LABORATORY Test Report

I measured maximum output levels from one of the headphone outputs in standard (unbalanced) mode. Into a high input impedance (47kΩ) there was the tiniest amount of clipping at the 0dB output level at 100Hz, so I backed it down to –1dB and it was clean. The (effectively) no load output was 8.74 volts RMS. Into a 295Ω test load, the output was 8.74 volts RMS. At least, it was with 1kHz and 10kHz test signals. With 100Hz there was a touch of clipping, so I backed it off to –2dB to clean it up. That reduced the output to 7.93V.

That works out to a 213mW output into high-impedance headphones, or some 23dB higher than the sensitivity rating. These days most headphones are at least in the high 90dBSPLs for 1mW of input, so you can expect a good 120dBSPL if you want it. (You shouldn’t want it.)

Into a 15.9Ω load there was clipping again. At –5dB this was completely gone. The output was 5.26 volts for an incredible 1.74 watts output. That’s 32dB higher than the sensitivity rating. If the headphones could cope. They probably couldn’t.

I calculated an internal resistance of around 0.8Ω, so even headphones with a widely varying impedance curve shouldn’t produce major frequency response changes.

As for the line output, if you set the RCA line output to ‘Bypass’—that is, the volume

Graph 3. Frequency response showing effect of FRMP filter setting left channel (white trace) versus right channel (green trace) using 192kHz/24-bit test signal.

Graph 2. Frequency response showing effect of FRMP filter setting (white trace), versus SRMP filter setting (green trace) versus BRCK filter setting (blue trace) vs Apodising (purple trace) using 96kHz/24-bit test signals.

Graph 4. Noise floor spectrum showing noise when battery-powered (white trace) and when mains-powered (green trace) using 96kHz/24-bit test signals.

Graph 1. Frequency response showing effect of FRMP filter setting (white trace), versus SRMP filter setting (green trace) versus BRCK filter setting (blue trace) using 44.1kHz/16-bit test signals.