I have a phase shift oscillator circuit from a drum machine that I would like to reproduce as faithfully to the original as possible. My problem is that the 40 year old design uses a transistor (2N3394) that is no longer available. But... the others from the same data sheet (2N3391, 92 & 93) still are. The only difference between them is that the hFE is higher than the original.

Here is the circuit (as it is presented in the original schematic):


simulate this circuit – Schematic created using CircuitLab

Note: I added S1 and the pulldown resistor (R11) to replace the original trigger. The output was originally fed to a summing bus, and then amplified.

The Intention: At this point, the plan is to faithfully reproduce the circuit on a breadboard (or PCB) and experiment with modification options and values without damaging the original device. I am considering a) only modifying the original drum machine; b) re-creating the drum machine with the modifications; or c) re-creating a few of the individual sounds as single units, and providing amplification for each one, rather than summing to a mono output.

So the main question is: Would replacing Q1 with a higher hFE transistor have any bearing on either the decay time, amplitude or frequency of the output?

Secondary to that is: When seeking out a replacement, what specifications, if any, have a bearing on those same parameters?

Additional Note: While composing this question, I reproduced this circuit using the Falstad Simulator, and found that even changing the hFE within the range of the original part (hFE=55-110) presented major differences in the output. But… from what I understand, the hFE can vary greatly within the given range, and is unreliable. This causes me to question the results I am seeing. If this were not the case, I would have taken the simulation result for my answer.

Here is a snapshot of the output at the C6/R9 junction:

Phase Shift Oscillator Output hFE=55-110

Left: output at hFE=55. Right: output at hFE=110

The waveform result at hFE=55 seems to best represent what I hear in so far as the "sharp impact" which fades out, and also looks like the waveform image presented in the original schematic - down to the phase and number of peaks.

Schematic Source: Here is a link to the service manual that I referenced this design from. It's difficult to read due to the low quality of the scan, but the values in my own schematic were taken from a hard copy.

  • \$\begingroup\$ You say that it was "originally fed to a summing bus." Are you still planning on doing that? Or are you planning something different for the output, now. Are you still planning on the same 14 V rail? Or is that to be changed? Etc. Can you discuss your plans in more detail (and not just the prior circuit and its simulation details?) I'm curious. \$\endgroup\$
    – jonk
    Commented Jun 26, 2017 at 19:30
  • \$\begingroup\$ By the way, this page may help you a lot on understanding (and answering your own question) the issue about the oscillation frequency for your setup: electronics-tutorials.ws/oscillator/rc_oscillator.html (Note that this is where \$\beta\$ may affect things as it affects the base loading on the last stage.) You can do the computation on that web site assuming the loading works out to a final resistor value that is also the same and you will get an oscillation rate that is close to your graph. \$\endgroup\$
    – jonk
    Commented Jun 26, 2017 at 19:58
  • \$\begingroup\$ beta is pretty important for this circuit Q1 gets biased via C2 & R6, so the discharge time of c2 will vary with beta \$\endgroup\$
    – sstobbe
    Commented Jun 26, 2017 at 21:22
  • \$\begingroup\$ You can easily adjust the duration of the oscillation by changing the value of \$C_2\$. Larger values will lengthen the duration by holding the base of \$Q_1\$ up longer. \$C_2\$ is jacked up by the switch and raises the base of \$Q_1\$ from "just barely off" to "just barely on" so that it can provide higher current gain (low gain at low collector currents) for the oscillation. Subtle design element. I'm growing to like it more and more. (Just dawned on me. But I'm a slow-minded hobbyist and that schematic layout could be improved to make this fact more obvious.) \$\endgroup\$
    – jonk
    Commented Jun 27, 2017 at 6:44
  • \$\begingroup\$ Also, the connection of \$R_4\$ and \$R_5\$ to a low-impedance, nearly in-phase copy of the base signal significantly aids the oscillation when started and does so without adding components. Nice. \$\endgroup\$
    – jonk
    Commented Jun 27, 2017 at 7:05

1 Answer 1


The classic 3 section phase shift oscillator requires a voltage gain of 29 to oscillate continually. Of course you do not want continual oscillations since you need a damped sine wave out.

The Hfe of the transistor does not directly affect the voltage gain but will affect the input impedance which will affect the loss in the phase shift network and thus the Q of the resultant circuit.

Your circuit is slightly unusual in that the emitter resistor is not bypassed which should introduce negative feedback (emitter degeneration) and stabilize the circuit with respect to Hfe variations.

I'm surprised that R4 and R5 return to the emitter and not to ground - it's not immediately obvious to me what effect that has.

  • \$\begingroup\$ I agree to the unusual nature of the circuit - but only because I cannot find any similar arrangements as of yet. Being that ß does seem to be the issue at the moment, are there any further parameters I should try harder to match when choosing a replacement? \$\endgroup\$
    – Jay
    Commented Jun 27, 2017 at 14:45
  • \$\begingroup\$ Are you sure that R7 = 68 k ... and not 680 k ? \$\endgroup\$
    – Antonio51
    Commented Jul 11, 2021 at 7:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.