Mr. Andrzej Stelmach is an electronic engineer, hobbyist and music lover. So I did not hesitate for even one second when he turned to me to test his D/A converter, a few years ago. And in the issue 14 from June 2005 a preliminary version of his DAC, in a DIY kit version, was tested (HERE). It was great. He continued working on the product, and now we have a new, improved version, as well as a switch to a “production” unit, complete with a safe cabinet, etc. And here it is – named DAC-04, in a cabinet (with the company name Stelaudio printed on it) – the newest DAC from Mr. Andrzej. I will not write more, as the listening session is the most important part of the test.

SOUND

Discs used for testing:

Compact Disc

• e.s.t., Viaticum, ACT Music+Vision, ACT 6001-2, CD; review HERE.
• Dire Straits, On Every Street, Vertigo/Universal Music Japan, UICY-93734, SHM-CD.
• Jimmy Giuffre, Western Suite, Atlantic/Warner Music Japan, WPCR-25160, CD.
• Chris Connor, Chris Connor, Atlantic/Warner Music Japan, WPCR-25163, CD.
• Maanam, Łóżko, Pomaton EMI, 52718, CD.
• Miles Davis, Miles Davis All Star Quintet/Sextet, Prestige/Universal Music Japan, UCCO-9294, CD.

WAVE, FLAC (LP)

• G. F. Haendel, Messiah (Dublin Version, 1742), Dunedin Consort&Players, Linn Records, CKH 312, 3 x 180 g LP, review HERE.
• G. F. Haendel, Messiah (Dublin Version, 1742), Dunedin Consort&Players, Linn Records, CKH 312, FLAC 24/88,2.
• The Beatles, Love, Apple/Parlophone/EMI, 3798081, 2 x 180 g LP.
• The Beatles, Love, Apple/Parlophone/EMI, 798102, CD + DVD-A 24/96; review HERE.
• The Beatles, Love, Apple/Parlophone/EMI, 798102, WAVE 24/96.
• John Coltrane, Blue Train, Blue Note/Classic Records, HDAD 2010, DVD-V 24/96+DVD-A 24/192.
• John Coltrane, Blue Train, Blue Note/Classic Records, HDAD 2010, FLAC 24/192.
• Helge Lien Trio Hello Troll, Ozella Music, OZ021CD, FLAC 24/96; review HERE.

Like I mentioned, I know this DAC, or at least its 2004 version. Already then I appreciated Mr. Stelmach’s idea, who saw the main possibility to properly process digital signals in preparing it when it is still digital, before the conversion stage. The converter had a very good sound and wasn’t expensive, not to say – cheap. But it was a DIY project, there could be no talk about selling it. This time I got a shop ready version, equipped in a simple, yet nice looking enclosure, the company has its name and the unit can be sold in audio shops as is. This is how it looks from the manufacturer side. From my side the changes are much bigger – my reference system is much more refined, and I listened to many dozens of products. And this has a huge impact. Only now I can fully appreciate the mastery of the design. Yes – mastery. Because still many things can be done better, and CD players for 10000zl can sound more precise, yet some aspects of the sound of the DAC-04 are unique, regardless the price tag. But only now I could ascertain that.

The most important characteristic of the DAC, supplied with the signal from the S/PDIF output of my Lektor Prime, is the incredibly sublime midrange. Incredible, for a digital source, was the sound of old recordings, usually tonally flattened, “washed out” from timbres, even by expensive units, or at least “tuned”. Here voices, like Chris Connor from the disc Chris Connor or Mark Knopfler from On Every Street Dire Straits had a deep timbre, a certain velvety border, which not only combined everything into a reasonable whole, but did also add some refinement patina to the music, some kind of a “quality certificate”. There is no place for brightening or sharpening here. The sound isn’t also warmer. I say – this is not about any “manipulation” of the signal, which would allow for extracting elements from the middle of the frequency range to the surface of the mix. This is just the ability to differentiate the timbres and thoroughness of showing the relationships between the base sound and its harmonics (this is just coherence), that gave this, and not a different result – although there was no added warmth, everything sounded in this very natural way. I mentioned vocals: beautifully, really very nice sounded the disc Łóżko Maanam. It is an underestimated disc, which is for me, together with “Róża”, the essence and culmination of “late” Maanam, still with Jackowski. Both discs are very nicely recorded, clean and with nice dynamics. With Mr. Stelmach’s DAC Łóżko sounded with swing, on the edge, but still on “the good” side of neutrality. Kora’s voice was warm, did not jump out and was not sharpened. This is how it should sound. Its separation from the background was not as good as with my reference player, also the stage was not equally deep, but within the boundaries of the band, the depth of the timbres was incredibly good.

Due to low distortion, I think – mostly digital, the sound of the DAC-04 seems slightly warm. I wanted to show, that this was not a primitive trick, like in some tube amplifiers. This way of thinking seems justifiable, because the upper bass, somewhere on the borderline with the midrange, is slightly stronger than its medium range. If I had to compare it to something – and strong lower bass should be added to the equation – then it would be the Ayon CD-1s or the USB DAC Wavelength Audio Cosecant v3. The bass guitar from Dire Straits and the contrabass from Viaticum e.s.t. sounded softer at higher range, their energy was not as big as higher and lower. This is a slight deviation from neutrality and linearity. There is no doubt about that, because my Prime played the same material evenly. I had the same complaints with the Benchmark DAC-1 USB. Here, and in the Ayon, the lower passages were not that good controlled, like with the DP-500 Accuphase. Ayon’s bass was bigger and more impulsive. Some of the Readers writing about this player pointed to this element as a bit problematic. But others took it as it is. I think this is also how you should approach bass in the DAC-04. I think, that the constructor had no full range, precise speakers, like my Harpia, at his disposal while designing the device – otherwise he would hear that soon enough. And I have no doubts, that he knows what he is doing.

I talked about this element of the sound a little longer, not because it is “bad” in an absolute scale, but because it is worse than the midrange. And only when compared to units costing about 10000zl and more. Two more DACs, cheaper than the mentioned level – the mentioned Benchmark and Audionemesis DC-1 Up-grade played the lower octave in a similar way. This is highly interesting, because they are different with regard to the construction details and philosophy of the sound. But let’s return to the treble. It is strong and full. It reminds me of the DC-1 Up-grade, but is much more resolving. /p>

At first I thought, that the energy of this sub-range is close to that what is presented by the Prime and the DP-500, but after a longer session I was certain, that the Stelaudio DAC has darker treble, that the amount of information is lower there. There is something “vinatage” in that, because the earlier sources from Mark Levinson sounded similar. This is not bad – like I say, we are in the 10000zl region, even when the DAC-04 is much cheaper. But there is something in the percussion cymbals and in the upper registers of the voices, that suggests not so precise reproduction, not so resolved connecting tissue. Because there are no visible differences between reverbs, sibilants, etc. Those are not underlined, this is why all discs sound at least pleasurable, but we are not fully aware of what is on them, and why. And in this way I return to the beginning: the best range of the DAC is its midrange. This is an outstanding achievement, and when we aren’t bass fanatics, when we have smaller speakers, then this will be the main decision making point. The DAC sounds a bit like tube based devices – I mean the lack of annoying distortion, fullness and richness of harmonics.

The unit sounds slightly different when connected to a computer using an USB cable. Like I showed in the test of the Cambridge Audio Dacmagic (HERE) this is not an easy task, and only specialized USB converters, like the mentioned Cosecant v3 Wavelenght Audio, or the Brick are able to show the quality of music collected on a HDD, including the high resolution one. DAC-04 sounds USB music in a bit “plastic” way, underlining the upper midrange, which is also not so resolved. This is the way the Cambridge sounds, and although Mr. Stelmach’s device handles tonal balance and dynamics better, still discerning high resolution and 16/44.1 versions is very difficult, if not impossible, at least for me. It is not that I dissuade the USB input! I am just telling, that for being able to extract everything it can deliver, then we need much more expensive solutions and special treatment of the link, for example placing the USB receiver in asynchronous mode, where the sampling frequency and word length is determined by the computer, and not the DAC. The DAC-04 is not bad, but the outstanding quality of the sound using the S/PDIF rises the threshold to such a level, that the USB cannot reach it.

This is a splendid device for little money. Besides the Audionemesis DC-1 it is for me the benchmark DAC for less than 5000zl, which can compete with units costing up to 10000zl. But you absolutely have to provide it with a high class transport – my Prime with the CD-Pro2 LF is a minimum. And now: Stelaudio is a small micro-company. If I would be an amplifier manufacturer, like Struss, or somebody “bigger”, I would not wait but I would buy the project, or at least I would start a cooperation with Mr. Stelmach. This would be an ideal supplement to that offering, and if a better enclosure would be provided, with a display for the sampling frequency, word length, etc, then it could cost 5000zl easily. I would do that before somebody else snatches the design or the designer. Just don’t spoil the project…

DESCRIPTION

This is a low-profile unit and placed in an aluminum enclosure with a thick front. On the front we have two buttons – one for powering the unit, the other to switch on the LEDs, an input selector with indicator LEDs, and a blue led indicating synch with the drive. I missed an indicator of the input signal parameters. On the back we have the following input sockets: S/PDIF, TOSLINK and USB and the RCA analog output.

Inside we have a small toroidal transformer, the USB input and digital receiver board and the main converter PCB.

Receiverk
The receiver/decoder of the S/PDIF signal is the CS8416. The unit has three inputs selected by a rotary switch – two S/PDIF inputs (RCA and optical TOSLINK) and an USB input (B type). The USB port allows to play the music back from a computer – the signal received by USB is converted to S/PDIF and the received by the CS8416 just like from other inputs. After plugging the unit to the computer a device is installed in the system called “USB audio device”. The USB interface is powered from the computer. The USB input allows signals in with the maximum sampling frequency of 48kHz – limited by the used chip. All inputs, including the USB, are galvanically separated from the signal source. The logic of the receiver mutes the signal in case of S/PDIF receive errors and if the signal received is not an audio signal. The PLL generator power supply inside the CS8416 has a separate stabilizer to minimize jitter distortion.

Converter
The DAC uses two ICs, the stereo CS4398 – one stereo chip plays one channel in phase and counter-phase. Channel separation and phase reversal are made digitally by logic chips placed before the CS4398. An additional IC in the logic part improves the shape of the digital signals send to the converter chips, what lowers jitter. The “mirror” signals from the CS4398 are summed (or rather subtracted) in the differential amplifier Analog Devices AD825 – such system allows to achieve some parameters (dynamics, sound to noise ratio, distortion) twice, meaning 3dB, better than in a standard setting, where one chip handles both stereo channels. This allows to play signals back, which are otherwise masked by noise and distortion. The power supply to the analog sections of the right and left channel are separated by RC filters. There are no capacitors in the sound path.

Power supply
Voltages are supplied by a toroidal transformer with a fourfold power reserve, with separate secondary windings for:
- digital section of the CS4398 and the receiver logic
- analog section of the CS4398
- operating amplifiers
In the rectifiers for analog section quick rectifying diodes are used. The values and makes of the electrolytic capacitors used in the last part of the power supply for the analog section of the CS4398 and the op-amps were chosen based on listening session – those are Rubycon ZL and Nippon Chemicals LXV capacitors. The analog ground is separated from the digital one to minimize the propagation of interference from the digital to the analog section.