Roksan Attessa streaming integrated amplifier Measurements

Sidebar 3: Measurements

I performed measurements on the Roksan Attessa with my Audio Precision SYS2722 system. I preconditioned the amplifier by operating it at 1/8 power into 8 ohms for 30 minutes. At the end of that time, the top panel's temperature was 108.4°F/42.4°C and that of the grille over the heatsinks almost too hot to touch, at 134.4°F/55.9°C. The Attessa needs sufficient ventilation.

Looking first at the line-level analog "Variable Input" and with the sensitivity set to "Low" with the MaestroUnite app, the Roksan Attessa's maximum voltage gain at 1kHz was 34.9dB from the loudspeaker output into 8 ohms, 5.85dB from the Pre output, and 11.85dB from the headphone output with the sensitivity set to "Mid." The amplifier preserved absolute polarity, ie, it was noninverting from its loudspeaker, headphone, and Pre outputs. The input impedance was 25k ohms at 20Hz and 1kHz and 23.6k ohms at 20kHz.

The headphone output's source impedance was a usefully low 6 ohms from 20Hz to 20kHz. From the Pre output the impedance was 23 ohms, again at all audio frequencies. The loudspeaker output impedance was a very low 0.09 ohms at low and middle frequencies, increasing very slightly to 0.17 ohms at the top of the audioband. The variation in the Roksan amplifier's small-signal frequency response with our standard simulated loudspeaker was just ±0.1dB (fig.1, gray trace). Into resistive loads (fig.1, blue, red, cyan, magenta, and green traces), the Attessa's frequency response rolled off above the audioband. The fastest rolloff was into 2 ohms (green trace), the output being down by 3dB at 39kHz. There is the beginning of a boost at frequencies below 20Hz in this graph; the measurements in our sister magazine Hi-Fi News indicated that this boost reached +9.2dB at 3Hz.

1122RokAttfig01

Fig.1 Roksan Attessa, frequency response at 2.83V into: simulated loudspeaker load (gray), 8 ohms (left channel blue, right red), 4 ohms (left cyan, right magenta), and 2 ohms (green) (1dB/vertical div.).

Fig.1 was taken with the volume control set to its maximum; the excellent channel matching was preserved at lower settings of the control with all three outputs. I found no overshoot or ringing with the Attessa's reproduction of a 10kHz squarewave (fig.2). The Attessa's channel separation (not shown) was 80dB in both directions below 3kHz, decreasing slightly to 68dB at 20kHz.

1122RokAttfig02

Fig.2 Roksan Attessa, small-signal 10kHz squarewave into 8 ohms.

The unweighted, wideband signal/noise ratio (ref. 1W into 8 ohms), taken with the inputs shorted to ground and the volume control set to its maximum, was 66.5dB (average of the two channels), improving to 71.7dB, left, and 68.4dB, right, when the measurement bandwidth was restricted to 22Hz–22kHz, and by another 2.4dB when A-weighted. Spectral analysis of the low-frequency noisefloor while the Roksan amplifier drove a 1kHz tone at 1Wpc into 8 ohms with the volume control set to the maximum (fig.3) revealed the presence of AC-related spuriae at 60Hz and its odd- and even-order harmonics. These were higher in the right channel (red trace) than the left (blue), correlating with the different S/N ratios in the two channels. The levels of these spuriae didn't change when I repeated the analysis with the volume control set to –12dB, though the level of random noise dropped by the same 12dB.

1122RokAttfig03

Fig.3 Roksan Attessa, spectrum of 1kHz sinewave, DC–1kHz, at 1Wpc into 8 ohms with volume control set to its maximum (left channel blue, right red) (linear frequency scale).

Roksan specifies the Attessa's maximum power as 80Wpc into 8 ohms (19dBW) and 130Wpc into 4 ohms (18.13dBW). With clipping defined as being when the THD+noise reaches 1%, I measured clipping powers of 80Wpc into 8 ohms (19dBW, fig.4) and 128Wpc into 4 ohms (18.06dBW, fig.5) with both channels driven. (The AC wall voltage had dropped from 119.3V with the Attessa idling to 118.5V with it clipping into 4 ohms.) The shape of the traces in these graphs suggests that the actual distortion lies beneath the noisefloor below clipping.

1122RokAttfig04

Fig.4 Roksan Attessa, distortion (%) vs 1kHz continuous output power into 8 ohms.

1122RokAttfig05

Fig.5 Roksan Attessa, distortion (%) vs 1kHz continuous output power into 4 ohms.

Fig.6 shows how the percentage of THD+N in both channels varied with frequency into 8 and 4 ohms at 12.64V, which is equivalent to 20W into 8 ohms (blue and red traces) and 40W into 4 ohms (green and gray traces). The THD+N was very low into both impedances, though a little higher in the right channel (red, gray) than the left (blue, green). The distortion waveform was predominantly the subjectively innocuous second harmonic (fig.7), which lay close to the level of the AC supply–related spuriae (fig.8). Intermodulation distortion was also very low in level (fig.9).

1122RokAttfig06

Fig.6 Roksan Attessa, THD+N (%) vs frequency at 12.64V into 8 ohms (left channel blue, right red), 4 ohms (left green, right gray).

1122RokAttfig07

Fig.7 Roksan Attessa, left channel, 1kHz waveform at 20W into 8 ohms, 0.04% THD+N (top); distortion and noise waveform with fundamental notched out (bottom, not to scale).

1122RokAttfig08

Fig.8 Roksan Attessa, spectrum of 50Hz sinewave, DC–1kHz, at 40Wpc into 4 ohms (left channel blue, right red; linear frequency scale).

1122RokAttfig09

Fig.9 Roksan Attessa, HF intermodulation spectrum, DC–30kHz, 19+20kHz at 40Wpc peak into 4 ohms (left channel blue, right red; linear frequency scale).

Turning to the digital inputs, the coaxial S/PDIF input accepted data sampled at all rates up to 192kHz; TosLink was restricted to 96kHz data. All the digital inputs preserved absolute polarity. With the volume control set to its maximum, data representing a 1kHz tone at –20dBFS resulted in a level of 819.2mV at the headphone output, 11.47V at the speaker outputs. (The latter suggests that the digital stage has about 13dB too much gain.) All the following measurements were performed at the headphone output.

Fig.10 shows the Attessa's impulse response with 44.1kHz data. It is typical of a short, minimum-phase reconstruction filter, with a small amount of ringing following the single full-scale sample. With 44.1kHz white-noise data (fig.11, magenta and red traces), the filter rolled off slowly above the audioband, not reaching full stop-band attenuation until 31kHz, but with then a scalloped noisefloor. With a 19.1kHz tone at –3dBFS (cyan, blue), the slow rolloff means that the aliased image at 25kHz is only suppressed by 9dB, and other, lower-level aliased products can be seen both in the audioband and between 60kHz and 70kHz. The harmonics associated with the 19.1kHz tone are all very low in level, however. The Roksan's frequency responses with data sampled at 44.1, 96, and 192kHz (fig.12) all followed the same basic shape, with a 1dB rolloff at 20kHz.

1122RokAttfig10

Fig.10 Roksan Attessa, impulse response (one sample at 0dBFS, 44.1kHz sampling, 4ms time window).

1122RokAttfig11

Fig.11 Roksan Attessa, wideband spectrum of white noise at –4dBFS (left channel red, right magenta) and 19.1kHz tone at –3dBFS (left blue, right cyan) into 100k ohms with data sampled at 44.1kHz (20dB/vertical div.).

1122RokAttfig12

Fig.12 Roksan Attessa, frequency response at –12dBFS into 100k ohms with data sampled at: 44.1kHz (left channel green, right gray), 96kHz (left cyan, right magenta), and 192kHz (left blue, right red) (1dB/vertical div.).

The red trace in fig.13 plots the error in the analog output level as a 24-bit, 1kHz digital tone steps down from 0dBFS to –140dBFS. The amplitude error is negligible until the signal lies below –120dBFS. An increase in bit depth from 16 to 24 with dithered data representing a 1kHz tone at –90dBFS (fig.14) dropped the level of the Attessa's noisefloor by 12dB, which implies a DAC resolution of around 18 bits. With undithered data representing a tone at exactly –90.31dBFS, the waveform was symmetrical, with the three DC voltage levels described by the data cleanly resolved (fig.15). Repeating the measurement with undithered 24-bit data gave a well-formed if rather noisy sinewave (fig.16).

1122RokAttfig13

Fig.13 Roksan Attessa, left channel, 1kHz output level vs 24-bit data level in dBFS (blue, 20dB/vertical div.); linearity error (red, 1dB/small vertical div.).

1122RokAttfig14

Fig.14 Roksan Attessa, spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS with: 16-bit TosLink data (left channel cyan, right magenta), 24-bit TosLink data (left blue, right red) (20dB/vertical div.).

1122RokAttfig15

Fig.15 Roksan Attessa, waveform of undithered 1kHz sinewave at –90.31dBFS, 16-bit data (left channel blue, right red)

1122RokAttfig16

Fig.16 Roksan Attessa, waveform of undithered 1kHz sinewave at –90.31dBFS, 24-bit data (left channel blue, right red)

As implied by the blue and cyan traces, in fig.11, harmonic distortion via the digital inputs was very low, the second harmonic of a full-scale 50Hz tone lying at –104dB (0.0006%), left, and –114dB (0.0002%), right (fig.17). With an equal mix of 19kHz and 20kHz tones sampled at 44.1kHz and peaking at 0dBFS, a large number of high-level aliased images were present. Reducing the signal level by 3dB eliminated most of these images, leaving those at 24.1kHz and 25.1kHz the highest in level (fig.18) and all the others at or below –90dBFS.

1122RokAttfig17

Fig.17 Roksan Attessa, 24-bit TosLink data, spectrum of 50Hz sinewave, DC–1kHz, at 0dBFS into 100k ohms (left channel blue, right red; linear frequency scale).

1122RokAttfig18

Fig.18 Roksan Attessa, 24-bit TosLink data, HF intermodulation spectrum, DC–30kHz, 19+20kHz at –3dBFS into 100k ohms, 44.1kHz data (left channel blue, right red; linear frequency scale).

With undithered, 16-bit J-Test optical data, the odd-order harmonics of the undithered Fs/192, LSB-level squarewave mostly lay a little higher than the correct levels, these indicated by the sloping green line in fig.19. While the levels of the sidebands at ±229.6875Hz are much higher, repeating this test with data sent to the Attessa over my network with BluOS lowered these two sidebands by 10dB.

1122RokAttfig19

Fig.19 Roksan Attessa, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz: 16-bit TosLink data (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.

The Roksan's MM-compatible phono input preserved absolute polarity at all three outputs. The phono input's audioband RIAA correction was superbly accurate (fig.20), though with a slight rise in the very low bass. The input impedance measured 45k ohms at 20Hz and 1kHz, dropping slightly to 40k ohms at 20kHz. The maximum gain at 1kHz, with the sensitivity set to "Mid" with the MaestroUnite app, was 55.5dB at the Pre output, 61.4dB at the headphone output, and 84.5dB at the loudspeaker outputs. I performed all the subsequent testing using the headphone output and, other than S/N ratio, with the volume control set to –20dB to avoid clipping the output.

1122RokAttfig20

Fig.20 Roksan Attessa, phono input, response with RIAA correction (left channel blue, right red) (1dB/vertical div.).

I connected a wire from the Audio Precision's ground terminal to the grounding lug on the Roksan's rear panel to obtain the lowest noise with the phono input. The wideband, unweighted S/N ratio with the inputs shorted to ground and the volume control set to the maximum was a very good 71dB in both channels, ref. 1kHz at 5mV. Restricting the measurement bandwidth to the audioband increased the ratio to 76dB, while an A-weighting filter further increased the ratio to 82.6dB. Spectral analysis of the phono input's low-frequency noisefloor (fig.21) revealed very low levels of random noise and power supply–related spuriae.

1122RokAttfig21

Fig.21 Roksan Attessa, MM phono input, spectrum of 1kHz sinewave, DC–1kHz, for 5mV input, measured at headphone output with volume control set to the maximum (left channel blue, right red, linear frequency scale).

The phono input's overload margins, measured at the headphone output and ref. 1kHz at 5mV, were superbly high at 20Hz and 1kHz, at 23.5dB. The margin dropped to a still-good 10.5dB at 20kHz. The phono input's distortion signature was an equal mix of second and third harmonics, both lying at –106dB (0.0005%) with 1kHz at an input level of 20mV (fig.22). The level of the 1kHz difference product with an equal mix of 19 and 20kHz tones (fig.23) was very low, at close to –80dB (0.01%), with the high-order intermodulation products even lower in level.

1122RokAttfig22

Fig.22 Roksan Attessa, MM phono input, spectrum of 1kHz sinewave, DC–10kHz, for 20mV input, measured at headphone output with volume control set to –20dB (left channel blue, right red, linear frequency scale).

1122RokAttfig23

Fig.23 Roksan Attessa, MM phono input, HF intermodulation spectrum, DC–30kHz, 19+20kHz witht 100mV input (left channel blue, right red; linear frequency scale).

With separates, it is possible for the designers to maximize resolution and minimize noise. With a single-box solution like the Roksan Attessa, it is difficult to do these things, with the current-heavy power supply in such close proximity to the low-level analog and digital decoding stages. However, the Attessa's measured performance, especially that of the low-noise, low-distortion phono input, indicates that the inevitable compromises have been managed well.—John Atkinson

COMPANY INFO
Roksan/Monitor Audio Group
North American distributor: Kevro International Inc.
902 McKay Rd., Unit #4
Pickering, ON L1W 3X8, Canada
(800) 667-6065
ARTICLE CONTENTS

COMMENTS
volvic's picture

I enjoyed reading this review, but was disappointed Mr. Atkinson could not find a step up transformer to listen and review the phono section of this amp. As an owner of one of their turntables, I was keen to read how well it could sound. I am hopeful for a follow-up.

John Atkinson's picture
volvic wrote:
I enjoyed reading this review, but was disappointed Mr. Atkinson could not find a step up transformer to listen and review the phono section of this amp.

I appreciate your point. However, it is fundamentally important for a product being reviewed that the reviewer not change anything else in the system. If I had borrowed a moving-magnet cartridge or a step-up transformer, there would then be another change to the system in addition to the Roksan amplifier.

John Atkinson
Technical Editor, Stereophile

EDunbar's picture

John: True, but some (several?) buyers who have a moving coil cartridge might be likely to add a step up transformer, even if it were a “basic” unit, if they were to buy the Altessa, so that is a real world configuration and it would be valuable to a review.

JRT's picture

The step-up transformer pair could have been kept in the comparison(s), using the same turntable, MC cartridge, step-up transformer pair in the LP playback subsystem, and using the MM inputs to the phono preamplifiers in both systems under comparison. No?

Maybe could have also tested the accuracy of the phono preamplifiers' RIAA filters using something like (James) Hagerman Audio Labs' iRIAA2 ($49+s/h).

Maybe in some future tests? This probably won't be the last device with a phono preamp on your test bench.

edit: I want to clarify that I do appreciate your subjective reviews and your efforts in including useful and interesting objective testing to the subjective reviews (yours and others), and do not want my disagreements, comments and suggestions to be misinterpreted as complaining about anything in your efforts in this. Sincere thanks for what you do provide. I just sometimes want more.

Kursun's picture

I believe it is an art form to design a good back panel, as well as the front panel.

In this example the speaker posts are tucked away into corner, cramped together along with a ground post.

A very poor design indeed.

Jack L's picture

Hi

I do not agree. BACK panel is for connection terminals. Function is the top priority over "art form" !

This is a pretty decent back panel design: right hand side for digital &
left hand side for analogue connection terminals to reduce RFI/EMI noise emitting from the digital side to the analogue side inside the amp & the connection cables outside the amp.

"cramped together with a ground post" qtd Kunsun.

With audio output power only 80Wrms @8ohm/channel, no need large gauge size ground cable at all. So why "cramped together" ???

Jack L

JRT's picture

The manual for the Roksan Attessa states that loudspeaker cable assemblies terminated with 4mm banana plugs are the Roksan's recommended method of connecting loudspeakers to their amplifier. The binding posts each have a 4mm banana jack/receptacle at the end with a plastic plug which must be removed to accept the banana plug. The manual also mentions the alternative method of using unterminated loudspeaker cable with wire ends stripped bare and inserted into the binding posts, and further states that the binding posts may accept no larger than 12AWG wire, and they recommend using no smaller than 16AWG.

To your point, there is plenty of clearance if the recommendations stated in the manual are followed.

Jack L's picture

as I design/build phono-preamp & power amps for decades.

The problem is many readers here just want spit out whatever like or dislike without the right knowledge.

Jack L

JRT's picture

Roksan's Attessa integrated amplifier is available without the network attached streamed digital audio receiver-processor functionality for significantly lower price. That wasn't mentioned in the review, but a very cursory web search turned up some online vendors. I only looked at pricing from the first online vendor in the search results.

Without that added functionality, the vendor's asking price is $2.1k. With that functionality, $3.2k (article mentions $3.4k MSRP). Both versions include the DAC and Bluetooth. The networked streamed audio functionality could be provided by a separate component rather than the integrated internal component.

When this is out of warranty, out of production, an obsolete product, out of active support, no longer receiving regular updates/patches to software/firmware, the integrated amplifier functionality may continue to function for several decades until some internal electronics component eventually fails. Unpatched software/firmware in the network attached hardware may cause some big problems beyond the audio playback setup.

I am not confident that the network security patches will be promptly released as new malware and network security vulnerabilities continually emerge, especially after the product is no longer offered for sale new by the manufacturer. I wouldn't want something attached to my LAN sniffing packets and skimming passwords and other account credentials on my network. I would rather have a separate nonproprietary device providing this networked streaming functionality, something I can easily update and eventually upgrade or replace separately from the audio amplifiers, control preamplifier, DA converter, etc.

Jack L's picture

Hi

BINGO ! Great minds think alike !?

That's exactly what yours truly cheapskate has done since day one of streaming: no costly brandname all-in digital preamps/integrated amps which are always prone to obsolescene sooner if not later.

Being a vinyl addict, why should I spend any decent money for any digital audio ? So with very little money, I got a dirt-cheap no-name DVD player with LAN/streaming function/auto-upgradable firmwares, + a dirt-cheap DAC (24bit-192KHz). Yet they both serve me nice & neat on playing music CD/DVDs & allow me to watch any on-line classical music programmes - FREE via say, YouTube.

Play smart is the name of the digital game - never spend big bucks in it due to its every changing "vulnerabilities".

Listening is believing

Jack L

X