There is no dedicated line output as such on the SR15, but it doesn’t really matter because there’s a ‘line out’ mode using the regular headphone socket. You enable this in the settings menu and choose the preferred level — 0.7 volts, 1 volt, 1.25 volt and 2 volts are available. Then when you change the volume using the knob, a line out button appears on the screen (in addition to the volume changing normally). Tap the button and the SR15 switches to fixed output mode. (Why those levels? I don’t know. There are intervals of 3dB, 2dB and 4dB between the pairs above.) I measured the line output with the 2 volt setting into a 47.1kHz load, and it came to 2.01 and 1.99 volts RMS for the two channels according to my oscilloscope. In other words, spot on the claim. (Even if the slight divergence is in the unit rather than my measuring rig, it amounts to less than 0.05dB.)
Setting the unit to headphone output mode, playing the same 0dBFS 1kHz sine wave into the same load, with the volume control set to 150 — the maximum — the result was identical. I’m guessing all the line output setting does is fix the output level. In other words, it’s a convenience for use with other audio systems.
Switching from the 47.1kHz load to 295 ohms, there was no clipping even at maximum output, and just about no reduction in output (only in the third significant figure, from 2.01 to 2.00 volts). Into a 15.9-ohm load there was massive clipping. I turned the level from the maximum of 150 down to 130 to eliminate it. You can safely say that playing into just about any headphones at an indicated 130 will guarantee no clipping (although it may well guarantee ear damage!). The output was still 0.565 volts, which is 20 milliwatts into that load. And that works out to 13dB above the sensitivity rating of whatever headphones you’re using.
Switching the load back to open circuit while retaining the same volume level, the output increased to 0.637 volts. From that it was easy enough to calculate an effective internal resistance of pretty much spot on 2 ohms. Perhaps a touch higher than with other high quality portable players, but still sufficiently low that it should result in no frequency response anomalies in any headphones with varying impedances across the audio band.
Sadly, the software I have been using in recent years to conduct other measurements simply refuses to work any more on any of my computers. That means I cannot measure signal-to-noise ratio or distortion, but I’d note that to the extent there is any noise or distortion created by the SR15 (as there must be), it is below any audible levels.
I went old-school on measuring frequency response (above), using pink noise signals and filtering them after recording to remove the 3dB per octave tilt. For 44.1kHz sampling, there’s a straightforward clinical treatment: flat to 20kHz and a sharp filter thereafter. For 96kHz sampling, the treatment is much the same, except that it was flat to 42kHz. The same pattern probably repeated for 192kHz, except that my ADC was softening the curve at the top end. Even then it was flat to 56kHz and down 3dB at 80kHz.
Those who swear by slow, gentle filters are likely to be disappointed. The rest of us can be confident that whatever is up there in the high frequencies, or indeed, suprasonic frequencies, is being delivered.