Roksan Attessa Streaming Integrated Amplifier Measurements

Link: reviewed by Vince Hanada on SoundStage! Simplifi on September 15, 2022

General Information

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

The Roksan Attessa streaming integrated amplifier was conditioned for 1 hour at 1/8th full rated power (~10W into 8 ohms) before any measurements were taken. All measurements were taken with both channels driven, using a 120V/20A dedicated circuit, unless otherwise stated.

The Attessa offers two unbalanced line-level analog inputs (RCA), one unbalanced moving-magnet (MM) phono input, two coaxial (RCA) and two optical (TosLink) S/PDIF digital inputs, one pair of line-level pre-outs (RCA), and two pairs of speaker-level outputs. On the front of the unit is a 1/8″ TRS headphone output. For the purposes of these measurements, the following inputs were evaluated: digital coaxial, and the analog line-level and MM unbalanced inputs.

The Atessa offers different gain settings (Low, Mid, High) for both the line-level and phono analog inputs. Unless otherwise stated, the default settings of Low for the line-level input and Mid for the phono input were used.

Most measurements were made with a standard 2Vrms line-level analog input, 5mVrms MM input, and 0dBFS digital input. The signal-to-noise ratio (SNR) measurements were made with the same input signal values, but with the volume set to achieve the rated output power of 80W (8 ohms). For comparison, on the line-level input, a SNR measurement was also made with the volume at maximum, where only 0.45Vrms was required to achieve 80W into 8ohms.

Based on the accuracy and non-repeatable results at various volume levels of the left/right channel matching (see table below), the Attessa volume control is likely digitally controlled but operating in the analog domain. The volume control offers a total of 40 steps using 20 LEDs as indicators for positioning. These indicators will either glow dimly or brightly, depending on voluem position. The volume range measured from -40.1dB (position 1) to +34.9dB (maximum) measured between the line-level analog input and the speaker outputs, in increments of 3dB below step 6, 2dB from 6 to 28, 1dB from 29 to 36, then back to 2dB increments from 37 to 40.

Volume-control accuracy (measured at speaker outputs): left-right channel tracking

Published specifications vs. our primary measurements

The table below summarizes the measurements published by Roskan for the Attessa compared directly against our own. The published specifications are sourced from Roskan’s website, either directly or from the manual available for download, or a combination thereof. 

Our primary measurements revealed the following using the line-level analog input and digital coaxial input (unless specified, assume a 1kHz sine wave at 2Vrms or 0dBFS, 10W output, 8-ohm loading, 10Hz to 90kHz bandwidth):

For the continuous dynamic power test, the Attessa was able to sustain 139W into 4 ohms (~1% THD) using an 80Hz tone for 500ms, alternating with a signal at -10dB of the peak (13.9W) for 5 seconds, for 5 continuous minutes without the protection circuit shutting down the unit. This test is meant to simulate sporadic dynamic bass peaks in music and movies. During the test, the top of the Attessa was warm to the touch, but not enough to induce pain.

Our primary measurements revealed the following using the phono-level input, MM configuration (unless specified, assume a 1kHz 5mVrms sine wave input, 10W output, 8-ohm loading, 10Hz to 90kHz bandwidth):

Our primary measurements revealed the following using the analog input at the headphone output (unless specified, assume a 1kHz sine wave, 2Vrms output, 300 ohms loading, 10Hz to 90kHz bandwidth):

Frequency response (8-ohm loading, line-level input)

In our measured frequency-response chart above, the Attessa is nearly flat within the audioband (+0.16dB at 20Hz, -0.44dB at 20kHz). At the extremes, the Attessa is +3dB at 5Hz and -6.3dB at 80kHz. The -3dB point was measured around 54kHz. In the graph above and most of the graphs below, only a single trace may be visible. This is because the left channel (blue or purple trace) is performing identically to the right channel (red or green trace), and so they perfectly overlap, indicating that the two channels are ideally matched.

Phase response (8-ohm loading, line-level input)

Above are the phase response plots from 20Hz to 20kHz for the line level input, measured across the speaker outputs at 10W into 8 ohms. The Attessa does not invert polarity and exhibits, at worst, less than 30 degrees (at 20kHz) of phase shift within the audioband.

Frequency response vs. input type (analog and 16/44.1, 24/96, and 24/192 inputs with 8-ohm loading, left channel only)

The chart above shows the Attessa‘s frequency response as a function of input type measured across the speaker outputs at 10W into 8 ohms. The green trace is the same analog input data from the previous graph. The blue trace is for a 16-bit/44.1kHz dithered digital input signal from 5Hz to 22kHz using the coaxial digital input, and the purple trace is for a 24/96 dithered digital input signal from 5Hz to 48kHz, while the pink trace is for a 24/192 dithered digital input signal from 5Hz to 96kHz. The 16/44.1 data does not exhibit brick-wall type filtering, with a -3dB point at 20.5kHz. The 24/96 and 24/192 kHz data both yielded -3dB points at 31.1kHz. The analog input shows more extended response at high frequencies than the 24/192 input. There’s no evidence that the Attessa digitizes incoming analog signals.

Frequency response (8-ohm loading, MM phono input)

The chart above shows the frequency response for the MM phono input. What is shown is the deviation from the RIAA curve, where the input signal sweep is EQ’d with an inverted RIAA curve supplied by Audio Precision (i.e., zero deviation would yield a flat line at 0dB). We see a maximum deviation of about +1dB at 20Hz and -0.5dB at 20kHz, from 20Hz to 20kHz.

Phase response (MM input)

Above is the phase response plot from 20Hz to 20kHz for the MM phono input, measured across the speaker outputs at 10W into 8 ohms. The Attessa does not invert polarity. For the phono input, since the RIAA equalization curve must be implemented, which ranges from +19.9dB (20Hz) to -32.6dB (90kHz), phase shift at the output is inevitable. Here we find a worst case of about -60 degrees at 200Hz and 5kHz.

Digital linearity (16/44.1 and 24/96 data)

The chart above shows the results of a linearity test for the coaxial digital input for both 16/44.1 (blue/red) and 24/96 (purple/green) input data, measured at the line level pre-outs of the Attessa for a 1Vrms (at 0dBFS) output signal. The digital input 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. Both data were essentially perfect as of -100dBFS up to 0dBFS. At -120dBFS, the 16/44.1 data were about +2dB above reference, while the 24/96 data were within 1dB of reference. This is a good linearity test result.

Impulse response (24/44.1 data)

The graph above shows the impulse response for a looped 24/44.1 test file that moves from digital silence to full 0dBFS (all “1”s) for one sample period then back to digital silence, measured at the line-level pre-outs of the Attessa. We can see that the Attessa DAC utilizes a reconstruction filter that has less pre-ringing than post-ringing.

J-Test (coaxial)

The chart above shows the results of the J-Test test for the coaxial digital input measured at the line level pre-outs of the Attessa. J-Test was developed by Julian Dunn in the 1990s. It is a test signal—specifically a -3dBFS undithered 12kHz square wave sampled (in this case) at 48kHz (24 bits). Since even the first odd harmonic (i.e., 36kHz) of the 12kHz square wave is removed by the bandwidth limitation of the sampling rate, we are left with a 12kHz sine wave (the main peak). In addition, an undithered 250Hz square wave at -144dBFS is mixed with the signal. This test file causes the 22 least-significant bits to constantly toggle, which produces strong jitter spectral components at the 250Hz rate and its odd harmonics. The test file shows how susceptible the DAC and delivery interface are to jitter, which would manifest as peaks above the noise floor at 500Hz intervals (e.g., 250Hz, 750Hz, 1250Hz, etc.). Note that the alternating peaks are in the test file itself, but at levels of -144dBrA and below.  The test file can also be used in conjunction with artificially injected sine-wave jitter by the Audio Precision, to show how well the DAC rejects jitter.

The coaxial input exhibits peaks within the audioband as high as -110dBrA near the 12kHz fundamental. This is an average J-test result, indicating that the Attessa DAC may be susceptible to jitter.

J-Test (optical)

The optical input peformed worse than the coaxial input, exhibiting peaks within the audioband as high as -100dBrA near the 12kHz fundamental.

J-Test with 10ns of injected jitter (coaxial)

The coaxial input was also tested for jitter immunity by injecting artificial sine-wave jitter at 2kHz, which would manifest as sidebands at 10kHz and 14kHz without any jitter rejection. Jitter immunity proved poor here, with visible sidebands at -70dBrA with only 10ns of jitter level.

J-Test with 100ns of injected jitter (coaxial)

The coaxial input was also tested for jitter immunity by injecting artificial sine-wave jitter at 2kHz, which would manifest as sidebands at 10kHz and 14kHz without any jitter rejection. Jitter immunity proved poor, with visible sidebands at -50dBrA with 100ns of jitter level.

J-Test with 100ns of injected jitter (optical)

The optical input was also tested for jitter immunity by injecting artificial sine-wave jitter at 2kHz, which would manifest as sidebands at 10kHz and 14kHz without any jitter rejection. Jitter immunity proved poor again, with visible sidebands at -50dBrA with 100ns of jitter level.

Wideband FFT spectrum of white noise and 19.1kHz sine-wave tone (coaxial input)

The plot above shows a fast Fourier transform (FFT) of the Attessa’s line-level pre-outs with white noise at -4 dBFS (blue/red) and a 19.1 kHz sine wave at -3dBFS (at 0dBFS, or 1Vrms, visible distortion was visible) fed to the coaxial digital input, sampled at 16/44.1. The slow roll-off above 20kHz in the white-noise spectrum shows that a brick-wall type reconstruction filter is not used. As a consequence, there are clear aliased image peaks in the audioband at -95dBrA at 13.2kHz, and at -110dBrA between 5 and 10kHz. The main 25kHz alias peak is just below -10dBrA.

RMS level vs. frequency vs. load impedance (1W, left channel only)

The chart above shows RMS level (relative to 0dBrA, which is 1W into 8 ohms, or 2.83Vrms) as a function of frequency, for the analog line-level input swept from 5Hz to 100kHz. The blue plot is into an 8-ohm load, the purple is into a 4-ohm load, the pink plot is an actual speaker (Focal Chora 806, measurements can be found here), and the cyan plot is no load connected. The chart below . . .

. . . is the same but zoomed in to highlight differences. Here we can see a maximum deviation within the audioband of about 0.1dB from 4 ohms to no load, which is an indication of a relatively high damping factor, or low output impedance. The maximum variation in RMS level when a real speaker is roughly the same, deviating by about 0.1dB within the flat portion of the curve (50Hz to 10kHz). Note that the dip in RMS level at higher frequencies, and rise at lower frequencies, is a result of the frequency response of the Attessa, and not a damping factor issue.

THD ratio (unweighted) vs. frequency vs. output power

The chart above shows THD ratios at the output into 8 ohms as a function of frequency for a sine-wave stimulus at the analog line-level input. The blue and red plots are for left and right channels at 1W output into 8 ohms, purple/green at 10W, and pink/orange at 77W. The power was varied using the volume control. At 1 and 10W, THD ratios are similar, with the right channel out-performing the left channel between 200Hz and 5kHz by as much as 5dB. Left channel THD ratios were roughly flat from 20Hz to 20kHz at 1 and 10W, hovering around 0.006-0.007%. At 77W, THD ratios varied wildly, starting above 1% at 20Hz, then down to 0.008% between 50 and 100Hz, then peaking again at nearly 1% at 3-4kHz, then back down to 0.01% at 20kHz.

THD ratio (unweighted) vs. frequency at 10W (MM phono input)

The chart above shows THD ratios as a function of frequency plots for the MM phono input measured across an 8-ohm load at 10W. The input sweep is EQ’d with an inverted RIAA curve. The THD values for the MM configuration vary from around 0.03% (20Hz) down to 0.002% for the right channel at 1.5kHz. The left channel was flat around 0.01% THD from 50Hz to 20kHz.

THD ratio (unweighted) vs. output power at 1kHz into 4 and 8 ohms

The chart above shows THD ratios measured at the output of the Attessa as a function of output power for the analog line-level input, for an 8-ohm load (blue/red for left/right channels) and a 4-ohm load (purple/green for left/right channels). The 8-ohm data out-performed the 4-data by about 5dB, and the right channel outperformed the left channel by about the same amount. THD ratios ranged from 0.02% down to 0.002% before the “knees,” which are seen at just below 80W into 8 ohms, and around 130W into 4 ohms. The 1% THD marks were seen at 83W and 134W into 8 and 4 ohms, respectively.

THD+N ratio (unweighted) vs. output power at 1kHz into 4 and 8 ohms

The chart above shows THD+N ratios measured at the output of the Attessa as a function of output power for the line-level input, for an 8-ohm load (blue/red for left/right channels) and a 4-ohm load (purple/green for left/right channels). While THD ratios were lower for the right channel compared to the left, noise levels were the opposite, with the right channel exhibiting nearly 10dB more noise (also see FFTs below) than the left channel. Overall, THD+N values for both loads ranged from 0.2% down to 0.01% before the “knees.”

THD ratio (unweighted) vs. frequency at 8, 4, and 2 ohms (left channel only)

The chart above shows THD ratios measured at the output of the Attessa as a function of frequency into three different loads (8/4/2 ohms) for a constant input voltage that yields 10W at the output into 8 ohms (and roughly 20W into 4 ohms, and 40W into 2 ohms) for the analog line-level input. The 8-ohm load is the blue trace, the 4-ohm load the purple trace, and the 2-ohm load the pink trace. The 8-ohm data clearly yielded the lowest THD ratios, from 0.007% at 20Hz, down to 0.005% at 3kHz, then up to 0.01% at 20kHz. The 4-ohm data was roughly 5dB higher in THD compared to the 8-ohm data, while the 2-ohm data was nearly 10dB higher than the 4-ohm data between 20Hz and 1kHz, then as much as 20dB higher at 20kHz.

THD ratio (unweighted) vs. frequency into 8 ohms and real speakers (left channel only)

The chart above shows THD ratios measured at the output of the Attessa as a function of frequency into an 8-ohm load and two different speakers for a constant output voltage of 2.83Vrms (1W into 8 ohms) for the analog line-level input. The 8-ohm load is the blue trace, the purple plot is a two-way speaker (Focal Chora 806, measurements can be found here), and the pink plot is a three-way speaker (Paradigm Founder Series 100F, measurements can be found here). Both speaker THD plots show signficant differences compared to the 8-ohm resistve load, which ranged from 0.005% to just over 0.01%. The 2-way Focal presented the most difficult load for the Attessa in terms of THD, with ratios fluctuating from as high as 0.07% at 20Hz, down to 0.001% at 2kHz. The three-way Paradigm yielded flatter results, with THD peaks at 0.02% at 100Hz and 20kHz, and dipping down to 0.003% at 4kHz.

IMD ratio (CCIF) vs. frequency into 8 ohms and real speakers (left channel only)

The chart above shows intermodulation distortion (IMD) ratios measured at the output of the Attessa as a function of frequency into an 8-ohm load and two different speakers for a constant output voltage of 2.83Vrms (1W into 8 ohms) for the analog line-level input. Here the CCIF IMD method is used, where the primary frequency is swept from 20kHz (F1) down to 2.5kHz, and the secondary frequency (F2) is always 1kHz lower than the primary, with a 1:1 ratio. The CCIF IMD analysis results are the sum of the second (F1-F2 or 1kHz) and third modulation products (F1+1kHz, F2-1kHz). The 8-ohm load is the blue trace, the purple plot is a two-way speaker (Focal Chora 806, measurements can be found here), and the pink plot is a three-way speaker (Paradigm Founder Series 100F, measurements can be found here). The 8-ohm resistive load yielded flat IMD results across the measurement range, around 0.006-0.007%. The two-way Focal yielded lower IMD results at low frequencies (0.0015% at 2.5kHz), rising up to 0.008% at 20kHz. The three-way Paradigm yiedled IMD results as low as 0.003% at 3kHz, and as high as 0.02% at 20kHz.

IMD ratio (SMPTE) vs. frequency into 8 ohms and real speakers (left channel only)

The chart above shows IMD ratios measured at the output of the Attessa as a function of frequency into an 8-ohm load and two different speakers for a constant output voltage of 2.83Vrms (1W into 8 ohms) for the analog line-level input. Here, the SMPTE IMD method is used, where the primary frequency (F1) is swept from 250Hz down to 40Hz, and the secondary frequency (F2) is held at 7kHz with a 4:1 ratio. The SMPTE IMD analysis results consider the second (F2 ± F1) through the fifth (F2 ± 4xF1) modulation products. The 8-ohm load is the blue trace, the purple plot is a two-way speaker (Focal Chora 806, measurements can be found here), and the pink plot is a three-way speaker (Paradigm Founder Series 100F, measurements can be found here). All three results are nearly identical, with flat IMD results howevering around 0.015%.

FFT spectrum – 1kHz (line-level input)

Shown above is the fast Fourier transform (FFT) for a 1kHz input sine-wave stimulus, measured at the output across an 8-ohm load at 10W for the analog line-level input. We find the second and fourth signal harmonics (2/4kHz) dominate at -80/-90dBrA (left/right), or 0.01/0.003%, and at -100/-110dBrA (left/right), or 0.001/0.0003%. On the left side of the main signal peak, we find the odd harmonics (180/300/420Hz) of the power-supply fundamental dominating with levels around -90/-80dBrA (left/right) and below, or 0.003/0.01%. As was seen in the THD and THD+N versus Output Power graphs above, the left channel exhibits higher THD while the right channel exhibits higher noise.

FFT spectrum – 1kHz (digital input, 16/44.1 data at 0dBFS)

Shown above is the fast Fourier transform (FFT) for a 1kHz input sine-wave stimulus, measured at the output across an 8-ohm load at 10W for the coaxial digital input, sampled at 16/44.1. We find roughly the same signal and noise-related harmonics dominating within the audioband, and at the same levels, as seen in the analog FFT above. The significant peaks at 43.1 and 45.1kHz are IMD products between the 1kHz signal and the 44.1kHz sample rate.

FFT spectrum – 1kHz (digital input, 24/96 data at 0dBFS)

Shown above is the fast Fourier transform (FFT) for a 1kHz input sine-wave stimulus, measured at the output across an 8-ohm load at 10W for the coaxial digital input, sampled at 24/96. We find roughly the same signal and noise-related harmonics dominating and at the same levels as seen in the 16/44.1 FFT above.

FFT spectrum – 1kHz (digital input, 16/44.1 data at -90dBFS)

Shown above is the FFT for a 1kHz -90dBFS dithered 16/44.1 input sine-wave stimulus at the coaxial digital input, measured at the output across an 8-ohm load. We see the 1kHz primary signal peak, at the correct amplitude, and noise harmonics from the right channel at near and slightly above the signal level.

FFT spectrum – 1kHz (digital input, 24/96 data at -90dBFS)

Shown above is the FFT for a 1kHz -90dBFS dithered 24/96 input sine-wave stimulus at the coaxial digital input, measured at the output across an 8-ohm load. We see the 1kHz primary signal peak, at the correct amplitude, and noise harmonics from the right channel at near and slightly above the signal level.

FFT spectrum – 1kHz (MM phono input)

Shown above is the FFT for a 1kHz input sine-wave stimulus, measured at the output across an 8-ohm load at 10W for the MM phono input. The second (2kHz), third (3kHz), and fourth (4kHz) signal harmonics from the left channel dominate at around -80dBrA (2kHz), or 0.01%, and -95dBrA (3/4kHz), or 0.002%. On the left side of the main signal peak, we find the fundamental (60Hz) and its odd harmonics (180/300/420Hz) dominating with levels from -75dBrA, or 0.02%, down to -90dBrA, or 0.003%.

FFT spectrum – 50Hz (line-level input)

Shown above is the FFT for a 50Hz input sine-wave stimulus measured at the output across an 8-ohm load at 10W for the analog line-level input. The X axis is zoomed in from 40Hz to 1kHz, so that peaks from noise artifacts can be directly compared against peaks from the harmonics of the signal. The second (100Hz) signal harmonic dominates at -85dBrA, or 0.006%. We also see the odd 60Hz power-supply related harmonics at -90/80dBrA (left/right), or 0.003/0.01%, and below. Also evident are IMD product between the signal and noise harmonic peaks.

FFT spectrum – 50Hz (MM phono input)

Shown above is the FFT for a 50Hz input sine-wave stimulus measured at the output across an 8-ohm load at 10W for the MM phono input. The X axis is zoomed in from 40Hz to 1kHz, so that peaks from noise artifacts can be directly compared against peaks from the harmonics of the signal. The most predominant (non-signal) peak is from the 60Hz power-supply fundamental at -75dBrA, or 0.02%. The second (100Hz) signal harmonic peak is below this level at around -85dBrA, or 0.006%.

Intermodulation distortion FFT (18kHz + 19kHz summed stimulus, line-level input)

Shown above is an FFT of the intermodulation distortion (IMD) products for an 18kHz + 19kHz summed sine-wave stimulus tone measured at the output across an 8-ohm load at 10W for the analog line-level input. The input RMS values are set at -6.02dBrA so that, if summed for a mean frequency of 18.5kHz, would yield 10W (0dBrA) into 8 ohms at the output. We find that the second-order modulation product (i.e., the difference signal of 1kHz) is just above (left) and below (right) -90dBrA, or 0.003%. The third-order modulation products, at 17kHz and 20kHz, are lower at around -100dBrA, or 0.001%.

Intermodulation distortion FFT (18kHz + 19kHz summed stimulus, coaxial digital input, 16/44.1)

Shown above is an FFT of the intermodulation distortion (IMD) products for an 18kHz + 19kHz summed sine-wave stimulus tone measured at the output across an 8-ohm load at 10W for the digital coaxial input at 16/44.1. We find that the second-order modulation product (i.e., the difference signal of 1kHz) is just below -90dBrA, or 0.003%. The third-order modulation products, at 17kHz and 20kHz, are lower at just above -100dBrA, or 0.001%. IMD products due to the 44.1kHz sampling frequency are also evident.

Intermodulation distortion FFT (18kHz + 19kHz summed stimulus, coaxial digital input, 24/96)

Shown above is an FFT of the intermodulation distortion (IMD) products for an 18kHz + 19kHz summed sine-wave stimulus tone measured at the output across an 8-ohm load at 10W for the digital coaxial input at 24/96. We find that the second-order modulation product (i.e., the difference signal of 1kHz) is just below -90dBrA, or 0.003%. The third-order modulation products, at 17kHz and 20kHz, are lower at just below -100dBrA, or 0.001%.

Intermodulation distortion FFT (18kHz + 19kHz summed stimulus, MM phono input)

Shown above is an FFT of the intermodulation distortion (IMD) products for an 18kHz + 19kHz summed sine-wave stimulus tone measured at the output across an 8-ohm load at 10W for the MM phono input. We find that the second-order modulation product (i.e., the difference signal of 1kHz) is right around -80dBrA, or 0.01%. The third-order modulation products, at 17kHz and 20kHz, are around -105dBrA, or 0.0006%.

Square-wave response (10kHz)

Above is the 10kHz square-wave response using the analog line-level input, at roughly 10W into 8 ohms. Due to limitations inherent to the Audio Precision APx555 B Series analyzer, this graph should not be used to infer or extrapolate the Attessa’s slew-rate performance. Rather, it should be seen as a qualitative representation of the Attessa’s mid-tier bandwidth. An ideal square wave can be represented as the sum of a sine wave and an infinite series of its odd-order harmonics (e.g., 10kHz + 30kHz + 50kHz + 70kHz . . .). A limited bandwidth will show only the sum of the lower-order harmonics, which may result in noticeable undershoot and/or overshoot, and softening of the edges. While there is no ringing in the corners of the square wave, the edges are softened.

Damping factor vs. frequency (20Hz to 20kHz)

The final graph above is the damping factor as a function of frequency. Both channels show a higher damping factor at low frequencies (over 200 at 20Hz), dipping down to just below 100 at 20kHz.

Diego Estan
Electronics Measurement Specialist