I just took some pictures of actual AKG K550 listening level voltages, for anyone wanting to see some details. Instead of just sine wave testing I'm using a DMM that records maximum, average, and minimum true rms voltage levels, then playing an entire song to capture those levels after a reset. I figure that gives a better representation of the voltage levels involved in listening to real world music on the phones. I've run the measurements two ways here, at normal listening levels then again at a level that I could only keep the headphones on for about 1-2 seconds safely.
Here is the test setup. Just a laptop with a breakout cable I made, going from the headphone jack to the phones, feeding a meter with full audio AC bandwidth capability and a scope. The scope is there just to verify no clipping occurring in the laptop's audio output stage while driving the headphones. Not likely at these low voltages and currents, but one never knows. The laptop is running on batteries to avoid any ground loops with the test equipment:
You can see from the photo here that the PC volume slider is somewhere between 3/4 and 7/8 (81/100 on the scale) for normal, comfortable listening levels on the AKG K550 headphones with this particular source material. Over the course of a song the meter from the photo recorded true rms (AC coupled) values of: maximum = 48.2mV, average = 29.7mV, and minimum = 17.4mV. Since these are true rms voltages and the impedance of the AKG K550s is very flat vs. frequency, the power input to the phones (per channel) is just E^2/R: maximum = [(48.2mV)^2 / 32R] = 72.6uW (yes that is MICRO watts!), average = 27.6uW, minimum = 9.5uW.
And from the photo here, with the volume slider all the way up and the phones too loud for more than 1-2 seconds use, the meter recorded over the course of the same song: maximum = 147.1mV, average = 83.0mV, and minimum = 13.9mV. Note that maximum voltage level on peaks, 147mV, is actually exceeding the "94 dB 135mV hearing damage level" on the box for extended listening. So this one can be considered an absolute maximum set of numbers. Power input to the phones is: maximum = 676uW = 0.676mW, average = 215uW = 0.215mW, minimum = 6.0uW.
So for normal listening levels, the 50mV rms number I posted previously, above, is more of a maximum. For this particular song the average is around 30mV rms! I'm amazed these phones put out this amount of sound with so little input voltage. Some pretty amazing sensitivity.
I will also have to modify what I posted above about an amp used with an iphone needing any voltage gain. At those super-low drive voltage levels I would say it is unlikely the iPhone - or any other device - couldn't muster 50mV (rms) maximum. Instead I'll go back to what I initially posted - what is needed is most likely a unity gain (voltage gain = 1) current buffer. In other words, the problem isn't that the iPhone can't muster enough voltage swing for the AKG K550s, but rather it is much more likely that the iPhone can't supply even the small 50mV / 32R = 1.56mA (rms) maximum current requirement. If the device's output stage can't supply enough current like this, it will drag the output voltage down (forms a resistive voltage divider with the headphones) and the net result is not enough maximum volume.
So, in summary, back to suggesting an amp where the voltage gain can be set to 1 as purely a current buffer.
Edited by agdr - 9/8/12 at 5:33pm