such a radio is similar in complexity to a simple one-transistor radio but the performance is superior (with the exception of the amazing one-transistor reflex ). The TL431 is available in a TO-92 package and it looks like an ordinary transistor so your hobbyist friends will be impressed by the volume you are getting with only one transistor and the amplifier may be used for other projects, too. Higher impedance headphones and speakers may also be used. An earphone from an old telephone will give ear-splitting volume and great sensitivity! The 68 ohm resistor may be increased to several hundred ohms when using high impedance earphones to save battery power.


Here is the amplifier used to boost the output from a simple crystal radio. The volume control is at the bottom left and the other components are on the terminal strip at the bottom of the picture. This is a really quick and easy audio amplifier! wpe10.jpg (12985 bytes)

Class-A Audio Amplifiers 

A class-A audio amplifier is pretty wasteful of power but when plenty of power is available the simplicity is attractive. Here is a simple darlington transistor example intended for use with a 5 volt power supply:


This circuit and the following aren't for beginners; they are of limited usefulness and require an understanding of the underlying principles and potential applications. They all pass DC through the speaker which is wasteful and can cause problems for the inexperienced builder. If built without variation, they should perform as described but make sure to read the text.

The 5 volts should be provided by a regulated power supply. The efficiency is below 25% and significant DC current flows in the speaker and that additional power should be figured in to the power rating of the speaker. But look how simple it is! The voltage gain is only about 20 and the input impedance is about 12k. The schematic shows two values of bias resistor to be used with the corresponding speaker impedance. With the 150k bias resistor and 8 ohm speaker, the circuit draws about 210mA (1 watt) and can deliver about 250 mW to the speaker which is plenty of volume for most small projects.   The speaker should be rated at 500 mW or more and should exhibit a DC resistance near 8 ohms (perhaps 7 ohms). Check the candidate speaker with an ohmmeter; much below 7 ohms will cause excessive current draw. With the 220k resistor and 16 ohm speaker, the circuit draws about 100 mA (500 mW) and delivers about 125 mW to the speaker. The 16 ohms speaker should be rated at 200 mW or more and exhibit nearly 16 ohms of DC resistance. (Most small speakers have a DC resistance near the rated impedance and that resistance is used to set the quiescent current level in this circuit.) Other NPN darlington transistors will work but choose one that can dissipate 1 watt minimum. Most power types don't need a heatsink but tiny TO92's might overheat.

If the inefficiency of the class-A hasn't dissuaded you yet, here is a 4-transistor amplifier suitable for small signals:



The input impedance is about 5000 ohms and the frequency response is flat from 30 Hz to over 20,000 Hz. With the 8 ohm speaker the current drain is about 215 mA and the gain is about 1700 (64 dB). With the 16 ohm speaker the current gain is about 110 mA and the gain is about 2500 (68 dB).  A volume control may be added by connecting one end of a 5k potentiometer to ground, the wiper to the amplifier input. The other end of the pot becomes the input.

Lets face it; just about any of the various IC audio amplifiers make more sense than this inefficient design. But, this circuit uses parts with only 3 legs. Umm, it doesn't use large capacitors except for the power supply bypassing. Lets see, its more fun-ariffic.  Well, lets see if we can come up with a project that takes advantage of the inefficiency:


So, what is it?

It is a modulated light sender! Connect the input to an audio source or microphone (a speaker will work) and the audio will amplitude modulate the light intensity. The inefficiency of the class-A works in our favor now, lighting the lamp to mid-brightness with no audio present. Actually, with a 4.7 volt bulb, the lamp will be near full brightness and will be "overdriven" on sound peaks. A higher voltage bulb will last longer but will be dimmer. Try a 6.8 volt bulb as a compromise. With a sensitive detector like a phototransistor, this communicator will work several hundred feet (at night). Best range is realized if the bulb is mounted in a typical flashlight reflector and the detector is similarly mounted. The input capacitor is reduced to .01 uF to give the amplifier a high-pass character to compensate for the slow response of the bulb. The audio will sound a bit muffled, anyway. The clever designer could use this amplifier for the receiver, too, switching the speaker to the input for transmitting and to the output for listening. If you choose a detector with good infrared response, like a pin photo diode, you can add plastic IR filters to block out ambient light and make the communicator harder to see at night.

Increasing the voltage to 12 VDC, replacing the bulb with a  3 watt, 16 ohm speaker and replacing the .01uF with a 1uF gives an audio amp that will deliver nearly 1 watt of audio power. The speaker will get warm, however! (Due to the nearly 2 watts of DC power in the speaker coil.)