This experimental (3) transistor class A audio power amplifier delivers 25mW into an 8Ω load, or 50mW into a 4Ω load using only a 1.5V power source. At such low voltages, there are many issues to consider and much to learn. To the best of my knowledge the following information is new to the world.
Months ago, I indicated an intention to write a piece on low voltage transistor application for another of my Single Transistor Amplifier Revisited series. Then a few weeks ago, Daniela, who participates in our forum, asked the simple, but profound question: “what is the minimum operating voltage of a transistor?” While, Mr. Marian attempted to narrow the scope of the question, no answers were given. This is on the heels of a similar article concerning Low Voltage Operation of the 555.
Theoretical minimum vs. practical minimum voltage
For a silicon bipolar transistor, the initial voltage requirement is to exceed the Vbe junction voltage of 0.6V. Then, to be able to current regulate this voltage via a series resistor, the source voltage must be about double this or about 1.2V. (The rule of thumb for shunt voltage regulators is that the source voltage is recommended to be double the regulated voltage.) This is perhaps the theoretical minimum Vcc. The practical minimum takes into consideration low voltage power sources – in this case, the ubiquitous 1.5V single cell battery is perhaps the standard low voltage power source – this is the power source for my circuit.
I have been toying with this idea for months and finally got around to bread boarding it. While it functions well, I am not suggesting that it is all that useful due to battery life limitations. I merely suggest that it is a really great experiment that shows what happens at low voltages.
While the power is relatively low (25mW), the sound volume is adequate and would be quite loud with earphones. Beyond this, the loudspeaker efficiency and enclosure acoustics plays an important role in the output volume. The frequency response is good due to the use of direct coupling – it is limited on the low end by C1 to 8HZ, and on the high end to well beyond 50kHZ. The output stage quiescent current flows through the loudspeaker – not recommended for higher power applications.
The direct coupling of the transistor arrangement (NPN, PNP, NPN) provides superior base drive required to saturate the output transistor.
Efficiency is about 20% that is close to the theoretical maximum of 25% for a class A amplifier. Note that unlike class B or AB, the class A amplifier DC supply current remains unchanged regardless of the output signal level – this makes it a real battery-killer!
The MPS650 output stage
The first transistor I tried was a 2N4401, but saturation was very poor and power was therefore limited. The MPS650 is like a 2N4401 on steroids – it is not an ordinary TO-92 device as its current rating is 2A and can easily saturate 100mA to 0.1V. A 4A power transistor may work just as well.
Observe the low distortion sine wave output – it sounds as clean as it looks.
Observe also the voltage of the base drive to Q3 – it is anything but sinusoidal – the wonder of negative feedback – without negative feedback, the distortion is unacceptable. The negative feedback reduces Av by the factor of the open loop gain the closed loop gain or 190 /5, and reduces the distortion by the same factor.
Low voltage phenomena
Observe the low voltage gain of each stage, as indicated on the schematic. Voltage gain suffers at low voltages. The open loop Av is 190 total. Note that at normal voltages e.g. 6V, this degree of voltage gain can be accomplished with a single transistor. The voltage gain of Q2 (Av = 1.4) is particularly low due to its load resistance being the base to emitter junction voltage of Q3.
This amplifier was very difficult to stabilize – never before have I had so much difficulty. Finally, I settled upon a 0.01uf capacitor across base to collector of Q3. However, I was not able to clean up the base drive signal – observe the thick raggedy trace on the oscillograph – this is an indication of a low level oscillation. Some compensation variations actually caused sub-cycle blocked oscillations – this leads me to suspect that the base circuit of Q3 is actually a negative resistance – similar to a tunnel diode. As base drive increases, the collector voltage decreases and Vbe also decreases thus causing increased base drive which in turn further decreases the collector voltage – this is due to the unusually high hRE effect at low voltages. If this is indeed true, this effect could be used to create an oscillator – something I have never heard of before.
Other devices to consider
I would love to see how germanium or MOSFET transistors perform at low voltages. I believe, that germanium is superior at low voltages, but have not experimented with them for really a long time. However, this is yet another of an endless field of potential experiments.
A word of caution – DANGER, ultrasonic noise!
Before listening to this via headphones, check the AC output voltage with a DMM or oscilloscope to verify that it is not oscillating. Even though you may not hear oscillations (perhaps 100kHZ) does not mean that it is safe for your ears. Ultrasonic noise is a grossly undocumented health risk.
Note that this topic may be a good topic for a research thesis – check out this link for more info.
Recommended devices for the serious experimenter
NPN Transistor: On Semi MPS651, 2A, 60V, TO-92, hFE = 75 @ 1A, DigiKey MPS651GOS-ND, $0.51
PNP Transistor: On Semi MPS751, 2A, 60V, TO-92, hFE = 75 @ 1A, DigiKey MPS751FS-ND, $0.43
For the future