1
\$\begingroup\$

I made a simple emitter follower headphone amp using fairly high speed BJT- 2SC5994.

enter image description here

Vcc 16V

R1- 10K

R2- 40k

RE- 100

load - 300 ohm

After powering up there it was clear there was serious issue, the amp would pop in and out oscillation randomly. This manifested as buzzing sounds through headphones and after checking with cheap 2mHz oscilloscope there were HF oscillations clearly visible.

They are high speed transistors so this was anticipated, Firstly I added some resistance at the output to isolate from headphone's capacitive load. This seemed to make it less prone to popping into oscillation but it would push up the output impedance too much so wasnt a suitable solution.

Surprisingly adding resistance to the base input ('base stopper') also fixed the problem. With 40 Ohm there hasnt been any oscillation when loaded with HPs and playing test tones of various frequency.

So if it wasnt capactive load on the output causing the issue what else could it be?

\$\endgroup\$
4
  • \$\begingroup\$ Any decoupling capacitor on power supply near collector? | Is input source clean with no load at all? The 40 Ohm loads it. \$\endgroup\$
    – Russell McMahon
    Commented Jan 17, 2020 at 23:22
  • \$\begingroup\$ yes, lots of decoupling and I later added 0.1uf right on the leg of the BJT to see it if it would help with issue. not sure what you mean by is the input source clean with no load \$\endgroup\$
    – Jay
    Commented Jan 17, 2020 at 23:32
  • \$\begingroup\$ if you look at the input signal without the emitter follower attached, what does it look like? Are the grounds hard connected. \$\endgroup\$
    – Russell McMahon
    Commented Jan 18, 2020 at 4:54
  • \$\begingroup\$ What type of coupling capacitors are you using? What is the input signal source impedance? \$\endgroup\$ Commented Jan 18, 2020 at 10:02

5 Answers 5

0
\$\begingroup\$

So if it wasnt capactive load on the output causing the issue what else could it be?

This is serious oscillation, if it can be seen on a 2Mhz-bandwidth 'scope. My numerous encounters with similar circuits oscillate at very-high or ultra-high frequency. The audible "pops" and "buzzes" suggest oscillating amplitude is large enough to seriously modulate audio. Less-serious oscillations can lurk under the audible-radar. Check carefully that oscillations have really been snubbed by your efforts, and not simply moved to a much-higher frequency.

Oscillation is often the result of some inductance resonating with small transistor capacitances. But oscillation can also result from a ground loop feeding current back into earlier gain stages that drive this circuit. Lower-frequency oscillation (rather than VHF or UHF oscillations) make ground-loop a serious possibility. Re-wiring the headphone return current to a different circuit ground can reduce or eliminate such oscillations. Substituting a lower-frequency \$ F_T \$ transistor won't likely cure a ground-loop oscillator, just lower its frequency.

In search of inductance

If a small series base resistor cures oscillation, this is more likely a local LC oscillator rather than a ground-loop oscillator. Where could the "L" of an LC oscillator be hiding? An oscillator in the MHz region requires significant L.
Possibly the headphone itself. It is a coil of wire. Or possibly the headphone cable. It is a long transmission line that appears inductive at some frequencies.
Possibly the inductance of the wire back to the DC supply. A bypass capacitor from transistor collector to ground might bypass this inductance. But be careful where the capacitor ground is grounded - you can add a ground-loop if it intercepts headphone current with the wrong phase. Where \$ R_E \$ is grounded might be the right spot.

BTW: a series base resistor less than about 100 ohms seems a decent solution.

\$\endgroup\$
2
\$\begingroup\$

Yes, adding a base resistor (usually a 50 - 1000\$\Omega\$) is one solution to RF oscillation of an emitter follower caused by capacitive loading. It doesn't take much capacitance, tens or hundreds of pF can be enough, which is not much length of cable.

Often a ferrite bead is used, either in the base or emitter lead because it doesn't degrade the high frequency performance as much as a resistor. You don't need it for audio.

See also Brian Drummond's answer here

Your actual question is "what else could it be?". I think it's the most likely cause, and I can't think of anything else that would be improved by the 'base stopper' resistor.

\$\endgroup\$
4
  • \$\begingroup\$ the thing is that the oscillation comes and goes depending on orientation of headphones, and could be stopped by touching the cable or headphones. if the source was circuit layout or cable capacitance you would assume it would be constantly/consistently oscillating since these values would not change, or is that wrong? \$\endgroup\$
    – Jay
    Commented Jan 18, 2020 at 1:59
  • \$\begingroup\$ You’re affecting the output capacitance by moving the cable around and touching it, so yes that fits exactly, particularly for an unshielded cable like a headphone. \$\endgroup\$ Commented Jan 18, 2020 at 2:24
  • 1
    \$\begingroup\$ Think of it as a little accidental theremin. \$\endgroup\$ Commented Jan 18, 2020 at 2:53
  • 1
    \$\begingroup\$ Love the "accidental theremin"! \$\endgroup\$
    – user16324
    Commented Jan 18, 2020 at 12:29
1
\$\begingroup\$

Any common collector ( emitter follower) or FET source follower with a reactive load where the slightest resonant gain is >1 can result in emitter oscillation with any wire L and C, capacitance loads.

This can be avoided by attenuating the loop gain, improving phase margin or matching emitter towards load impedance. This may be done by;

  • adding either lowering the input shunt impedance to hundreds of Ohms
  • or adding at least 22 Ohms series R to the emitter to closer match the emitter to the transmission line
  • or add a ferrite R which adds similar values at unity gain BW freq.
  • by minimizing C(Vbe) copper coupling capacitance in the layout

e.g. Design by impedance

\$\endgroup\$
2
  • \$\begingroup\$ 'adding either lowering the input shunt impedance to hundreds of Ohms' this would mean lowering values of R1/R2 divider? I currently have no input shunt resistor (Ri above) before the input cap, \$\endgroup\$
    – Jay
    Commented Jan 18, 2020 at 2:04
  • \$\begingroup\$ no . R1,R2 can be 50x Re Values. Raising Ze(f) with small series R or ferrite bead is most common solution to drive cable \$\endgroup\$ Commented Jan 18, 2020 at 2:35
0
\$\begingroup\$

It is probably layout related. If you have inductance on the collector it tends to turn the circuit into a Colpitts oscillator. By adding the resistor you are damping the loop gain below 1.

\$\endgroup\$
0
\$\begingroup\$

Your schematic is missing important power supply decoupling capacitors close to the transistor. 0.1uF ceramic (marked 104) for VHF plus 10uF electrolytic for audio.

\$\endgroup\$
1
  • \$\begingroup\$ This does not answer the question directly. If you believe it does, please indicate why decoupling is a possible cause of his oscilations in this particular circuit. If not, this is at best a comment and not an answer. \$\endgroup\$
    – efox29
    Commented Jan 20, 2020 at 5:03

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.