This review page is supported in part by the sponsor whose ad is displayed above

Reviewer: Edward Barker
Turntables: Scheu Premier II, Garrard 301, Garrard 401, Kuzma Stabi, Systemdek Transcription, Thorens TD320, Thorens TD160
Arms: Shroeder DPS, Cartridge Man Conductor, Hadcock 242 SE, Ortofon 212, Mission 774, ET2
Cartridges: Allaerts MC2 Finish, Cartridge Man Music Maker 2 & 3, Koetsu Urushi, Madrigal MC1, Empire MC1000, Shure V15
Phono amplification: Tom Evans Groove Plus, Tron Seven [in for review], DIY Lite Audio, Garrard Missing Link II, Gram Era Gold V; Altec 4722 stepup transformer
Digital: Resolution Audio Opus 21

Preamp: Canary 903
Power amps: Canary 339 monoblocks with WE 300Bs, BBC AM8/4A monoblocks, Alter Philips monoblocks, Beard 100, various Rotel and Yamaha solid state integrateds
Speakers: Living Voice OBXRW, DIY project open baffle Isophon, DIY project horn with TAD 4001, Martin Seddon Le Cleach horns, Coral mids and high frequency units (suggestion for bass units welcome)
Ancillaries: Clearlight Audio NFT cabling; Silver Arrow cabling and mains leads; Audiomagic Mini Stealth conditioner for digital; Incognito wiring on Conductor and Hadcock 242, Living Voice Mystic Matt, Boston Audio Graphite Matt, Cartridge Man Isolators and setuo tools, Dr. Feickert protractor
Review component retails: $8,295

Subjective audio reviewing is an art form that can well generate a good deal more heat than light. How many times do we read an article -- or even write one -- only to feel not much more enlightened than before? The problems are several of course. If only we had measurements which were calibrated to our own ears, taste and mood so we could find out that x component gives us 98% of live sound at x price. Wouldn't that be fun? We could all go home and concentrate on the music, with only a minimal effort required to putting the system together.

Unfortunately, that's never going to happen. Science of course is an attempt to replace qualitative distinctions with quantitative ones but one of the most difficult aspects of this endeavour is to find the right experiments, the right questions and the right means to answer them. To date, the experiments we use in audio simply don't correlate particularly well to the subjective experiences we call listening. It's then really fairly pointless to say that a good sounding but bad measuring amp measures bad. Rather, the discrepancy between what we hear and measure is simply an indicator of how inadequate our measuring techniques are.

The fact is, no matter how we go about it -- and as long as we are humans subject to our moods and fancies -- we are going to continue being stuck in the audio maze. If that's the case, why not enjoy it? Sure, we all curse the expensive mistakes but those experiences and scars leave us stronger. Hopefully.

In the course of this review, I've got a few simple goals - to give a sense of the differences I can perceive between this Canary 903SE preamp and what I remember of its predecessor, the 803 I owned before trading up to the 903SE. I'd also like to characterize its sound in general and as an aside, try to make a case for the valve preamp as an essential and crucial part of the system that makes what I hear as 'beautiful sound'. And it turns out that each of these goals is far harder to achieve than you'd credit me with. Okay, enough of the complaints...

To audiophiles, it may seem an odd thing that classical music fans can get so obsessive about different performances of a piece that may vary in the most subtle ways while they listen to it on a cheap boom box or radio. So many of the auditory cues and the gear's ability to produce a really large emotional impact are severely compromised. Of course, once a composition is being replayed over high quality components, the quality of the recording gets more important and then the performance itself becomes key. Yet as all audiophiles know, it's not just the performer that is central. The strange thing is, even the type of instrument reproducing the music can have a profound impact on the performance as well. I've recently got into period instrument recordings and somehow, the unique timbre of these ancient instruments brings a whole new level of enjoyment for me, reviving the pleasure of listening to music I thought I'd pretty much permanently lost interest in.
There's a view circling about that the most neutral preamp is a pair of passive transformers and a volume pot and that anything else, in particular a tube preamp, has in some way an additive function and as such could even be see as an interpreter. I completely disagree based on a couple of issues. First, there is a huge conceptual problematic around the concept of neutrality (neutrality being defined as having minimal effect on the signal path). In practice, it seems to me that the linear concept of neutrality isn't a particularly neutral one. The signal begins and then passes through a series of stages, each of which alter, distort and degrade the signal slightly, especially when these stages filter and amplify. This is a complex dynamic full of feedback and thus not a linear one at all. A volume device has a dramatic effect. The lesser ones will smooth out the inner core of a note's vibrations and instead present a sound that is flat and even; that has lost its inner dynamics and instead appears like a color-by-numb(ers) presentation.

My basic recipe for a good preamp is to start with tubes, add a tube power supply, keep the circuit as simple as possible and eschew remote control. Everyone at some point believes that they can design a preamp with remote that won't sonically damage things. If they are driven first and foremost by sonic considerations (which is a big issue because most audiophiles insist on a remote) they give up. Or they keep trying but you certainly won't find a remote on any Canary preamp. Nor Tron. Nor Tom Evans. Nor Lavardin. Nor Kondo. One starts to get the picture. Let's instead just get a little exercise, shall we?

I'm philosophically fairly hostile to cables and really unhappy with the idea that they make an audible difference in the signal path. But I have to admit I've heard it and I'm pretty confident it wasn't psychoacoustic auto suggestion because I'd much rather not have heard it. Of course connectors will have a big effect too but it is extraordinary how much of a difference tone arm cables can make. In fact, the nearer to the source and the smaller the signal, the more crucial it is to stop this smearing of the inner core of the note.

From what I've heard, the large majority of preamps produce something similar to this smearing effect and the result is simple. Where we ought to have a sound that is vibrant, holographically three-dimensional, pulsating, rhythmically lightning fast and timbrally drenched in colour, in fact most of the time we get a well put-together and believable reproduction that has completely lost those qualities. It is no longer vibrant but merely tonal. It is flat, boring, slow-footed and pale in timbre. The culprit? Let's swing 'round and look at the preamp.

If CD is the main source, it's at its output that the signal is at its most delicate. If it's LP, then we take a step back and the phonostage becomes the most crucial stage. Traditionally, the idea of a preamp was to provide a small amount of gain to get several components up to line level, allow for switching and for volume control. But we wanted to believe that each of these roles (or at least the switching and volume aspects) have only the tiniest of audible effects. I'm sure this is wrong. The signal is weak at this point and just pushing it through two breaks, a couple of cables and connectors, a switching device and an attenuator is going to produce that blurring and smearing effect right there. I've had one of the finest passive silver transformer-based preamps available today in my system and that is what I heard. A crucial chunk of inner detail just blurred out of the picture. The inner vibrations of the notes were gone. That's what I've heard when people go from a digital or even analog volume control directly off a CD player. It's not that the signal has been passed on in its purest form. No, it's got mangled already and lost the magic. So when you put a good preamp back in the system and listen to it, the reality is it's not adding color and beauty, it simply causes less damage to the signal in the first place.

Canary is not a well-known company in the US but as a boutique operation, they are characterized by a concentration on extreme audio and really surprising value for money. The 903SE represents their top of the line attempt to make the finest preamplifier available. It's pretty close to an absolutely no-compromise design and asks to be judged as such. From an engineering point of view, the 903 makes no effort to hide the fact that it is designed to take a series of principles to their most extreme logical conclusion. You simply can't achieve greater channel separation than with a completely dual-mono chassis in the line stage itself with its incredibly delicate signal. But the power supply is of almost equal importance and it is potentially noisy. The best thing to do is to separate it into a separate chassis. Yet Canary being Canary, they take the dual-mono thing to its logical conclusion and separate the power supply into two dual-mono chassis as well. As far as I'm aware, the only other preamp that has ever done this was a Jadis some years ago. We should remember that as long as production components are being used, the biggest cost in an amplifier will be the box itself. Giving you four of them is a pretty serious commitment on the manufacturer's part. I have heard of a number of UK amplifier designers who have examined the Canarys and simply no idea how they can supply this quality at the prices they do.

On unpacking the 903, the first thing we notice -- beyond the fact that the four boxes are a whole lot larger than standard components and will never fit on two 19-inch shelves (be warned, they really are massive so in all likelihood, you will need to do some re-arranging with your rack) -- are the rather discrete champagne fascias with their elegantly silk-screened logos as well the engraved "Handmade in California" slogan so characteristic of Canary. Each of the four boxes sports a single blue power indicator. The line stage section contains a 24-step attenuator with the first step being off (useful as a mute though not direct of course) and a four-position selector switch (none of whose inputs are labeled phono, sadly). There is CD, tuner and aux 1 and 2. I hear Canary is designing a phonostage. I frankly hope that the next top-line preamp iteration does include a phonestage. The power supply boxes sports the master on/off switch and that's it. Round back you get high-quality gold-plated binding posts, two outputs for biamping. I notice that the connectors between power supply and line stage umbilical have been redesigned. It's easy to get used to the dual volume and selector knobs. Obviously having dual attenuators means you can use them as a balance feature but I would recommend solving any channel imbalance problem at its source.

What else has changed from the 803? The crucial difference is in the power supply which is much heavier than the 803's. It now sports two transformers and a choke and the addition of a third valve, a 5Y3GT. The transformers are much heavier and exude real quality. The circuit boards are made of fiberglass, the signal path is milspec (as indeed is the whole package), and build quality is on a level I've seen rivalled only by Tron. From the Teflon pads isolating the transformers from the power supply chassis to the gold and Teflon tube sockets, it's hard to identify what could have been done better. Tube components include the aforementioned 5Y3 completed by a 12BH7 and an EF86 in each channel of the power supply. The line stage uses a 12AX7 and 12AU7 per channel.

In use
Fire it up and leave it running for say 50 hours to break in the transformers. Then on for a listen. The first thing I notice is that just as with the 803, I don't really get enough play with the stepped attenuators. With CD, we are down to using only the first 8 steps. With vinyl, it will depend more on the phonostage gain but I'd like to see a different set of resistors used to give a lower taper and wider degree of control over level. This is a typical problem with stepped attenuators and still present with the 903. Some people may find it a problem. Overall, I can always find a comfortable volume so personally, it's not a big issue. [Running two 12AU7s per channel should drop down the overall gain, shouldn't it? - Ed.]

One of the things I like about Canary is that how things are put together and the decisions behind them are strictly about sonics. Each unit is a master class in what's possible. You will not find remote controls on Canary preamps. You will not find digital display circuitry. You will not find a solder joint that isn't perfect. You will not find dubious trim pots, balance knobs or absolute phase switches in the signal path. There's nothing contrarious to absolute maximum performance.

It's common for audiophiles to talk about transparency and signal degradation and this is a concept I believed I understood. Yet it was only after being exposed to really extreme audio components and systems that I gradually became aware of how much of what I had considered to be a perfectly good preamp was in fact still blurring the signal in musically damaging ways.

This Canary preamp is characterized by fabulous liquidity and finesse. It is incredibly revealing but never edgy or bright. Instead, it majors in delicacy and speed, with this really remarkable transparency combined with a really palpable and physical quality. Sonny Rollins sax in The Standard Sonny Rollins really hits you with its physicality while never sounding bloated or heavy. It sounds a lot like it would in real life, with the reedy transient and the welter of harmonics spraying out. Sure, it doesn't have quite the color you get standing a few feet from a live sax but I wouldn't think you'd be losing too much if you fed that signal through this preamp. Another characteristic I like a lot is that it can go huge in scale without ever bending or bloating things out of shape.

With Haydn's complete Quators by the Quatour Festetics on period instruments [Arcana A411], there is a quality to the strings that is almost palpable. It's like you can touch and rub the vibrations between your fingers. This is the absolute opposite of the bland and washed out. But room ambience and echo also contribute to the quite convincing sensation that these three-hundred year old instruments are there with you. Every aspect of the Canarys' performance contributes to this - their timing and poise and their great ability to communicate the inner workings of the vibrato of a note; their punch and presence, powerful and sinuous but always supple. Transients have just the right combination of sharpness and air. The transparency is such that room ambience and acoustics are preserved to remarkable effect. That is key to why a system with the Canary 903 in it will sound palpably and believably more lifelike in almost all circumstances.

To some extent, it's slightly disingenuous to talk of the sound of a preamp that uses 12AX7 and 12AU7s. Valve rolling affords an incredible plethora of options and possible sounds. In particular being able to work with the gain relationship between the12AX7s and 12AU7s means that an enormous degree of fine tuning is possible. I've no doubt that this is one of the great attractions of this preamp.

Joshua Breakstone's Self-Portrait in Swing provides a good example. Using a good passive preamp, the guitar will sound excellent but a bit flat and to my ear CD-like. Insert the Canary and Breakstone's guitar will bloom. It will come out of the flat background and take shape in the middle of the room with vibrations emanating in all directions. It will feel palpable, with a realistic weight. Its timing will be intimate and eager and tinged with some slightly unnameable longing. Remove the Canary and we are back to a rather banal virtuoso. With the 903, the drum stick hitting the high hat has an incredibly in-the-room sensation, not quite as though you can tell what type of wood the drum stick is made out of, but more that you know it's really hard and the high hat itself is of good quality and not dented. That is, it has a crisp, newly minted character which is immediately and obviously revealing.

Change over to Miles' Steaming and Philly Joe Jones' high hats become way more present again, with more air and harmonics but not quite as revealing of their character and age. Either way, they are present and real enough to easily do the job of fooling you they are in the room. I go to Jazz evenings fairly regularly and don't feel I'm missing much. No, it's not an objectively 'real' sound but the quality of realism is such that pretty much all the musically and emotionally aspects of the sound are there. The reality is that they aren't but a good system -- and in particular a good preamp -- will do such a good job that it will fool you into filling in what's missing. That the Canary does in spades.

To get a contrast, let's substitute the Canary with another preamp. This is valve preamp that I love and that has won several "best product of year" awards. What happens? The sound becomes smaller in overall scale and considerably less dense. In other words, it sounds blander, more like reproduction than being part of the original experience. The musicality is still there, the high hats still sound excellent but now it is their rhythmic role that is highlighted rather than the full spectrum of the harmonic coloratura.

The sound is still beautiful (which is a remarkable achievement in the first place), but it's much blander and far less real. Several layers of ultra-fine information that go into making something sound real are gone. Make no mistake, it's still a great sound and comparable to what I'd describe as really good sound coming from an excellent system at a show. It's also just a bit fast and loose with overtones so notes don't appear quite as on key as they should. There's a bit of flatness in the timing and there's not quite the crispness or almost complete lack of mechanical and electronic signature that the 903 exhibits.

Cymbals are an immediate giveaway. They don't have the majesty and scale, the richness and tonal complexity of harmonics. It's a simplified instrument and sound. And it sounds tinny, much more ethereal than in the flesh. Initial transients are still excellent but on Chet Baker's From the Berkeley Sessions, they just aren't able to get into the same groove with the sax that happens with the 903. Yes, I could live with the cheaper preamp but I'm really happy that a component of the quality of the 903 is being made. The 903 makes the difference between a lovely reproduction and something much closer to a heart-gripping sensation that you've been transported on a flying time carpet right to a stool somewhere in Chet's recording room.

Put on Abdullah Ibrahim's African Sun, and where has the silk gone, where's that luxurious smoothness and dynamic seduction? Instead, the sound is harsh, brittle and quite unpleasant. Almost ear-piercing. The sound volume has to come down to be tolerable. Very high piano notes are 'dinks' rather than 'diiingggggs'. Hmm. No, it's got to come off. True, this is a strident recording on most systems but not if I put the 903 back in. Then it's on the edge but incredibly palpable. No, I find that while in percentage points, there's not a lot of difference, but in fact the difference in gestalt is that between a recording and an experience.

Compared to the 803
This is an easy one. For those coming from the 803, you'll find there's an extra level of detail, a deeper and richer musical experience. It's more three-dimensional and sensuous, finer, more textured, silkier, smoother and truer. Dynamics are faster and richer, timbres deeper and more profound. It has a quality that only truly great preamps have, which is to bring a sense of utter poise to the music, a layer of timing and texture, of presence and aliveness that is absolutely seductive. Once you've heard it and lived with it, you can hear preamps that do 99% of the sound quality but miss this elusive ingredient. And you think, boy, I could live with that cheaper sound, only to find that a few minutes later, your attention is straying.

The general presentation is the same and one of the hallmarks of this design .There is a consistency pretty much regardless of what power amp one uses. It's why I prize this preamp so much. Scale seems exact, that is, true to the source. Female vocals shift in size depending on the recording. This is not a preamp that gives its own interpretation on top of what comes before. It is among the least clouded I've heard. It is absolutely clear. Since I personally only have time and ears for valve preamps, it's something to revel in. It has all the vibrancy, all the tonal and timbral richness of valves and all their spatiality so that guitar highs and rim shots flare out like instant audible aurora borealii across the room.

I mentioned in my visit to Living Voice that my system with the Canarys sounded fairly entry-level by comparison. And it did. I pretty much put that down to the difference in Kondo amplification (my personal amplifiers above). Well, I was in some important ways wrong again. A significant portion of the log jam turned out to reside in the speaker. Had I written today's review before the arrival of the Living Voice RWs, I would have said, hand on heart, that while the 903 is my second favorite preamp, comparing it to the Kondo M77 in another system, it still sounds like the difference between a real Jackson Pollock and the poster. And that would have been an accurate and honest assessment. Most of us would have thought, the magic is basically in the amp. Well, put the RWs into the pot and it's a whole different picture. No, my Canary push/pull 300Bs don't turn from pumpkin to prince but they are not embarrassing themselves like before. At a guess, right now, if forced to choose, I'm pretty sure that I'd rather live with the OBXRW version of the speakers and the Canary amps than the older OBXR2 speakers and the mighty Kondo amps. Had the timing not been right, I would have never known that. Which goes to show how much of a difference the speaker makes. And it goes on to show how easy it is to get things wrong because we don't know what we don't know.

No, the 903 is not the Kondo M77 but now I can begin hearing the differences. It's a cruder, more metallic picture, without the serenity, peace and stature of the M77 but musically, it's by no means banal. It has insight; it is able to communicate the passion. Unfortunately, I'll have to finish this review before the M77 arrives but it would have been really interesting to have listened to these preamps side by side.

Among all the manufacturers making high-end preamps using off-the-shelf components, the Canary 309 sets a remarkably high standard from a sonic point of view, from the brilliance of its design and the quality of construction and last but not least when judged by value for money. Please remember that currently (June 2006), Canary doesn't have a distributor in the UK where I live (and I'm not sure what the US situation is). It does mean that people will need to arrange for their own servicing. In my opinion, a high-quality valve preamplifier is the crucible of an audio system and there aren't that many preamplifiers out there that are quite as rewarding. Not to mention very, very few that surpass the four-box Canary Audio 903.
Manufacturer's website