In an investigation into parasitic oscillation of some headset driver amplifiers, the equivalent circuit of a nominally eight-ohm headset was empirically examined using mimicry. The method is shown in the sketch below:
The resulting RLC model of the headset was the following circuit which has an impedance versus frequency characteristic as shown. The headset impedance was very clearly not just a simple resistance of some nominal value.
The headset driver oscillations were at 1 MHz, correlating approximately with the resonance of the headset itself. The addition of a capacitance, Cshunt, across the headset however, not only lowered the frequency of the headset resonance, but also lowered the peak value of the impedance magnitude.
With 1.0 µF shunted across the headset, the headset resonance was totally suppressed, the impedance bandwidth went to approximately 10 KHz, sufficient for speech service, and the driver oscillations ceased.
Caution .
Some headphone driver IC's also may not "like" the "heavy"
1 uF shunt . I'd recommend a few ohms in series with the 1 uF to dampen those possible effects. In fact, perhaps a smaller cap like 0.1uF and a few ohm series resistor damper could be a good solution. That was done all the time in early 5 watt 12 Volt supply audio power IC's like TDA2002
Otherwise, good story to open up folks eyes to the power of modeling ! Jerry at JDM LABS LLC
Posted by: Jerry Meyerhoff | February 17, 2011 at 10:15 AM
Go for a .1u in series with a 20 ohm. Should take out the resonance and make the driver happy.
Posted by: Clint Brown | February 17, 2011 at 03:36 PM