pl | en

Power amplifier & sound processor

Mozzaik Audio
MOZZAIK1

Manufacturer: MOZZAIK AUDIO
Price (when reviewed): 7999 € (+VAT)

Contact: via website

mozzaik-audio.com

Provided for test by:
MOZZAIK AUDIO


Review

Text: WOJCIECH PACUŁA
Images: Wojciech Pacuła | Mozaik Audio
Translation: Marek Dyba

No 204

May 1, 2021

WORLD PREMIERE

The MOZZAIK AUDIO was founded in 2019 by MARKO DVEČKO - engineer and designer - and MIHAEL MUNT - marketing director. Its first and so far the only product is a power amplifier with an integrated signal processor, the MOZZAIK1.

DON'T KNOW IF YOU NOTICED, but in every interview I conduct with an audio editor in the "The Editors" series, I ask whether my interlocutor has recently encountered any important, interesting, new technology or solution. My guests give different answers, but the most common ones concern room acoustics correction, and more broadly - the topic of signal processors (the latest interview № 30, with WOLFGANG KEMPER from the HIFISTATEMENT.net magazine can be found HERE).

In most cases, it is about digital compensation of room modes, as in the case of the processor used in the ARCAM SA30 (PL). From time to time, however, I encounter a completely different approach, but also focused on the CORRECTION, namely with systems aimed at correcting errors in devices, loudspeakers and even cables.

Let me refer you to three examples on this occasion, it so happens that all of them come from Poland: ANCIENT AUDIO, YAYUMA AUDIO (both are active solutions) and the ideas of Mr. STANISŁAW RUTKA (his further development of the ideas used in MIT Cables) ; let me „dot the 'i'” with the help of the SPEC brand (and their passive solutions). Each of these three manufacturers offer their own unique way to minimize phase errors and other distortions, and each does it in their own way.

ANCIENT AUDIO The oldest solution among them is a DIGITAL sound processor with an algorithm developed for Ancient Audio, which the company uses in its devices, for example in the HEADPHONE AMPLIFIER / P-1 PROCESSOR. The basis for the solution is creating a theoretical model of a loudspeaker. It takes into account the size of drivers’ membranes, what they are made of, attenuation, reflection of the sound waves from the cabinet baffles, cabinet resonances and many more. These phenomena are modeled by a unique algorithm that is executed by a DSP speaker processor in real time.

YAYUMA AUDIO AWARNESS LINE ASP 01 is also a signal processor, but an ANALOG one (hence the processor name: ASP - Analog Signal Processor). Its use is different, although it is also related to the speakers. It is intended to minimize the errors in converting a „velocity” signal to „intensity” signal that occur when signal room a microphone gets to an electro-acoustic transducer. It is plugged between the source and the amplifier (or preamplifier) or between the preamplifier and the power amplifier.

PASSIVE SYSTEMS Finally, the simplest technically, which does not mean that it is easy to develop, filter used in the cables of Mr. STANISŁAW RUTKA, resembling those used by the American company MIT CABLES. These circuits compensate for phase shifts, impedance, capacitance, etc. The simplest filter would be the circuits that I use to prevent back electromotive force from transducers, proposed by SPEC REAL-SOUND PROCESSOR RSP-AZ9EX.

Last but not least I need to mention two, even more complex solutions, which use digital processing in DSP circuits (similarly to Ancient Audio), and which are proposed by Linn and Devialet in their products; the former offers their SPACE OPTIMIZATION system (SPACE = Speaker Placement And Custom Environment; accessed: 08/01/2021), and the latter the EXPERT PRO CONFIGURATOR (accessed: 01/08/2021). It is important that all the solutions mentioned above go beyond the classic scheme of the acoustics correction system.

The new company MOZZAIK AUDIO also has its own solution. Its first product, Mozzaik1, is a power amplifier with an integrated filter system aimed at correcting the imperfections of classic loudspeaker systems and, to some extent, their placement in the room.

»«

| A few simple words…

MIHAEL MUNTA | Marketing director

⸜ MIHAEL MUNTA (on the left) and MARKO DVEČKO

UP TILL THE FOUNDATION WE WORKED AS SOFTWARE DEVELPERS. Electronics were always Marko’s initial passion so throughout his career he continuously fiddled with audio amplifiers on the side. Marko was putting a lot of thought into audio systems and how audiophiles choose the perfect combination of audio equipment to build the perfect sound. It wasn’t until 2018 when he got an idea for an amplifier that would be different than all the others.

The first obstacle that occurred to him was finding the perfect amplifier for various speaker models. It is well known to all audiophiles how challenging this pursuit can be. Marko saw this challenge as something that, if overcome, would present great value to audiophiles. So he started thinking about ways this problem could be solved. At this point we decided that we wanted to make an amplifier that could adapt to different speakers and to do that we had to process the signal in addition to amplifying it. The innate distortion of the speakers was identified as the point that needed fixing in order to achieve the desired effect.

Every speaker model produces different amounts of distortion and at differing frequencies. The first step was to create a way to let our amplifier „know” which kind of speakers the signal will eventually reach. We needed a sort of an input variable with which we would be able to distinguish different speaker models. To accommodate this the first setting, the FREQUENCY COMPENSATION, was developed.

It enables a listener to set a frequency at which the next setting in the chain will reduce speaker distortion. By exploring the issues of the human ear and speakers we identified that the distortion which the speakers produce is mostly heard in the high frequencies. This is why the compensation frequency should be very close to the frequency at which the tweeter takes control, in other words, the crossover frequency.

After compensation frequency is set we have successfully identified the point where the speaker is at its worst. The next challenge was to reduce the distortion at that frequency. The idea arose from the universal truth that two negatives give a positive. In this case one negative is the speakers’ innate distortion and the second negative is the distortion produced by our processing module.

The second setting, SPEAKER COMPENSATION, lets you control the amount of this produced distortion you want to apply to the before selected compensation frequency. When the most optimal amount of produced distortion is applied the speakers’ innate distortion will be minimalized and will therefore extract the full potential of your speakers.

⸜ The Mozzaik1 before it is put into its chassis

There is also a third setting in the chain called PSYCHOACOUSTIC CHARACTERISTICS. There are various factors that dictate our perception of sound. Whereas with the first two settings we achieved the objective adaptation of sound to various speakers, this setting was created to achieve adaptability to the human ear.

After extensive research and looking over mathematical models of sound perception at the ear we have developed a model of our own that enables modification of the signal only in this subjective domain. At the electronics level it is once again achieved by producing a very complex yet harmless distortion and applying said distortion to the audio signal. With all of the settings together we have enabled the listener to have full control over the sound, be it objective or subjective.

The electronics part was developed first and at that time we haven’t given much thought to the design of the case. We were just happy to have accomplished something innovative regarding sound. Initially, the case should have been made out of wood, but as time passed by we thought if we already have originality in sound, we should also make the case look original as well. That’s where we decided we wanted the combination of aluminum and carbon fiber.

As for the future we already have the algorithms developed for a digital to analog converter. It will also be a very innovative product, utilizing sigma delta modulation for a very interesting result. For now, our focus lies on placing Mozzaik1 to the global market and having it as our flagship product. Once we are established and our brand is recognized we will start developing more products. We hope to have the full chain of audio products under our brand in the future.

»«

SO MOZZAIK1 IS A POWER AMPLIFIER - this is its primary function. It has a form of a small, flat "box", placed on tiny rubber pads glued to the bottom. Its form is on the one hand modern and on the other hand not too modern. The sides are made of thick aluminum slabs with cooling slots. They overlap with the actual housing, made of two elements - an aluminum bottom (and back) and a carbon fiber braid part that covers the top and front; on the front there is a company logo - a mosaic with one missing element - and a white power LED.

Unlike conventional power amplifiers, the rear of the Mozzaik1 is really interesting. The sound compensation knobs are especially noteworthy. There are six of them, in two rows. The three on the top are for initial adjustments and the bottom ones for fine adjustments - for each "initial" "click" (step) there are eight fine ones. The knobs are graduated with lines and marked with symbols, but this is not a very convenient solution. I would prefer the knob collars to be marked with numbers.

Knobs are manipulators used to change the settings of analog high-pass filters - the sound is processed using analog circuits. As Michael said, these filters introduce specially selected signals to the audio signal to compensate for the distortions generated by loudspeakers. Let me remind you that Nordost followed a similar path in the product known as QPOINT. On the manufacturer's website (which is very good indeed) you will find a guide and a simulator that will help you to set the initial values.

The amplifier used by the manufacturer works in the analog class D. As the designer told me, he wanted to show how good this type of design can sound like. And here is an important thing - in one chassis there are as many as four amplifier channels, hence the presence of four speaker outputs per channel on the back panel.

Two channels offer lower output and are intended for mid-woofers and tweeters, delivering up to 2 x 150 W @ 8 Ω, and two more powerful ones, with an output of 2 x 220 W @ 8 Ω and are intended for the bass-woofers. Thanks to this arrangement, bi-amping will be very easy, and we will save a lot of money - for the price of one amplifier we actually get two. The left and right channels are electrically separated and each has its own power supply. This gives us two separate two-channel amplifiers in one chassis.

| SOUND

HOW WE LISTENED The Mozzaik1 is, in fact, a classic device from the audiophile arsenal, although belonging to the high-end class - it is a power amplifier. And so it was tested, replacing in the HF reference system a powerful power amplifier, working in the AB class, the Soulution 710.

I sent the signal to it from the Ayon Audio Spheris III preamplifier with the Crystal Cable Absolute Dream RCA interconnect (Mozzaik1 is an unbalanced design). After some time, however, I tried the connection bypassing the external preamplifier, adjusting the volume directly in the CD-35 HF Edition SACD player, and this connection turned out to be perfect - this is how I carried out the whole test.

As my Harbeth M40.1 speakers feature single speaker terminals, I only used two amplifier channels by plugging the cables into the top speaker terminals.

But one of the elements that make the tested device stand out from the crowd of similar products are the filters that allow user to correct distortions generated by our speakers. The design of the amplifier is minimalistic and probably, in order not to mess with it, it was decided to place the manipulators on the back panel - after all, one sets them up once and forgets about them (one shouldn’t keep using them back and forth, as it will take away the joy of listening to the music). But to see what the filters really bring to the sound, I positioned the amplifier with its rear in the front and connected it with flexible NOS cables from Western Electric.

Recordings used for the test | a selection

⸜ ABRAXAS, 99, Metal Mind Records MMP CD 0102, CD (1999);
⸜ BILL EVANS, Everybody Digs Bill Evans, Riverside/JVC JVCXR-0020-2, XRCD (1958/2007)
⸜ DEAD CAN DANCE, Into The Labyrinth, 4AD/Beggars Japan WPCB-10076, „Audiophile Edition”, SACD/CD (1994/2008)
⸜ DIANA KRALL, This Dream Of You, Verve Records UCCV-1181, SHM-CD (2020);
⸜ ELLA FITZGERALD & JOE PASS, Take It Easy, Pablo/JVC JVCXR-0031-2, XRCD (1974/1987)
⸜ LAURIE ANDERSON, Homeland, Nonesuch 524055-2, CD + DVD (2010); recenzja
⸜ TAKESHI INOMATA, The Dialogue, Audio Lab. Record/Octavia Records OVXA-00008, SACD/CD (1977/2001);
⸜ THE MONTGOMERY BROTHERS, Groove Yard, Riverside/JVC, JVCXR-0018-2, XRCD (1961/1994)

»«

PROCESSOR THE EFFECT OF THE PROCESSOR SETTINGS ON THE SOUND OF THE MOZZAIK1 AMPLIFIER is not something that we will appreciate in a short demo, for example in an audio showroom or at some show. These changes are not comparable in scale with a replacing an amplifier with a better one, or even switching cables. I would estimate their influence on the sound - still remaining in the "quantitative" domain - at about 5%. That's a lot, but still not enough to make hype about it. We will introduce a much greater correction by simply adjusting the tone.

Fortunately, this is not the first time I have encountered such changes and quite smoothly, as it seems to me, I managed to move from the "quantitative" domain to the domain that I have to, I simply have no other choice, call "qualitative". The point is, quantity in high-end does not always translate into quality. I would even say that this happens quite rarely. The processor used in the Mozzaik1 amplifier is a beautiful example of such a case - an example that although quantitatively results in a 5% change, qualitatively we can talk about two different devices - with the processor in play or not.

This is a change of 10 or even 15%, and the 10 or 15% is a huge gap in terms of high-end. Therefore, the tested device sounds in a really refined way. I tested the filter settings as recommended by the designer, Marko Dvečko, i.e. starting their adjustment with a drums and a vocal recordings - in the first case I used the track from the Takeshi Inomata’s Dialogue album, and in the second one - a beautiful recording of Ella Fitzgerald's voice from the Take It Easy album, accompanied only by Joe Pass's guitar.

The differences that I heard were exactly as what the designer had suggested, which testifies well to both his listening skill and his efficiency in the describing his impressions. The first two knobs influence primarily focus of the sound - this is the first knob. Changing these settings also moves the foreground more towards us, or more backwards - that's the second knob. The latter one gave me an opportunity to determine the subjective size of the presentation and its tangibility.

The final settings looked like that:


The test was carried out using them and the following notes apply to the amplifier and processor being treated as a "black box", it is as if it were a normal power amplifier.

AMPLIFIER I have already said something about it, and the strongest term I used was probably: "high-end". This is an amplifier with a refined sound, and what’s more a sound that I personally like, one that I am looking for and which - but this is only my opinion, each of us probably has a different one - is the "right" one.

The Fitzgerald and Pass album, which I mentioned, is a minimalistic recording and although it is a recording from 1973, it is a monophonic one (or so it sounds) with reverb placed behind the instrument and vocals, not on the sides, which makes both these elements seem tangible and at your fingertips. This is a recording in which the noise of the Pass tube amplifier has not been cut out, perhaps prepared without Dolby noise reduction at all (although I am not sure of that), because there is a lot of noise here. The Mozzaik1 amplifier showed these properties of the disc perfectly.

| Our albums

⸜ ELLA FITZGERALD & JOE PASS Take It Easy

Pablo/JVC JVCXR-0031-2, XRCD (1974/1987)


TAKE LOVE EASY IT’S AN ALBUM RECORDED BY jazz singer ELLA FITZGERALD and jazz guitarist JOE PASS, released in 1974. This is the first in a series of four albums recorded by this duo, but also the last one in her career that the singer recorded in a duet, with the accompaniment of just one instrument.

The material was captured overnight, on August 28th 1973 by Rafael O. Valentin in Los Angeles; the producer was Norman Grantz, founder and owner of the Pablo Records. Valentino was nominated for a Grammy twice, in 1958 and 1961 - interestingly, not for his music, but - respectively - for voice and special effects. In the 1960s, he was responsible for the production of interesting, but less known titles, such as the Oscar Peterson & Nelson Riddle album.

The DISCOGS.com portal lists 33 versions of the Take Love Easy, and the vast majority of them come from the year in which the album was released. It is not a big number and it proves that the album did not sell well (by contemporary standards). The first reissue dates from 1983 and was intended for the Scandinavian market, and two years later in Japan the first version was released on CD, with material remastered for its needs.

In 1997, JVC releases the best digital version of it, remastered in 20 bits by Alan Yoshida at A&M Mastering Studios in Hollywood, from the original master tape; the producer of this release was Akira Taguchi (more about XRCD in the article entitled PRAISE (NON) FORMAT: XRCD). It's great music, with a very well captured Fitzgerald vocal, which is usually a bit harsh and sharp on other albums. Fun fact - the album was mixed in mono.

THE TAKE IT EASY ALBUM WAS PLAYED by the tested amplifier in a warm, dense way, with a fantastically shown foreground. Fitzgerald's voice was extremely clear, and this is not a "clinical" recording, it also had a large volume, was dense and rather warm. After a few more albums it turned out that the device does not warm up the sound in the way that tube amplifiers do. The album by Diana Krall, entitled This Dream Of You showed me, however, that the treble is open and strong and that it is not about closing the sound.

This is not a perfect recording and in Krall's voice you can hear "throat" elements being a bit underlined, a bit hard and bright. The tested amplifier showed it perfectly, did not try to smooth or warm anything up. However, it was not "underlined", which means that this aspect of the album's production did not stand out to the fore. I heard it because I wanted to hear it, because I was looking for it Without such "search" mode being on, the sound will seem warm again. As it seems, it is not about warming the sound, that is rounding the attack and / or correcting the timbre, but about resolution and internal coherence of the presentation.

It is absolutely surprising how open that sound of this amplifier is, while maintaining the density and harmonic richness. Without knowing what technology I am dealing with, I would never, ever say that it is a class D amplifier. And this is a unique achievement, because although I know a few excellent devices of this type, apart from rare cases, such as the flagship, costing 60,000 EUR, four-piece RPA-MG1000 amplifier of the Japanese brand SPEC, the intrinsic nature of a pulse amplifier is fairly easy to point out.

Mozzaik1 does not allow it, or at least not directly. The Bill Evans recordings I listened to before, released on XRCD by JVC sounded warm and dense. You can hear that this type of music and such recordings are something the amplifier feels best with. It was similar with another re-issue of this label, the Wes Montgomery Brothers’ Groove Yard.

Having already listened to the amplifier for some time, I paid more attention to something else - not the density, but to how the amplifier is both selective and resolving in this density. This is something that we can usually hear with small SET tube amplifiers, with semiconductor electronics - very rarely, and with class D, almost never.

The recordings of Dead Can Dance from the Into The Labyrinth were also exceptionally convincing. It was when listening to it that exceptionally good imaging was confirmed, something I have not talked about yet. It is an amplifier that builds large phantom images, but arranges them within a wide and deep soundstage - the latter is very difficult for amplifiers of this type. The bodies of instruments are not particularly tangible, nor are they as massive as with a high-class 300B amplifiers, don’t count on that. But also the momentum of the presentation is so big, it is so rich in its entirety that we do not pay attention to individual elements of the presentation.

The bass is also very interesting. It is very well controlled and compact despite not having a hard attack. The lowest range does not have as much energy as with the Soulution 710, it is not as massive. And yet, probably due to the excellent saturation, the sound of the whole seems to be set very low, which was confirmed when listening to Laurie Anderson's Homeland, and then the Abraxas’ 99, I wrote about in the part 7 of the series "Here is an album".

With both discs the amplifier showed something else, that is outstanding phase coherence, thanks to which the spacial effects and sounds in out of phase surrounding the listener sounded in an unusual - or at least very rare - way. They were extremely clear and distinct, and at the same time tight and coherent. There was no question of "holes" and discontinuities in the sphere surrounding me.

| Summary

The Mozzaik1 amplifier is one of the most interesting power amplifiers I have dealt with recently. It sounds open, delivers low bass, but most of all it sounds coherent, dense and engaging. It clearly shows the differences in production, so it is not suitable for correcting the sound of recordings. But the set of filters used in it allows to clarify this sound, regardless of the loudspeakers used.

There are some things we won't miss, such as the softness and energy of the bass, which we get with the best designs working in class A and A/B, or the differentiation of the bodies of the instruments, which in turn is the domain of SET amplifiers. But what we get in return is extremely interesting and worth trying in systems where we think everything is already in order and "finished". Congratulations!

Technical specifications (acc. to the manufacturer):

Nominal output:
• High power outputs 1+2: 2 x 470 W / 4 Ω | 2 x 220 W / 8 Ω
• Low power outputs 3+4: 2 x 120 W / 16 Ω | 2 x 150 W / 8 Ω
Frequency range: 3 Hz-60 kHz
Voltage gain: 32 dB
Input impedance: 50 kΩ
Transient Intermodulation: 0.003%/10 W, 8 Ω
THD+N: 0.003%/1 W, 1 kHz, 8 Ω
Efficiency @ 250 W: >81%
Idle power consumption: 30 W
Dimensions (W x H x D): 360 x 70 x 360 mm
Weight: 7.6 kg



Reference system 2021



1) Loudspeakers: HARBETH M40.1 |REVIEW|
2) Line preamplifier: AYON AUDIO Spheris III Linestage |REVIEW|
3) Super Audio CD Player: AYON AUDIO CD-35 HF Edition No. 01/50 |REVIEW|
4) Stands (loudspeakers): ACOUSTIC REVIVE (custom) |ABOUT|
5) Power amplifier: SOULUTION 710
6) Loudspeaker filter: SPEC REAL-SOUND PROCESSOR RSP-AZ9EX (prototype) |REVIEW|
7) Hi-Fi rack: FINITE ELEMENTE Pagode Edition |ABOUT|

Cables

Analog interconnect SACD Player - Line preamplifier: SILTECH Triple Crown (1 m) |ABOUT|
Analog interconnect Line preamplifier - Power amplifier: ACOUSTIC REVIVE RCA-1.0 Absolute-FM (1 m) |REVIEW|
Speaker cable: SILTECH Triple Crown (2.5 m) |ABOUT|

AC Power

Power cable | Mains Power Distribution Block - SACD Player: SILTECH Triple Crown
Power (2 m) |ARTICLE|
Power cable | Mains Power Distribution Block - Line preamplifier - ACOUSTIC REVIVE
Power Reference Triple-C (2 m) |REVIEW|
Power cable | Mains Power Distribution Block - Power amplifier - ACROLINK Mexcel 7N-PC9500 |ARTICLE|
Power cable | Power Receptacle - Mains Power Distribution Block: ACROLINK Mexcel 7N-PC9500 (2 m) |ARTICLE|
Power Receptacle: Acoustic Revive RTP-4eu ULTIMATE |REVIEW|
Anti-vibration platform under Acoustic Revive RTP-4eu ULTIMATE: Asura QUALITY RECOVERY SYSTEM Level 1 |REVIEW|
Power Supply Conditioner: Acoustic Revive RPC-1 |REVIEW|
Power Supply Conditioner: Acoustic Revive RAS-14 Triple-C |REVIEW|
Passive filter EMI/RFI: VERICTUM Block |REVIEW|

Anti-vibration

Speaker stands: ACOUSTIC REVIVE (custom)
Hi-Fi rack: FINITE ELEMENTE Pagode Edition |ABOUT|
Anti-vibration platforms: ACOUSTIC REVIVE RAF-48H |ARTICLE|

Isolators:
  • PRO AUDIO BONO Ceramic 7SN |REVIEW|
  • FRANC AUDIO ACCESSORIES Ceramic Classic
  • HARMONIX TU-666M "BeauTone" MILLION MAESTRO 20th Anniversary Edition |REVIEW|

Analogue

Phono preamplifier: Phono cartridges: Tonearm (12"): Reed 3P |REVIEW|

Clamp: PATHE WINGS Titanium PW-Ti 770 | Limited Edition

Record mats:
  • HARMONIX TU-800EX
  • PATHE WINGS

Headphones

Headphone amplifier: AYON AUDIO HA-3 |REVIEW|

Headphones: Headphone Cables: Forza AudioWorks NOIR HYBRID HPC