3 Transistor Op-Amp




  The 3 transistor op-amp shown was designed for use as a 20 milliwatt
  headphone amplifier (40 ohm headphone). The circuit requires 10 parts
  but lacks efficiency since the output is single ended and draws around
  60mA from AA batteries, or 120mA for a stereo application.

  Parts Selection:

  For a power output of 20mW (RMS), the peak output power will be 40mW
  and the peak output current will be I^2*R =.04 and I(peak) = 32 mA.
  The negative peak of the output signal occurs when transistor Q3 is
  completely turned off, so the output voltage is just the result of
  a voltage divider 40 ohms and R4. If the peak current is 32 mA, then
  the peak output voltage will be E=IR = .032*40 = 1.28 volts. Since the
  negative supply voltage is 3 and the peak load voltage is 1.28, there
  will be 3-1.28 = 1.72 volts across R4. So, we need a resistor that drops
  1.72 at 32 mA or R4 = E/I = 1.72/.032 = 54 ohms, or 56 standard value.

  Next consideration was the base and collector currents for Q3. Since the
  output voltage swing is symmetrical, the peak voltage across R4 will be
  3 + 1.28 or 4.28 and the current in R4 will be I= E/R = 4.28/56 = 76mA.
  The peak collector current of Q3 will then be 76mA plus 32 milliamps in
  the load or, 108mA total. This seems a little high for a 2N3906, maybe
  some other transistor would be better.

  Next consideration was the steady state DC current for the differential
  pair of transistors (Q1,Q2). The DC currents of Q1,Q2 should be somewhat
  balanced (with no signal) so that both transistors carry near equal DC
  currents. And this current should be somewhat greater than what is needed
  by Q3. The measured hFE gain of the 2N3906 at 100mA was about 100, so the
  peak base current of Q3 should be around 1 mA. The DC current for Q1,Q2
  was then chosen to be about 3 times greater, or 3mA for each transistor
  (6 mA total). It could be larger, but would increase battery drain.

  Next step was to work out the common emitter resistor R3. Since the base
  of Q2 is at ground, the emitters will be about -700mV and the voltage
  across R3 will be 3-0.7 = 2.3, and the value of R3 will be
  E/I = 2.3/0.006 = 390 ohms. The total current in R3 will then be 6mA,
  or 3mA for each transistor Q1 and Q2.

  Next step was to assign a value to the Q3 e/b resistor (R2). In the
  balanced (no signal) condition, Q3's base current will be about 0.5mA and
  the remaining current for R2 will be 3mA - 0.5mA = 2.5mA, so the resistor
  value would be R2=Vbe/.0015 = 0.7/.0025 = 280 ohms (270 standard value).
  Resistor R1 will have the same value and is used to measure the DC current
  in Q1 which should be about 3mA, so the voltage across R1 should read
  around 810mV if things are working right.

  The amplifier gain was selected to be ten or (20dB) which defines the ratio
  of R5 to R6, or 1K and 10K. This makes the input impedance about 1K since
  the inverting configuration is used. For a higher input resistance, larger
  resistor values could be used (10K 100K) etc. Or you could ground the 1K
  imput resistor R6 and use the non-inverting input point (base of Q2) as a
  high impedance input.

  Notes:

  A small 1000pF cap was added to limit the bandwidth to around 100kHz and
  suppress oscillations. The value was found by experiment.

  The circuit seems fairly linear (low distortion) as indicated in the
  scope picture below. The top waveform is the ramp input from a function
  generator, and the lower waveform is the inverted output of the op-amp
  with a 40 ohm load. So, it's actually a plot of the rising input ramp
  and falling output ramp (2.25v p-p). Notice there is a slight bend in
  the output line indicating some non-linearity (distortion) but hard
  to see.

  The DC offset at the output will move negative as the battery voltage falls.
  It measured about +50mV at full supply voltage and drifted negative to
  -150mV with the supply at +/- 2.5 volts. This can be corrected with a
  small adjustment to R3. Lowering R3 will increase the current in Q1/Q2
  and move the DC offset positive, and visa versa. Another approach is to
  replace R3 with a constant current source (one extra transistor) which
  should maintain the DC offset near zero as the supply voltage falls.
  But then it wouldn't be a 3 transistor opamp anymore. It would be a
  4 transistor opamp.
  

Back to Homepage