# Why does my BA speaker truncate the negative part of the waveform?

I'm developing a custom earphone and I'm having some issues with producing a clean sound from my balanced armature (BA) driver (Sonion 31D005/8). I've come up with a pretty standard circuit (see below) that receives a sine wave at three possible frequencies (227,660,1000) from a 12-bit DAC.

When I probe the output of capacitor C2 without the speaker connected, I see a clean wave with a zero volt average, however as soon as I plug in the BA the peak of the negative portion is truncated, and the voltage (p-p) is reduced (see below). This outcome is more prominent for the lower frequencies, ultimately reducing the loudness of my speaker for lower frequencies, contrary to what's shown on the datasheet (should remain relatively constant).

Another important factor is that the BA's impedance decreases for lower frequencies (500Hz:38 ; 1000Hz:66) so I expected the opposite effect, however I'm assuming there's some time constant effect or an inductive feedback that is distorting the waveform. I've confirmed with a microphone that the received waveform is quite distorted in a similar way, and the lower frequencies do not sound very clear. Am I missing a passive component or something that should be considered when designing headphone (high impedance) as opposed to speaker (4-8 ohm) circuits? Perhaps a different amplifier (tried the LT1010 with worse performance)?

• What voltages are powering the amplifier? Single rail or dual rail? – Transistor Jul 23 at 11:23
• How much current can an OPA344 deliver? What does the impedance curve for your speaker and C2 look like? What will be the current through the speaker given the impedance and the supply voltage? – JRE Jul 23 at 12:15
• The amplifier is single rail, however the voltage is above zero until after the C2 capacitor. I'm not sure how to determine the OPA344 current from the datasheet but op-amps commonly supply 10's of milliamps, and I'm well below that for a Balance Armature. I haven't been able to measure it directly but that'll be my next step. The impedance curve for a similar speaker (that I also tried) can be found here since the Sonion one doesn't have a curve in the datasheet. – meekeech Jul 23 at 19:15