Links: reviewed by Mark Phillips on *SoundStage! Solo* on February 1, 2021

**General information**

All measurements taken using an Audio Precision APx555 B Series analyzer.

The Coda was conditioned for 30 minutes at 2Vrms at the output before any measurements were taken.

The Coda offers a USB Type A input and a 3.5mm TRS headphone output. The Coda also offers three digital filters: minimum phase, linear phase, and hybrid. Unless otherwise stated, the minimum-phase filter was used for our primary measurements and graphs. The Coda and the Audio Precision analyzer were connected to the same Microsoft Surface Pro 6 laptop via USB Type-A connection. The Coda also offers a volume control; however, volume changes are not performed by the Coda internally, but rather by the computer and its operating system in response to commands sent by the Coda. Volume was maintained at maximum for all measurements.

**Published specifications vs. our primary measurements**

The table below summarizes our primary measurements performed on the Coda. Here we can compare directly against Clarus’s own published specifications for the Coda, which are stated as follows:

- Maximum output voltage: 2.0Vrms
- Output impedance: <1 ohm
- Input impedance: MM/MC HIGH: 47k ohms, MC LOW: 1k ohms, MC VERY LOW: 110 ohms
- SNR: 120dB
- THD+N: -112dB (~0.0003%)

Our primary measurements revealed the following using the USB input (unless specified, assume a 0dBFS 1kHz sinewave input, 2Vrms output, 300-ohm loading, 10Hz to 90kHz bandwidth, minimum-phase filter):

Parameter | Left channel | Right channel |

Crosstalk, one channel driven (10kHz, 44.1kHz and 96kHz sample rate) | -69dB | -69dB |

Dynamic Range (A-weighted, 44.1kHz) | 101.6dB | 101.6dB |

Dynamic Range (A-weighted, 96kHz) | 104.8dB | 104.8dB |

IMD ratio (18kHz and 19kHz stimulus tones, 44.1kHz sample rate) | <-94dB | <-94dB |

IMD ratio (18kHz and 19kHz stimulus tones, 96kHz sample rate) | <-93dB | <-93dB |

Maximum Output Voltage (0dBFS) | 2.06Vrms | 2.05Vrms |

Maximum output power into 600 ohms (0dBFS) | 7.15mW | 7.07mW |

Maximum output power into 300 ohms (0dBFS) | 14.20mW | 14.04mW |

Maximum output power into 32 ohms (1% THD+N @-2.2dBFS, unweighted) | 74.04mW | 72.67mW |

Output impedance | 1.56 ohms | 1.56 ohms |

Noise level (A-weighted, 44.1kHz) | <19uVrms | <19uVrms |

Noise level (unweighted, 44.1kHz) | <40uVrms | <40uVrms |

Noise level (A-weighted, 96kHz) | <6uVrms | <6uVrms |

Noise level (unweighted, 96kHz) | <25uVrms | <25uVrms |

THD ratio (unweighted, 44.1kHz) | <0.0012% | <0.0013% |

THD+N ratio (A-weighted, 44.1kHz) | <0.0015% | <0.0018% |

THD+N ratio (unweighted, 44.1kHz) | <0.0022% | <0.0023% |

THD ratio (unweighted, 96kHz) | <0.0013% | <0.0015% |

THD+N ratio (A-weighted, 96kHz) | <0.0013% | <0.0015% |

THD+N ratio (unweighted, 96kHz) | <0.0018% | <0.0019% |

The Clarus Coda’s maximum output voltage of 2.0Vrms was corroborated, where we measured 2.05Vrms for a 0dBFS input signal into a 300-ohm load.

Clarus’s claim of a 1-ohm output impedance was not quite corroborated by our measured 1.56-ohm output impedance, however, this is still very low and desirable for a headphone amplifier.

The signal-to-noise ratio (SNR) claim of 120dB (we are assuming A-weighted) was not corroborated by our dynamic range measurements of 105dB (A-weighted, 96kHz sample rate). For a DAC, dynamic range is a more appropriate measurement than SNR, because in some DACs, the output of the device is switched off when there is no signal. The Audio Precision uses the AES17 method for dynamic-range measurements, whereby a 0dBFS 1kHz sine wave input signal is applied, and the noise measurement is performed using a -60 dBFS stimulus, which is then notched out. That said, when we measured SNR (A-weighted) with the Coda, we measured slightly worse (lower) values compared to our dynamic-range measurements, though the values were still commendably high.

The THD+N claims from Clarus were not corroborated into a 300-ohm load at 0dBFS for a 1kHz input signal, where we measured (A-weighted, 96kHz sample rate) at worst 0.0015% (-98dB, shown in table above) and at best 0.0008% (-102dB, from our swept THD versus frequency graph shown below) compared to Clarus’s claim of 0.00025% (-112dB). However, as can also be seen in the THD versus frequency graph below, between 300 and 500Hz, into a 32-ohm load at -6dBFS, the Clarus did achieve 0.0003% (-110dB) THD, which is very low and extremely close to the specified value by Clarus.

**Frequency response (44.1, 96, and 192kHz sample rates)**

In our measured frequency-response plots above for a 0dBFS input signal sampled at 44.1kHz (blue/red = left/right channels), 96kHz (purple/green), and 192kHz (orange/pink) into a 300-ohm load, the Coda deviates less than +/-0.1dB from flat from 5Hz to 20kHz (with the exception of the 44.1kHz data). The -3dB points are at 17kHz (44.1kHz sample rate), 38kHz (96kHz), and 72kHz (192kHz). In the graph above and some of the graphs below, there may be two visible traces representing the left channel (blue, purple, or orange traces) and the right channel (red, green, or pink traces). On other graphs, only one trace may be visible. When there is only one trace visible, it is because the left and right channels are tracking extremely closely, so as not to show a difference with the chosen axis scales. When channels are not matched as well, two traces become visible.

**Frequency response (600-, 300-, 32-ohm loads, 96kHz sample rate)**

The chart above shows RMS level (relative to 0dBrA, which here is 1Vrms or -6dBFS at the input) as a function of frequency for the left channel only. The blue plot is into a 600-ohm load, the purple is into a 300-ohm load, and the orange is into a 32-ohm load. Here we find that all plots are very closely grouped together (within about 0.3dB), which is an indication of a very low output impedance.

**Frequency response (linear-phase, minimum-phase, hybrid filters with 96kHz sample rate)**

The plots above show frequency-response for a 0dBFS input signal sampled at 96kHz for the minimum phase filter (blue), the linear phase filter (purple), and the hybrid filter (orange) into a 300-ohm load for the left channel only. We can see that the hybrid filter provides the most “brick-wall” type response (*i.e.*, a very steep rolloff), and the linear phase filter shows the earliest attenuation, visible above 10kHz. The -3dB points are at 39kHz (minimum phase), 36kHz (linear phase), and 40kHz (hybrid).

**Frequency response (linear-phase, minimum-phase, hybrid filters with 44.1kHz sample rate)**

The chart above shows the frequency-responses for a 0dBFS input signal sampled at 44.1kHz for the minimum-phase (blue line), the linear-phase (purple), and hybrid (orange) filters into a 300-ohm load for the left channel only. The graph is zoomed in from 1kHz to 22kHz, and extra sampling points were introduced around the corner frequency (the “knee”) to highlight the various responses of the three filters. As with the 96kHz chart above, we can see again that the hybrid filter provides the most “brick-wall”-type response, and the linear phase filter shows the earliest attenuation, visible above 7kHz. The -3dB points are at 19.2kHz (minimum phase), 19kHz (linear phase), and 21kHz (hybrid).

**THD ratio (unweighted) vs. frequency (600-, 300-, 32-ohm loads)**

The chart above shows THD ratio as a function of frequency into 600-ohm (blue/red), 300-ohm (purple/green), and 32-ohm (orange/pink) loads for a -6dBFS input signal (1Vrms output). THD values at 20Hz are just below 0.001% for all three loads. The 600- and 300-ohm THD data match very closely, where it’s mostly constant up to 3kHz, then there is a rise to 0.004% at 20kHz for the 600-ohm load, and 0.005% for the 300-ohm load. Between 50Hz and 1kHz, the 32-ohm load THD values are demonstrably lower (by almost 10dB between 300-500Hz) than the 600- and 300-ohm data. However, past 2kHz, the 32-ohm load THD data rises up past 0.01% between 10kHz and 20kHz. Interestingly, there’s a rise in THD at 200Hz and 2kHz for all data sets. This phenomenon was repeatable across several measurements over several days.

**THD ratio (unweighted) vs. output power (600-, 300-, 32-ohm loads)**

Above we can see a plot of THD ratios as a function of output power into 600-ohm (blue/red), 300-ohm (purple/green), and 32-ohm (orange/pink) loads for a 1kHz signal sampled at 96kHz swept from -60dBFS to 0dBFS. The 600- and 300-ohm plots are very similar (600-ohm performed 2-4dB better), showing THD ratios from about 0.5% at 20nW, down to near 0.001% at 5-10mW (0dBFS). The 32-ohm plot is near 1% just above 100nW, then dips down to near 0.001% at 50mW (about -4dBFS input signal), then a sharp rise in THD as the Coda’s output is no longer able to sustain the required current. The 1% THD value is reached just past 70mW.

**THD+N ratio (unweighted) vs. output power (600-, 300-, 32-ohm loads)**

Above is a chart of THD+N ratios as a function of output power into 600-ohm (blue/red), 300-ohm (purple/green), and 32-ohm (orange/pink) loads for a 1kHz signal sampled at 96kHz swept from -60dBFS to 0dBFS. The 600 and 300-load plots are very similar (600-ohm performed 2-3dB better), showing THD+N ratios from about 1.5% at 20nW, down to 0.002% at 5-10mW (0dBFS). The 32-ohm plot is near 1% just above 100nW, then dips down to near 0.002% at 50mW (about -4dBFS input signal), then a sharp rise in THD as the Coda’s output is no longer able to sustain the required current. The 1% THD+N value is reached just past 70mW. The Coda is a low-noise device, and because of this the THD+N and THD (above) charts are similar (*i.e.*, THD ratios dominate above the noise).

**Digital linearity (44.1kHz and 96kHz sample rates)**

The plot above shows the results of a linearity test. For this test, the digital input to the Coda is swept with a dithered 1kHz input signal from -120dBFS to 0dBFS, and the output is analyzed by the APx555. The ideal response would be a straight flat line at 0dB down to the lowest possible level (-120dBFS). The results were identical for both 44.1kHz and 96kHz sample rates. The Coda approaches the ideal 0dB relative level just below -90dBFS, then yielding perfect results from -85dBFS to 0dBFS. At -120dBFS, both channels overshot the ideal output signal amplitude by about 25dB.

**FFT spectrum – 1kHz (0dBFS, 44.1kHz sample rate)**

Shown above is a fast Fourier Transform (FFT) of a 1kHz input sinewave stimulus at 0dBFS sampled at 44.1kHz, which results in the reference output voltage of 2Vrms into a 300-ohm load. Here we see the worst-case signal harmonic (2kHz) at -100dBrA. The third-order harmonic (3kHz) is at about -115dBRa. The other peaks (hovering just below -120dBrA) seen to the left and to the right of the primary 1kHz signal peak are likely digital-filter aliasing artifacts.

**FFT spectrum – 1kHz (0dBFS, 96kHz sample rate)**

Shown above is an FFT of a 1kHz input sinewave stimulus at 0dBFS sampled at 96kHz, which results in the reference output voltage of 2Vrms into a 300-ohm load. Here we see the worst-case signal harmonic (2kHz) at -100dBrA. The third- and fifth-order harmonics (3 and 5kHz) are at about -115dBRa. Beyond the second signal harmonic, the odd order harmonics (3, 5, 7kHz, etc.) are above -120dBrA and dominate the even-order harmonics (4, 6, 8kHz, etc.), which are below -120dBrA.

**FFT spectrum – 1kHz (-90dBFS, 44.1kHz sample rate)**

Shown above is an FFT of a 1kHz input sinewave stimulus at -90dBFS sampled at 44.1kHz, into a 300-ohm load. Here we see the worst-case signal harmonic (5kHz) at -100dBrA, with the 3kHz harmonic at -110dBrA. Most of the remaining peaks are likely digital-filter aliasing artifacts, ranging in level from -110dBrA to below -130dBrA.

**FFT spectrum – 1kHz (-90dBFS, 96kHz sample rate)**

Shown above is an FFT of a 1kHz input sinewave stimulus at -90dBFS sampled at 96kHz, into a 300-ohm load. The worst-case signal harmonic (5kHz) is at -100dBrA, with the 3kHz harmonic at -110dBrA. Like the 0dBFS FFT, the odd-order signal harmonics dominate over the even order harmonics, which are below -130dBrA. The majority of the remaining peaks are likely digital-filter aliasing artifacts, which range in level from -110dBrA to below -130dBrA.

**Intermodulation distortion FFT (18kHz + 19kHz summed stimulus, 0dBFS, 44.1kHz sample rate)**

Shown above is an FFT of the intermodulation (IMD) products for an 18kHz + 19kHz summed sinewave stimulus tone measured at the output into a 300-ohm load, sampled at 44.1kHz. The input values are set at -6.02dBFS so that, if summed for a mean frequency of 18.5kHz, would result in a 0dBFS signal, or 2Vrms (0dBrA), at the output. We found that the second-order modulation product (*i.e.*, the difference signal of 1kHz) is at -105dBrA, while the third-order modulation products, at 17kHz and 20kHz, are at -115dBrA.

**Intermodulation distortion FFT (18kHz + 19kHz summed stimulus, 0dBFS, 96kHz sample rate)**

Above is an FFT of the intermodulation (IMD) products for an 18kHz + 19kHz summed sinewave stimulus tone measured at the output into a 300-ohm load, sampled at 96kHz. The input values are set at -6.02dBFS so that, if summed for a mean frequency of 18.5kHz, would result in a 0dBFS signal, or 2Vrms (0dBrA), at the output. The second-order modulation product (*i.e.*, the difference signal of 1kHz) is just below -100dBrA, while the third-order modulation products, at 17kHz and 20kHz, are at -115dBrA. The other peaks at 2, 3, 4, 5kHz, etc. are likely signal harmonics of the second-order 1kHz modulation product.

**Wideband FFT spectrum of white noise and 19.1kHz sinewave tone (44.1kHz sample rate)**

The chart above shows an FFT of the Clarus’s output with white noise at -4dBFS (blue/red) and a 19.1kHz sinewave signal at 0dBFS (purple/green), sampled at 44.1kHz. The roll-off above 20kHz in the white-noise spectrum shows the implementation of the reconstruction filter. The significant peak at 25kHz (-35dBrA) is an aliased image due to the digital-filter implementation (as with all charts above unless noted, this was with the minimum-phase filter). The second-, third-, and fourth-order signal distortion harmonics (38.2, 57.3, 76.4kHz) of the 19.1kHz tone are much lower in amplitude, at around -95dBrA. Several other lower-level aliasing artifact peaks can also be seen at around -130dBrA. With the filter set to linear phase as well as hybrid (both not shown), the slope in the roll-off at around 20kHz in the noise spectrum was sharper (reflecting what the frequency response with the three filter settings sampled at 44.1kHz graph shown above demonstrates) and the main aliasing peak at 25kHz was lower in amplitude, at -90dBrA.

*Diego Estan*

Electronics Measurement Specialist