Updates and where to find them are listed here
1. 22/23 November: Added comments to 'Part (a) Highs' based on listening with my old Stax Lambda Pro. Obviously, the Corda Concerto was not involved.
2. 20/21 December: See post 48 (p.4) for tentative comments of Beyer T1 with Stagedac/Concerto versus Bifrost/Lyr.
The following review compares a non-usb Bifrost with an Eastern Electric MiniMax DAC (tube switched out). In a follow-up post, I will compare the Bifrost and Stagedac.
These units were connected via TOSLINK to a MacMini with mostly ALAC and WAV material hosted in iTunes. Where noted, some material was iTunes purchased (256Kbit). There was no special cabling - about cables I am in the “don’t know” camp - except that I use a New Zealand made ‘twisted pair’ interconnect called Taranui. It uses high quality plugs, looks good, and cost only slightly more than a department-store cable.
Amplifiers used were the Corda Concerto and, in part (b) of this comparison [still coming], the Schiit Lyr with stock GE 6BZ7 tubes.
Phones used were Audez’e LCD2 rev 1 and Beyerdynamic T1. In part (b), my aging Stax Lambda Pro makes a cameo appearance.
Both DACs were calibrated using pink noise to listening levels of 80 (actually consistently around about 80.4) dBA on my SPL meter (+/- 1.5dB claimed accuracy @ 1KHz).
The Bifrost delivers 2.0Vrms and the MiniMax 2.5Vrms. With the Concerto’s 8db gain (high gain mode) used for all the listening in part (a), this difference corresponded to 2dBA. Conveniently, this was ‘3 clicks’ out of 64 on the Concerto’s volume control. This was an easy adjustment to make when switching DACs, thanks to a LED adjacent to the volume control. The LED toggles between lit and unlit with each ‘click’. This previously slightly puzzling feature of the Concerto now makes complete sense.
Part (a) was conducted at 36-80 hours of active burn-in (see below). Assuming factory testing occurred as described, 24 hours can be added to this figure. Hours left turned on but with no music playing would add another 20 hours.
Next, I get straight into the details of listening and what I found. Read these details as what differences ‘sounded like’, rather than ‘what they are’ necessarily. Of course, this is all to my ears, IMHO, YMMV and so on. This said, I only report ‘large’ differences. The smaller or more subtle it is, the less it's worth reporting. Variation in hearing and perception from person to person will soon mask small effects.
Because it could strike some readers as unnecessary, boring or self-indulgent, I have placed details about me, my preferences and how I approached this review at the end. Indeed, skipping this part is recommended!
I have no view concerning burn-in. I am happy to follow designers’ recommendations whether to worry about it or not, and how long to use.
In this case, my first impression of the Bifrost (see below) led me to check Hero Kid’s excellent inventory of Jason’s announcements and responses. Although it struck me as a fairly casual issue to Jason - he had to be asked after all - his answer on this point was that all units are burned in for 24 hours at the factory, and their amplifiers benefit from 50-100 hours after that. He suggested a similar period would apply to the Bifrost.
Burn-in used mostly various iTunes playlists, and in one case two hours of pink noise.
I may have noticed subtle changes in the Bifrost’s sound beyond 36 hours, but I wouldn’t bet on this being the electronics. It could just as easily have been me learning its sound.
Listening impressions part (a) - with Meier Concerto
The Bifrost arrived one or two days sooner than expected. I had just listened to five tracks earlier that morning. With the Bifrost barely warmed up, I listened again. I had not equalized levels at this stage; this was not an A/B. Based purely on audio memory, my impressions were the Bifrost was exceptionally clear, had excellent attack, seemed over-bright and a touch thin, and - straight out of the box - just might be a little wearing with extended listening.
At this point all sense of over-brightness and/or thinness had gone. Clarity remained, attack featured less, and the unit now sounded very natural. At this stage, I started serious A/B listening.
Because I established brightness had “gone” with the LCD2 rev 1 - which is sometimes described as ‘dark’ - I re-checked using the T1 a few hours later. Again, the Bifrost no longer seemed bright to me compared to the MiniMax.
However, I shall come back to this point.
Sound-stage. I’m not really a sound-stage sort of person, especially with headphones. Even when I do notice, it soon disappears from my attention as I get into the texture and weave of the music. Nevertheless, the following was an interesting experience.
When I decided to start serious listening, I cued my first track to the MiniMax. Several repeats later, I switched to the Bifrost. “Aha”, I thought, “the Bifrost has a much wider sound-stage, and focuses much more precisely too”. Somewhat elated at how easily this first ‘difference’ had shown itself, I switched back - after a few more repeats - to the MiniMax.
“Aha”, I thought, “the MiniMax has a much wider sound-stage, and...” - well, I’m sure you get the idea.
I declare this a dead heat. Both the MiniMax and Bifrost do sound-stage very well to my ears. As I say, I’m not really a sound-stage guy, so if others reliably hear differences or pick up subtleties I don’t, I won’t argue.
Speaking strictly for myself, I found this ‘first-change’ or ‘novelty’ effect interesting, and something to watch out for. By the way, the effect disappeared after a while. I guess I learned to screen it out.
Detail. With one qualification which I detail below in ‘differences’, I found both DACs very detailed and resolving. Otherwise, I could not identify any reliable differences at all.
This is the crux of this comparison. In some ways, the differences are easy to describe; in another way, somewhat hard. I tried hard, but could not avoid a flowery analogy at the end.
Bass. The MiniMax does bass just slightly better. The difference is significant enough a bass-head just might prefer the MiniMax. It is easiest to detect with low-tuned bass-guitar and drums. With bass-notes, the Bifrost just seems to play the string’s core note, whereas the MiniMax presents more of the textural and timbral change as the note decays on the string. With drum, the Bifrost focuses more on the ‘tap’ or ‘beat’ of the drum stick on the skin of the drum, whereas the MiniMax reveals this and more of the ‘boom’ of the body. In one track, with kick-drum, the reverberation or resonance seemed to be picked up and repeated by the platform or stage. The Bifrost presented this information just as capably, but it’s the MiniMax that renders it more viscerally.
Two more examples are worth citing. Both DACs revealed the ringing of the body of a cello (Susan Sheppard, BachCello Suites, iTunes plus) equally capably, yet the MiniMax prolonged the ringing of the lower notes, the Bifrost the higher notes and harmonics. This was a somewhat unique listening experience.
This was equally evident listening to sections of Beethoven’s piano sonatas (no 29 and 32, his late work which frequently features hands wide apart on the keyboard., i.e. lots of low bass and upper mid combined). With the sustain pedal held down, the MiniMax holds on to the lower notes for longer, and tends to add a ‘richer’ sound to the chords than the Bifrost. However, the Bifrost produces the upper harmonics more articulately.
Which DAC is the more accurate or neutral sounding? Although I played piano for many years, it’s impossible to say. My suspicion is the MiniMax has been deliberately voiced on the warm side. It slightly exaggerates the bass and lower mid, and just slightly ‘swamps’ information the Bifrost may be presenting with more balance. The effect of this is to make the Bifrost seem the brighter of the two on a lot of material, or just ‘clearer’ on other material.
Clarity and ‘blackground’. The Bifrost has retained its initial impression of clarity throughout my listening. As I think my notes above suggest, the MiniMax has a warmth that adds just a touch of lushness. To repeat, the tube was left out of circuit throughout. Although I really don’t think there is any real difference in detail, the Bifrost can sound the more detailed.
The notion of ‘blackground’ - of dead silence between notes - seems to appear most often in discussions of some (usually solid-state) amps. Yet this is an easy description to apply to the Bifrost. When there really is a pause in the music, sound emerges noticeably more suddenly with the Bifrost. When the music is playing continuously, there is still a sense of sharper, clearer edges or better definition to the notes. I hasten to add this is not ‘digititis’. The Bifrost sounds very natural. Again, if I have pinned things down accurately in the section above, it is likely the MiniMax sustains bass and lower-mid notes and resonances longer. These play over the top of and ‘swamp’ micro-details and micro-textures which are there but one has to listen harder to hear them. The Bifrost does not have this ‘problem’, revealing these fine textures and details with relative ease.
Highs. I hesitated slightly whether to include this note, as it’s more subtle than what I’ve described above. However, I think there is just enough evidence to say the Bifrost does detail in the highs just noticeably better than the MiniMax. There is a fraction more detail: for example, more splash with cymbals, more definition (and ‘sharpness’, where near walls are involved) in the echo and reverberation of the venue, more air in general.
Update 1- Stax Lambda Pro comparisons: Use of the Stax 'phones did make some differences easier to hear. With slower, less energetic tracks especially with less direct high content e.g. from hi-hats, any difference - if it is detectable at all - is a subtle one of being more "lit" (Bifrost) or slightly darker (MiniMax), as noted in the original summary below.
There is an energetic hi-hats layer in Feral (Radiohead, WAV) present through most of the track. Quite noticeable and not at all subtle was greater extension, shimmer and hence 'detail' with the Bifrost. The MiniMax contrastingly seemed 'dull', almost as if the cymbals were being held.
The fast, energetic guitar/percussion track Hanuman (Rodriogo e Gabriela, iTunes+) brought out another aspect. The MiniMax gave more body to the guitar chords, whereas the Bifrost put more emphasis in the higher-frequency components of the percussion. Bifrost could thus seem to have more attack, although I am satisfied both were equal in this area. Interestingly, both 'versions' were equally terrific to listen to. I suspect this might be true of many energetic tracks, which provide enough fast-paced material that each DAC highlights a different aspect of what is in the music, rather than seeming to have anything missing.
Summary. As a final observation, my extended listening sessions often made me feel the MiniMax could be described as relatively dark, the Bifrost as ‘light’. I have deliberately avoided repeating the word ‘bright’ as I don’t think that accurately describes the Bifrost. Both DACs are very detailed and resolving. Yet there are differences between them: subtle but noticeable. With the exception of what I noted in the low bass, the same information can be heard with both DACs.
The difference - especially from the upper mids on up - is one of emphasis, and here I cannot resist a visual analogy. Picture a clear sunny day, perhaps in the fall. Everything can be seen clearly, but the MiniMax puts things in the shadows where they are lit more softly and perhaps a touch more romantically or - if you prefer - ‘organically’.
On the other hand the Bifrost puts things right out there in the sunlight. I’m not talking harsh, midday sun here; just crisp clarity. There is light and shade, and there is texture and there is detail.
I am surprised by some of these findings with my “usual” rig. The MiniMax was my preferred DAC in my headphone rig, partnered with the Concerto most often, the Schiit Lyr nearly as often, and both usually driving the LCD2. I usually described the MiniMax as transparent, detailed, resolving, and especially “lively”, driving the Concerto in particular with a lot of life.
I am now inclined to think the Bifrost has the better synergy in the Concerto - LCD2 chain. I am no longer sure I can describe the MiniMax as ‘transparent’, or perhaps I mean it has a very slight but musical coloration. On the other hand, in my listening so far I find this suits the (Concerto-driven) Beyerdynamic T1.
Is the Bifrost ‘as good as’ or even ‘better than’ the MiniMax? The MiniMax is a mid-priced and perhaps mid-fi DAC but it has won a reputation as something of a giant-killer and a real bargain. Reportedly, the recently released MiniMax-Plus may be better, sounding more detailed and having better blackground. In other words, bringing it closer to some of what I’ve described about the Bifrost.
I really cannot answer this question: I have far too little experience with DACs for one thing. What I can say is the Bifrost seems to me to be a different DAC with different qualities from the MiniMax but the same capabilities and no faults or weaknesses that I could detect. This seems to me a really fine achievement for a first DAC from a young company (albeit a very experienced design team) at this price-point.
Some may recognize the qualities they are looking for in the MiniMax after reading what I’ve described above; others the Bifrost. That, after all, is what a review should be about in my humble opinion.
Listening impressions part (b) - with Schiit Lyr
[due late November]
Boring bits about me
I have been a “speaker audiophile” since 1985, when I bought my first component system. Although I bought a pair of Stax Lambda Pro in 1992, and Grado RS1 about four years ago, I did not really take headphone listening seriously until I heard the LCD2. I am still a speaker person, and consider I am inexperienced with headphones.
Incidentally, considering highs with speakers tend to attenuate rapidly in a typical listening room, this may explain why the LCD2 sounds ‘natural’ to me. That said, my speakers ironically seem to significantly out-perform the Audez’es in this area, at least with much material.
I am not a bass-head, but considering my preference for Beethoven over (for example) Mozart, and the fact my Stax Lambda-Pro “didn’t do it for me”, it might be said I like bass. I appreciate the bass texture available with the LCD2 for example. I played piano for many years - not professionally but good enough to be able to play most of what I wanted to, and to accompany some of my classically trained friends - with far better skill levels than me I would add. I used to enjoy improvising for many hours.
I regard piano to be the one instrument I know intimately. Actually, to this I can add I’m pretty good at unscrambling words in recorded conversations of 4-6 people talking simultaneously, together with noting inflections and other changes in their voices. This needs a distinct set of listening skills. No, I’m not an electronic eavesdropper. This concerned experiments conducted in our social laboratory years ago. Unfortunately, I got to spend many, many hours transcribing these conversations in excruciating detail.
Apart from classical, other genres I listen to - probably more than classical these days - are Euro-jazz (e.g. Eberhard Weber, Thomas Diethelm); blues, blues-rock, gothic, some electronic, pop/rock. Actually, pretty much anything.
Most of the work I do is pure research, and development of robust, reliable and valid measures is often a key aspect. As will be obvious from the above, my review contains none of these things: in fact, it couldn’t, as the tools I use require development of scales that are then validated with samples of raters and ratings. I am largely untroubled by this. Audio is one of the things I do to relax. Until we have the makings of a comprehensive model that can account for all the data in this hobby - e.g. how people apparently hear things they shouldn’t be able to - we have little scientific basis to say anything at all.
Nevertheless, I did try to follow some method. My first problem was to identify tracks from those I am very familiar with which showed differences between the test units, supposing I could find any. By the way, it turned out finding these tracks wasn’t hard. Once I found the first few of these, I constantly returned to them as my ‘reference points’ as I gradually brought in other material. I spent about 3/4 of my time A/B’ing tracks or parts of tracks. The rest of the time was spent on extended listening.
I also re-calibrated frequently. This was more important than I expected. To my surprise, listening levels changed by 2-3 dBA from time to time, possibly a time-of-day effect. I live in an intensive farming area with a great deal of heavy equipment. It may be power into my home was sagging come - e.g. - milking time.
Two problems I paid some attention to were: (a) effective listening stops when I get tired; (b) effective listening also stops if my mind is “chattering” (thinking - or is it those voices I can't get rid of?) while the music’s playing. When either of these occurred, I took a break.
There are many cognitive priming and other effects that, possibly, might invalidate a review like this. Without disputing the possibility these applied here - meaning in fact there is no difference between these DACs, or alternatively that I failed to hear other large differences because these effects masked them - it is interesting to note that on three occasions I thought I had falsified a difference previously found...only to discover I was actually listening to the other DAC. This made me modify my procedure to ensure I knew exactly which ‘experimental condition’ (DAC) I was collecting data for at each step.
Edited by Argo Duck - 12/20/11 at 10:09pm