My problem is: When drawing significant current (above 1mA) there is a very disturbing crosstalk between both channels (above -30 dB). This can even be heard very easily when the signal is only on the left or right. Additionally this crosstalk shows a very strong distortion. (THD >> 10%).
The load itself is about 8 ohms (on JP7 and JP9). The 33 ohms series resistor is used to get a current driven behavior (previously 100 ohms were considered, but because of maximum about 2 volts voltage amplitude the resistance had to be reduced to 33 ohms to meet output power requirements).
The input signal (on Pin 1 of JP6 and JP8) has a source impedance of about 10 kOhms to AGND.
The two channels are only coupled via the supply voltages +5V, GND and AGND. (AGND is a pseudo supply voltage generated by another AD8656 and a voltage divider stabilized by a 10uF capacitor on the input side of the opamp)
Is this behavior of the AD8656 in the output amplifier ok? Or is it defect? If not, how could this be optimized?
Thanks for your advice and effort!
In the datasheet of the AD8656 in figure 53 the crosstalk between the two channels is tested, BUT with no load on each of the outputs. So i don't think these results are applicable to my problem.
I reproduced the measure environment of figure 53 in the datasheed by simply increasing the load resistance. The crosstalk immediatly disappeared. So there is a correlation between current draw and crosstalk / distortion of the crosstalk.