Welcome to buchanan1.net

J.R. Buchanan


A simple 2 watt audio utility amp

Early to mid '90s

Why only 2 watts? With a single 12V supply, you can get a peak-to-peak output voltage of almost 12 volts. With a sine wave output, this gives you an RMS voltage of:


With an 8 ohm speaker, this gives you:


There are two other effects. Since I wasn't going for a lot of power and I had a bunch of 1 ohm resistors, I used two of them for R12 and R13. This cuts into the available power. On the other hand, most 12 volt supplies are either vehicle batteries or power supplies intended to emulate them. They put out closer to 13.6V, which gets you a bit more power. On the gripping hand, I wasn't trying to build a powerhouse here, so I don't really care. I didn't even measure the power output when I was done.

By the way, the amp will safely drive a 4 ohm speaker, but the power won't quite double, as the losses in R12 and R13 will be greater.

I've built a stereo version and a mono version of this circuit so far. The mono version is just a utility amp driving an old PA speaker in my shop. Right now it's mostly on a breadboarding socket. It was the first. The stereo version is built on perf board and it's on my computer desk at work, connected to two car stereo speakers in foam-filled wood boxen on one end and a Walkman-style cassette player on the other (You can tell this is from quite a while ago...). The gain was a bit high for the later use, I just put a couple attenuators in the line to the amp input. Two watts per channel is plenty into these fairly efficient speakers. When no one else is around, the sound in my cubicle can get loud enough to be quite uncomfortable.

Schematic here

C1  5uF, 25V aluminum electrolytic
C2  1uF, 50V aluminum electrolytic
C3  270pF ceramic disk
C4  100uF, 16V aluminum electrolytic
C5  2200uF, 16V aluminum electrolytic
D1  1N4148
D2  1N4148
Q1  2N3904
Q2  2N3906
Q3  2N3906
Q4  TIP31
Q5  TIP31
Q6  TIP32
R1  15K   1/4W
R2  1K    1/4W
R3  10K   1/4W
R4  470   1/4W
R5  10K   1/4W
R6  2.2K  1/4W
R7  2.7K  1/4W
R8  4.7K  1/4W
R9  100K  1/4W
R10 15    1/4W
R11 240   1/2W
R12 1     1W
R13 1     1W

Note: The voltages listed on the electrolytic capacitors are the voltage ratings of the ones I had on hand. Lower voltage units could be used in many cases.

This circuit will draw a little under 200ma at full output.

When I first hooked this circuit up at work (for the cassette player) I got some fairly nasty hum in the speakers. I'd made a classic mistake and tested it on a clean bench supply, not a typical noisy voltage source.

In this case, I solved the problem by cleaning up the crude supply I had under my bench. If that were not an option, I suspect that using a simple RC filter for the power supplied to the earlier stages would solve the problem. If you come up with a solution, or build this circuit and need help with one, feel free to email me.

The output stage is a simple complimentary-symmetry amplifier.

C5 is charged to 1/2 the supply voltage under quiescent conditions. The output is pulled down almost as low as -.5Vcc when Q6 pulls its emitter to nearly ground potential. When the emitter of Q5 is pulled up almost to the supply voltage, the output is at almost .5Vcc

Bias current for Q5 and Q6 is provided by R11. If the upper end of R11 were connected to Vcc, then when the base of Q5 was allowed to rise to near Vcc, there would be very little voltage drop across R11. This would not provide sufficient base drive current to Q5, just when it needs it the most. To solve this problem, C4 and D2 are used. When the emitter of Q5 pulls the output of the amp high, the voltage across C4 is added to this. The junction of D2 and R11 can reach almost 2xVcc this way. This technique is called bootstrapping and is quite common in audio amplifiers, as well as vertical sweep amplifiers in TV sets. (Again, you can tell this is from quite a while ago, what's a "vertical sweep amplifier", you say...)

Q4 is the driver transistor. It pulls the bases of the output transistors down. It has a much easier job than R11 and doesn't need as much help. :-)

The bias current through D1 and R10 develops a bias voltage that is dropped across the base-emitter junctions of Q5 and Q6, as well as R12 and R13. This sets up a small bias current that flows through both transistors even when there is no output. This prevents what is known as "crossover distortion", a slight notch in the output waveform at the zero-crossing that occurs when on transistor takes over from the other.

Initially, I used two diodes instead of the D1 R10 combo. I couldn't find any in my junk box that provided a bias current I was happy with, so I used R10.

Q2 and Q3 are a differential amplifier. Each base is an input, each collector an output. I only needed one output, so I grounded the other (the current has to flow somewhere). The other output is connected to the base of Q4. This is a current node. The voltage here has a logarithmic relation to the signal. It's really not too worthwhile to measure the voltage here with a meter or a 'scope. Well, maybe for educational purposes...

Resistors R9 and R5 provide negative feedback. I've got it set about as low as possible without too much distortion. This amplifier is not too linear without feedback (naturally it is linear with feedback). Since C2 is in series with R5, the feedback is a little more at DC, in fact the DC coupled part of the amp has a gain of about 1 for DC.

R7 and R8 set the output voltage of the amp (as present on the + side of C5). You could look at Q2-Q6 as an op-amp, with the base of Q3 being the non-inverting input. The main trouble with this is that the gain isn't high enough for the standard formulas used with op-amps to provide accurate answers, but it's kinda close.

All amplifiers start to exhibit phase shift as the frequency increases. If you have negative feedback, it is possible at some point the amplifier will exhibit enough phase shift and still have enough gain left that an elaborate oscillator is formed.

By using one method or another, the gain at higher frequencies may be lowered so that oscillation is not a problem. That is what C3 does.

C3 was selected empirically. The circuit did not oscillate without it until I drove it with a square wave. Then it rung badly at a low RF frequency, but the oscillations damped out before the next cycle. I selected the value for C3 by raising the value until the ringing stopped then went a bit higher just in case. Yes, I took transient analysis and feedback theory in college, but believe me, this was the easiest way...

Q1 and R1-R4 are a simple common-emitter amplifier. I needed this since I wanted more gain in the prototype than I could get without lowering the feedback to the point of distortion. With this stage, the gain was a bit high for the Walkman-Clone output, I just used a simple attenuator in the line from the cassette player and left the amp alone.

The gain of this amplifier when loaded by the approximately 10K input impedance of the next stage is about 10.

Negative feedback is generated across the unbypassed emitter resistor, R4.

R1 and R2 are a voltage divider that set the base voltage of Q1. The emitter voltage is approximately .7 volts lower and this voltage across R4 provides the bias current.

When I had the hum problems, I think this is the stage that the hum got in through. I'd filter its supply and maybe the supply to the differential amp (have to think that through) before I got too concerned about the problem.