POWER AMPS [ discrete transistor, low voltage, low power for transistor radios ]
Click on an image to view a larger size
[1] Rail-to-Rail Power Amp Attempt (my best ever) - Split DC Supply
added April 2, 2026
Above — Schematic of my attempt to swing a 1 KHz sine wave as close to the DC rails as possible. This is perhaps a goal to strive for as you get better at making audio power amps.
Using 2 finals with collector output [ some call them 'collector followers' ] This avoids collector to emitter saturation voltage + VBE limitations we see in ordinary emitter follower power stages. Essentially, this is a Class AB power amplifier with a complementary pair of common emitter amps with their collectors driving the load -- the finals provide both power and voltage gain. The output impedance is low because of negative feedback -- negative feedback reduces the collector output impedance by the open-loop gain of the amplifier.
I drove these finals with the fastest opamp I stock - the LM4562. This is a beautiful part -- and I built our home HiFi system with this lovely op-amp. It loves to oscillate however -- and the 0.022 µF capacitor across the Q1 + Q2 bases proved essential to suppress HF oscillations.
I wanted close to a 12 VDC rail -- so voltage regulators were placed on each rail using 6.2 volt zener diodes. Thus My rails were +5.84v and -5.78v --- or 11.62 volts DC. I wanted my rails close to 12 VDC to make amp to compare to other 12 volt DC single-supply PA stages I designed & built.
Above — FFT of the amp. My signal swings 11.1 VAC -- which is 2 Watts output power between a 11.62v rail. Yay! This is the closest I've ever come to rail-to rail output. I'm chuffed! It's much easier to do this with a split DC supply.
Above — FFT of the rail-to-rail PA driven too hard at 11.3 VAC peak-peak. Serious clipping arose. This experiment proved instructional -- I am making a video about a Sony chips that uses "collector followers" for audio power and now I'm better informed to comment about this technique.
[2] Simple but good - discete BJT PA system with lots of voltage gain as needed
added March 30, 2026
Above — My current go-to audio power amp for my radios. Love this amp. Although it uses 12 transistors - the design is simple + optimized. Like many audio chip PA stages -- this PA uses a NPN emitter follower Q10 for positive going signals and a compound-pair (Q11 + Q12 in 1:1 feedback) for the negative going signal in the "PA PNP slot".
Q3,Q4 - the input stage emitter coupled pair get sunk by a current sink & collector loaded with a current mirror. The Q1 high impedance output connects to an emitter follower Q7-- and then to the main voltage amp Q6.
This makes the open-loop gain super high [Vg =1080] which is desireable. Cf acts as phase shift compensator + gives local negative feedback that works well to lower distortion from the middle to high frequencies.
Q8 + D3 set the bias for the finals. You could technically remove Q8 and just use D3 instead. D3 is normally glued to Q10 for thermal tracking.
Quiescent current = 13.3 mA. Super low DC ripple due to the 100 Ω | 220 µF RC low-pass filter on all but the finals.
Above — FFT of the output. Gives a little over 9 volts peak-peak with all harmonics under -60 dBc. Great performance from a simple final output stage.
[3] Compound Feedback Pair driven with an op-amp
added Dec 6, 2025
Above — Good 12 volt PA. Rather than drive 2 successive complimentary emitter followers pairs per standard, I went with the complimentary compound feedback pair Q2|Q4 and Q3|Q5. The gain of the finals is 1.x – not too high so that it creates oscillations above 100 KHz.
They’re easy to drive, very stable, and do not require a current source to drive fairly well. I went with an adjustable bias transistor with a trim pot to carefully control Vbias. I decided on a relatively low 15 mA quiescent currrent.
If I increase vBias to a quiescent current ~ 50 mA , I can reduce the ~2 KHz harmonic by 3-4 dB. All scope shots are at 15 mA quiescent current. It’s always fun to get a swing of 10 VAC peak-to-peak in a ~12 VDC single DC supply audio PA. This audio distortion performance beats all the commercial Ham radio receiver AF PA amps I’ve ever tested, or read test reports about -- even with 15 mA quiescent.
Further, output networks such as the Zobel [ 10 ohm resistor in series with a small cap shunt to ground ] , plus a feedback network across the 18K resistor to reduce closed loop gain at higher frequencies might help when running a speaker.
Reducing the entire PA stage ‘s voltage gain by lowering the feedback resistor value (from 18K) also lowered distortion on my FFT. Distortion is an inverse function of the closed loop gain.
Above — FFT of the compound feedback pair driven to ~ 3 Vpp ( 146 mW).
Above — FFT driven to ~9 Vpp or 1.34 Watts at ~1 KHz.
Above — FFT driven to 10 Vpp or 1.63 Watts at ~1 KHz. It's always fun to get 10 Vk-pk on a ~12 VDC single supply power supply. HiFi grade.
