The γ2 Compact High Performance DAC

Specifications and benchmarks

The tests were conducted using the following equipment:
  1. Tektronix TDS2014B 100MHz quad trace digital storage oscilloscope
  2. Fluke 187 digital multimeter
  3. Toshiba Satellite A65 laptop computer with Celeron 2.8GHz CPU and 768MB RAM
  4. M-Audio Firewire Audiophile mobile interface
  5. RightMark Audio Analyzer 5.5 software
  6. other custom test jigs, cables and software
The γ2 under test was populated with AD1896, WM8741, AD8656 (unless otherwise specified), and mated with a γ1 Full++ configuration board. The analog outputs were loaded with 10KΩ resistors.

Please also see the "Specifications" section of the γ1 website for related information.

Input modes
Resolution 16 or 24 bits, auto-detect
Sampling rates Up to 212KHz, auto-detect
Up to 192KHz when used with γ1
Upsampling, all modes Re-sampled to 96KHz
Local master clock
Jitter RMS 0.5pS typical, 1pS maximum
Digital filters
filter A Linear phase, slow rolloff
filter B Minimum phase, slow rolloff
filter C Linear phase, brickwall
Analog filter
All modes Low-pass filter, Fc = 100KHz, 12dB per octave
Anti-clipping mode
Attenuation (off / on) 0dB / -2dB
Digital input
Data format I²S
Connector type Pin headers and receptacles
Analog output
Connector type RCA phono jacks and 3.5mm stereo mini jack
Output Level 1.4Vrms at 0dBFS
Output impedance with OPA2365 9Ω at 1KHz
with AD8656 0.7Ω at 1KHz
Number of channels 2
Frequency response
at -1dBFS		 digital filter C 0.8Hz - 42KHz, +0.02dB, -3.0dB
digital filter B 0.8Hz - 32KHz, +0.02dB, -3.0dB
digital filter A 0.8Hz - 32KHz, +0.02dB, -3.0dB
S/N ratio
at -1dBFS, RMS, 1KHz, A-weighted		 139dB
Noise level
RMS, 20Hz - 20KHz, A-weighted		 106.3dB*
Total harmonic distortion
at -3dBFS, 1KHz		 0.0014%*
Intermodulation distortion + noise
60Hz / 7KHz, 4:1		 0.0033%*
Stereo crosstalk
at -1dBFS, 100Hz - 10KHz		 -103.4dB*
* limited by test equipment.

Note: γ2 will not decode AC3, ADAT or other non-PCM, non-audio data streams.

RMAA test results

Several tests were performed, one with the M-Audio Firewire Audiophile mobile interface running in loopback mode to establish the baseline (lowest limit of measurement resolution). The Firewire Audiophile's S/PDIF output was then connected to the companion γ1 coaxial input. γ2's analog output is fed back to the M-Audio Firewire Audiophile analog input for recording in the remaining tests (for the three digital filter settings).

RMAA loopback testing sets the measurement sound card's ADC sampling mode the same as the digital output. With the γ2, this artificially makes the test results worse for any mode less than 24-bit 96KHz. This is the mode which γ2's DAC operates at regardless of the input stream mode, due to the upsampling. Hence, only the 24-bit 96KHz results are shown here. The "32 bits" results are actually in 24-bit mode but contained in 32-bit words.

Click on the following links to see the results.

Comments on the RMAA test results

γ2's measured frequency response is flat with extended bandwidth and minimal passband ripple (see graph below). The noise floor, THD and IMD distortion products, and stereo crosstalk are all supremely low. In fact, the results are limited by the M-Audio Firewire Audiophile's loopback baseline. Some measurements were actually better than the baseline due to γ2's superior DAC. Indeed, the γ2's performance exceeds that of the measurement instrument, thus the results are arrtificially limited by it. Nevertheless, the numbers compare favorably to other far more expensive high-end DACs.

It is also noteworthy that (in separate tests), changing the ASRC chip to the SRC4192, or changing the DAC chip to WM8742, or changing the opamp to OPA2365 did not produce tangible differences in the RMAA test results.

Comparison of digital filter frequency response

Oscilloscope waveforms

Some oscillogram waveforms of the γ2 DAC are shown below. All input waveforms are digitally generated on the computer. A Tektronix TDS2014B digital storage oscilloscope is used to display the analog line output waveform.

The impulse and square wave response graphs exhibit the Gibbs Phenomenon ringing behavior. It is affected by the choice of digital filter setting, as shown. Please see a discussion about the digital filters in the Tech highlights section.

1KHz sine wave 10KHz sine wave
 
Impulse response (filter C) 1KHz square wave (filter C)
 
Impulse response (filter B) 1KHz square wave (filter B)
 
Impulse response (filter A) 1KHz square wave (filter A)


Main: γ2 Main | Prev: Initial check | Next: History