Anaylsis Audio

Associated Equipment:
Front End
Digital Front End
The Antique Sound Lab MG SI 15 DT Single-Ended Class A Tube Integrated Amplifier
Sonic Beauty at an Affordable Price


February 2007



Considering the small size of the high performance audio world compared to the mega corporation-monopolized mass market, the diversity of approaches and choices available to audio enthusiasts verges on the staggering. One can choose amplifiers from a wide variety of solid state, digital, and vacuum tube designs, incorporating circuits ranging from the cutting edge contemporary to the tried-and-proven antique, and hailing from anywhere on the planet. Products are available from a variety of distinct and, at times, conflicting design schools, including the UK’s “Flat Earth” school, the USA’s ultra-priced High End, and the now world-wide production of tube-based designs. The variation in design philosophy makes system-building and system-matching an ever more demanding tribulation. Using a simple logic, an inexperienced user might try to build a system incorporating the best of these schools - the timing and rhythm of a classic UK turntable, say, with a holographically-imaging High End US preamp, mated with the glorious midrange of a 300B-based SET amp, and driving transmission-line, 84 dB sensitive speakers in an attempt to get tight bass. That such a system would most likely fall completely on its face speaks strongly of the pressing need to be aware of the underlying assumptions of the varying design schools and to incorporate them in one’s system matching efforts.

Perhaps the most idiosyncratic of these design schools is that of the low-powered Single-Ended Triode vacuum tube (thermionic valve to our UK friends) amplifier, or SET. Though in reality a sub-school of vacuum tube-based design in general, the SET school has taken on a set of design principles distinctly soaked in aesthetics-derived values of engineering simplicity, with sonic goals focusing almost exclusively on purity of tone and sonorous beauty. The result is the ‘classic’ SET system: the low-powered, single-ended, Class A, no-negative feedback, triode tube-based amplifier driving a high sensitivity, crossover-less, single driver speaker (or its corollary the high sensitivity, horn-loaded speaker.) Devotees of the SET school are often enthusiastic to the point of evangelism: it is clear that for many their experience of tube electronics and SET amps involves an epiphany so striking and so moving that a near-conversion experience results. Many SET converts often take on a single-mindedness that becomes oblivious to the limitations of the approach.

The modern SET movement originated in Japan, emerging from that strange post-World War II Japanese nostalgia for American Hi-Fi products of the 50’s and 60’s. It quickly tapped back into even older tube designs, eschewing the push-pull, phase-splitting, negative feedback designs that had become the norm for tube amps. Strongly do-it-yourself oriented (both in amp and speaker building,) the Japanese SET school DIY-ers attempted to graft Zen aesthetic values of simplicity, purity, and wholeness onto their designs, utilizing the simplest of circuits, the minimum number of parts, and focusing on the quality of the individual part’s contribution to a larger whole. Just as the finest artistic results appear to flow with an ease that seems natural and that belies the effort expended to produce those results, the SET design school aesthetic attempted to maximize the “natural” simplicity of technology, hence the reliance on Single-Ended design, Class A output stages, the focus on the inherent quality of a specific triode output tube, and no negative feedback as central SET design principles.

Given the French insistence that everyday life contain an aesthetic dimension it’s not surprising that France was the first Western country to embrace the SET mindset. It quickly spread throughout Europe and England, on to the US, and to Asia where the nascent Pacific Rim economies quickly glommed onto the technology and mindset. SET and conventional tube amps are now truly a world-wide phenomenon. As China has bloomed as a technology center, their advantage in lower manufacturing costs has allowed vacuum tube products to be produced at readily affordable prices.

Antique Sound Lab is one such Chinese firm. Broadly adept in manufacturing capabilities, the company produces its own metal work and designs and winds its own transformers. The MG SI 15 DT integrated amp reviewed here is their least expensive Single-Ended, Class A Triode design. Priced at $1100, the MG SI 15 DT produces 5 watts per channel in Triode mode from its single KT88 output tube. It can also be run in Pentode mode with an output of 15 watts per channel by the simple flick of a switch. A single 12AX7 drives each KT88; thus, there are 4 tubes in all.

The current MG SI 15 DT has changed considerably from the original unit launched on these shores some 6 years ago. It now incorporates remote control of input, volume and muting (the transmitter is carved from wood,) a choke-filtered power supply, line-level subwoofer outputs plus its 3 line-level source inputs, higher quality jacks, and a removable tube cage. Significantly, it now uses no negative feedback, and thus has a larger power supply, an upgraded power transformer, and new higher-quality output transformers that do not include a tap to incorporate a feedback loop. The 15 thus is line with the major design credos of the SET school. Its only deviation from orthodoxy is that it runs the KT88 (6550 in the USA) output tube and can be switched to run in Pentode mode.

More than any other design school, the low-powered SET amps demand that the system be built on the needs of the amplifier. Speaker choice, in particular, has to be made on the limitations of a low-powered tube amplifier. Herein lies the very big rub. The standard mating of these low-powered amps has been with the high-sensitivity full-range single speaker – often a Lowther or Fostex installed in an expensive after-market cabinet - or various horn-loaded designs of the past: Altec Voice of the Theater, Klipsch folded horns, or various high-sensitivity Dual Concentric designs. Many of these speakers achieve sensitivities of more than 100 dB with 1 watt input.

I have to be honest: I remain agnostic and undecided about the ‘classic’ 300B-based SET amps largely because I’m unconvinced by single-driver speaker designs. Giving up the state-of-the-art tweeter response I’ve enjoyed from a variety of speakers for the last 30 years is too extreme a demand. I also find the distortions of horn-loaded designs incongruent with the sound of instruments, except for, of course, horns. Indeed, the central problem of using tube amps, and low-powered SET amps in particular, is finding loudspeakers that one would want to listen to for reasons other than that they “work” with tube amps.

Defining what “works’” with low-powered tube amps turns out to be more complicated than simply choosing a speaker with high sensitivity (say, 93 dB or more with one Watt at one meter distance.) A tube amp’s bass response is related to, and varies with, a number of factors in its design - tube type, power supply, transformer quality, damping factor, power bandwidth, output impedance, the use of negative feedback, etc. How these varying factors will interact with a specific loudspeaker’s own electrical and mechanical aspects (and that loudspeaker’s own interaction with its room) is very hard to predict with any usable or reliable certainty. There is no simple and fool-proof formula, so there’s no substitute for simply trying and seeing. Uh, hearing.

The MG SI 15 DT’s limited power output (5 watts in Triode mode, 15 watts in Pentode) can easily be overcome by using a smallish room – I did much of my auditioning in an 18’ long by 14’ wide room with the speakers placed on the long wall – and limiting playback volume to non-deafening (according to revised OSSHA standards) SPL’s of 85 dB or less. Used thusly and in pentode mode, I was able to play speakers ranging from 85 to 90 dB sensitivity without any clipping. As I normally listen at these volume levels, I didn’t find the 15 DT’s smallish power output to be a limitation. Indeed the amp’s high resolution and clarity invites lower volume listening: there’s no need to go to extreme volume to obtain clarity.

Construction and appearance of the amp is strictly ‘form follows function’. Switches are robust, tube bias setting is simple, and the amp exhibited no hum, buzz, noise, hiss or any other extraneous indication that it was in the system. The owner’s manual, however, yearns for a re-write, as the “Chinglish” in which it is written can produce confusion in the reader. A front panel meter allows one to set bias easily: turn down the volume, move the bias knob on the left side of the amp to the requisite tube and rotate the bias adjustment screw with a long-handled screw driver until the meter’s indicator is centered. Repeat for the other KT88. I set the bias twice during the amp’s initial 10 hours of play; after that it didn’t need re-adjustment.

Tube amps tend to be heavy things (good output transformers tending toward the massive) and tubes themselves are microphonic, so it was no surprise that the 15 DT benefited enormously from isolation devices. Played ‘neat’ – the SI 15 resting on its own feet atop a massive oak LP cabinet that holds roughly 1200 LP’s – the 15 DT exhibited the low-contrast, foggy and soggy performance that is typical of the effects of low-bass environmental interference. Isolating the amp with 3 Vibrapod Cone isolation devices ($8 each) transformed the amp: tightening up the bass, blowing away the fog, turning up the resolution, and focusing the individual instrument’s playing and location. This effect is typical with tube-based components, indeed, with virtually all components. Moreover, my experiments with a wide variety of isolation devices in 3 different rooms indicated that one can tune the amp’s resolution to match its source quality: upping its resolution just enough that so that the illusion of music is not snapped. This was particularly useful when mating the 15 DT with less-than-perfect (aren’t they all?) CD players. I was able to use more sophisticated isolation with the amp when using the Rega Saturn and the Pioneer Elite DV 79AVi CD players, for example, than with my older, slightly cruder CD players. Using the state-of-the-art Stillpoints Component Stand under the MG SI 15 DT and playing my reference LP playback system revealed a quality of playback unimagined by cavalier placement of the amp on its own feet. I would strongly recommend potential users invest at least the minimum of $24 for the Vibrapod Cones before embarking on any experimenting with AC cords, tube rolling, and other tweaks.

The use of the KT88 output tube is somewhat heterodox to the more purist wing of the SET school, who do not consider it a proper triode tube. The KT88 and its US counterpart, the 6550, is probably best known for its use in high-powered (for tube amps) push-pull, phase-split designs, dating back to the seminal and classic Stewart Hegeman-designed Harman Kardon Citation amps of the late 60’s. Hegeman was a strong proponent of wide bandwidth as a sine qua non for first–rate audio performance, and his tube designs were antithetical to what is popularly understood as ‘classic tube sound,’ i.e., a soft, rolled-off top, murky, boomy bass, and a heavily colored midrange. The KT88/6550’s wide bandwidth (especially its high frequency performance) served Hegeman’s design philosophy well. Herb Reichert’s writings in the now defunct “Listener” magazine remarked that the KT88 performed even better in a single-ended Class A pentode circuit without negative feedback, as this design configuration keeps the tube’s THD musically consonant without compromising its detail and high frequency extension.

I listened to 9 different speakers with the MG SI 15 DT amplifier. Consistent with them all was an exceptionally natural midrange performance with an equally exceptional level of detail and resolution, extending throughout the treble. The amp was superb at reproducing an instrument’s context as well as its own sound. Unlike the “sounds emerging from a black background” distortion often mistakenly praised by muddle-headed reviewing cliché, the 15 DT reproduced the ambience of the recording site as the context from which the instrument’s sound emerged. It managed to do this even with sources not noted for ambience retrieval, i.e., the CD.

The amp was also excellent at tying an instrument’s harmonic structure to its fundamental pitch frequency and in tracking the instrument’s transient energy coherently. It maintained this ability even with multiple instruments playing – each instrument kept its transient and harmonic coherence, without the common tendency of many amps to wash this delicate information into a whitish-sounding electronic hash. The amp was completely free from the kind of harsh artificial brightness that is still the most common complaint of audiophiles with their systems. Significantly, the 15 DT achieves this without rolling off the treble or masking the problem with soft-focus: it resolves the problem, by its, uh, high resolution. It’s clear that the combination of the KT88 tube’s bandwidth, Class A single-ended operation, and the freedom from T.I.M. (transient intermodulation) distortion gained by eliminating the application of negative feedback, works wonders in offering clear and natural timbre to a wide range of acoustic instruments, revealing their all too often occluded sonorous beauty.

A standard apothegm of tube and analogue LP lovers is that “CD finished what the transistor began.” It becomes increasingly clear that the CD format is an exceptionally fragile playback medium: it walks a very tenuous sonic high wire. The slightest flaw in playback and it’s falling head-first off that wire. The MG SI 15 DT’s performance with CD was revelatory, offering a strikingly more natural sonic experience from that un-natural sounding music format. It did this not by artificially coloring or “warming” up the sound, but by not adding the same kinds of distortions that are endemic to the format and thus compounding its problems. Particularly notable was the improvement in the decay of notes, an area in which CD has traditionally been very weak, along with its corollary, the “attack” of the note. String instruments, plucked and otherwise, in particular sounded vivid and natural, free from hash and harshness. This was true with the delicate sounds of the harp, acoustic guitar, oud, and lute as it was with bottle-neck and slide guitar and Rock and Blues electric guitar. Orchestral instruments were more easily identifiable than with the average transistor amp, with a significant increase in the accuracy of timbre from string and woodwind instruments. “Original instrument” performances and medieval instruments lost their slightly alien timbres, as did the wide range of world folk musics. The amp’s high resolution decoded dense studio mixes: perception of Brian Jones’ tamboura and Dave Mason’s shenai playing in the back of the mix on The Rolling Stones’ Street Fighting Man clearly emerged from the clutter. I’d never heard that before, even after nearly 40 years of playing the Beggars’ Banquet album. Impressive.

Vocal performance with CD playback was also superb: I was able to parse words and phrases even in languages that I don’t understand (the Gaelic of Irish folk band Clannad, the Nubian of Master Oud player Hamza el Din.) The legendary tube prowess with female vocals was in full display with the 15 DT, and was even more entrancing in triode mode playback. The timbre of voices was exceptionally rich and natural with subtle inflections of phrasing and vowel and consonant formation clearly rendered. The results were the same with all the “singing” instruments. One could close one’s eyes and easily “see” the performer’s physical movements as they sang and performed.

The limitations of the CD format, while not spotlighted, were, however, perceivable, as were the differences between the 4 CD players I used. Significantly though, even the most humble of these – a physically-modified Marantz CD67SE – produced a sound through the 15 DT that would encourage anyone alienated by CD’s synthetic qualities.

It was with analogue LP, however, that the 15 DT’s full potential was revealed: the format’s more natural timbre and more communicative melodic and rhythmic drive were a true joy to experience, the sonority of instruments even more heart-meltingly beautiful. The high frequencies of a variety of phono cartridges were very well portrayed; even the typical rising high-frequency response of moving coil cartridges was well integrated into the overall sonic fabric. This was true even with the ultra-revealing abilities of speakers like the electrostatic panels of the Sound Lab Dynastats, Infinity’s old EMIT tweeters, Celestion’s metal domes, and the Harbeth Monitor 30’s soft dome tweeter. While the amp’s high frequency response might not match the ultimate best available, treble was free from hash and harshness and clearly depicted tonal colors of varying hues, rather than depicting everything in monochromatic white.

The MG SI 15 DT preferred turntables, arms, cartridges, and phono stages that were fast, linear, and controlled in the bass. Sprung-subchassis tables like my Merrill/AR and my two Linn Sondek LP12’s sounded much murkier in the bass than did my two Origin Live tables. The lead-footedness of high-mass suspension-less tables are unlikely to be a good match. In budget cartridges, it preferred Audio Technicas to Grados, and with phono stages it performed better with wide-bandwidth designs like the Graham Slees than with the tube EAR 834P. I was not able to audition Antique Sound Labs own tube phono stage – the $500 Phono Lux DT.

Psst..There’s a 6,000-pound Elephant in the Room

The Achilles’ heel of tube amps traditionally has been their bass performance. Compared to solid-state amps, tube amps are curtailed in low frequency bandwidth, have very low damping factors, and offer a high output impedance to the loudspeakers. These factors will interact with any given loudspeaker in ways that are very difficult to predict.

Working through my stable of loudspeakers was an intriguing and occasionally perplexing experience. Speakers that I predicted would not work at all with the 15 DT ended up working quite well. The old Infinity Qb and RS7 are both 3-way designs with acoustic suspension woofer-loading and 4 ohm impedances. Since the 15 DT offers no dedicated 4 Ohm tap, I expected the result to be a fiasco. Instead I got very good bass (tighter and more linear than with reflex-loaded designs), luscious mid-range and high treble results, that while not fully exploiting the ultimate treble abilities of the EMIT tweeter, could only be described as excellent.

I strongly resisted trying the Harbeth Monitor 30 with the 15 DT because I knew it was designed to work with solid-state amps and because of its low-ish sensitivity of 85-6 dB. Given that the midrange of the Harbeth Radial driver produces tube-like beauty even with solid-state amps, there isn’t much point in running tubes into them. Still, the 15 DT worked very well into the Monitor 30’s as long I kept an eye on volume levels.

The 15 DT fell apart trying to drive my Sound Lab Dynastats. Its performance on the speaker’s electrostatic panels was superb, with more detail and focus than my antique 50 watt per channel, EL34-based EICO HF89 tube amp. Unfortunately it could not drive the Dynastat’s dynamic woofers, which handle the range below Middle C down to 20 Hz in my room, to any sort of balance with the electrostatic panels. Since the Dynastat is an unusual load that few amps can drive coherently, I don’t judge an amp’s merits by its ability to drive them.

Driving classic 2-way, stand-mounted, 6-inch woofer, budget ‘mini-monitors’ (like the Rega R1, Celestion 3 MKII and F15,) the MG SI 15 DT proved very consistent, yielding beautiful tonality and exceptionally vivid and convincing stereophonic illusions, the speakers’ limited bass response not taxing the amp’s bass capabilities. The 15 DT’s performance with the increasingly popular “two and a half”-way speaker systems – speakers using the same drivers as in budget “mini-monitors,” but doubling up the woofer and installing them in a floor-standing cabinet – raised some very interesting questions.

A central limitation of tube amplifiers is their high output impedance, the effect of which theoretically is production of frequency-response anomalies in a speaker’s output. It can also affect the speaker’s woofer alignment. I measured the 15 DT’s response into the Celestion F30 (a 90 dB sensitive, floor-standing, “two and a half-way” bass reflex-loaded design that uses the identical drivers, with an additional woofer, as the Celestion F15 ‘mini-monitor’ which I have reviewed: see I compared it to the speaker’s response with the low-powered Class T, Sonic-Impact Super T amp.

Although the testing gear I used is decidedly low-tech and the results should in no way be construed as typical of the 15 DT’s performance with any other speakers, the contrast of the two amps’ response into this specific load was intriguing. The 15 DT completely eliminated the 5 dB peak at 5 kHz in the F30’s response when driven by the Super T amp and playing CD. The tube amp also produced a minor dip from 1.5 KHz to 3.15 KHz where the Super T produced a 2 dB rise. Although neither of these responses are absolutely neutral, the sound of the 15 DT when playing CD was infinitely preferable in both a sonic and musical sense. A zippiness and ‘strangled’ quality of the F30 in the treble and upper midrange was completely gone. More illuminating was the bass response of the two amps. The F30’s are flat to 40 Hz in the aforementioned smallish listening room used for my auditions. The 15 DT was actually more linear in the 40 to 80 Hz octave than the Super T, but showed a significant trough in response in the 100 to 200 Hz octave, which made it hard to follow bass lines as they passed through this zone. The on-paper advantages of the two and a half-way designs (lower bass, no need for a speaker stand) over the single woofer mini-monitor are proving unmet even with high damping factor, low output impedance transistor amps. I assume the cause is a more complex impedance load and associated phase variations. The bass results of the 15 DT with the mini-monitor F15 were far more coherent.

The 15 DT’s bass response in general was weighty and with good punch, though there was some slurring of bass notes’ initial attack transients and notes were held slightly longer than they should be. Absolute control of the bass was far worse than with typical solid-state amps, but was also far better than some contemporary tube amps of higher power and three times the cost. Still, music with complicated bass lines, or with multiple bass instruments playing, often tripped up the amp. Rhythm and timing was also affected, the amp never going beyond metronome quality in its portrayal of tempo and flow. Subtle rhythmic shifts and expressions – leaning forward in the beat, or laying somewhat back against it – were glossed over. Physical response to the music was also muted: phrasing and points of arrival were often slightly off. While the amp’s exceptional midrange purity might inspire one to sing along with the music, playing “air” drums or bass is not likely.

The timing and rhythm of midrange instruments was clearly better than that of the bass instruments, leading to variable performance with varying music, performance not dependent on genre. With drum and bass-driven music, the amp, for example, played the Rolling Stones well, but slurred Little Feat; happy with Led Zeppelin, but less happy with Cream. Still, while the range of music that worked well with the 15 DT was large enough to keep the amp from being completely a special application product, it did not open the aesthetic door to all music, a criterion which is absolutely essential for my listening habits and tastes. Because of this limitation, I think the amp’s greatest strengths are maximized as a second, back-up system, used in a small-ish room, and probably CD-driven.

There was not a night and day difference between running the 15 DT in Triode or Pentode mode. The only really obvious difference was in depiction of female vocals where Triode mode sounded slightly richer and even more organic. Since I did not find the Pentode mode’s vocal performance marred or flawed in any way, I was happy to do most of my listening in Pentode, welcoming its far larger power output.

The Antique Sound Lab MG SI 15 DT is an exceptionally seductive and tantalizing amplifier. Its midrange and treble performance capture the sonic beauty of instruments in a way that re-awakens the ear’s delight in sonorous beauty and at a level of resolution that should arouse “Eureka!” responses, especially given its affordable price. Its rendering of CD should melt the heart of anyone frustrated by that format’s unnatural aspects. Though the amp’s bass performance, like every tube amp I’ve ever heard, falls short of a good solid-state amp’s drive, timing and control, and thus will limit the amp’s performance with some music, its opening up of a wide category of music’s deep sonic beauty is a rare achievement indeed, and one well worth the amp’s affordable price of admission.

Paul Szabady


Single-Ended, Class A, No negative-feedback, Remote-controlled, Tube Integrated amplifier.
Power Output: 5W/channel (Triode)/ 15W (Pentode) User switchable.
Frequency response @ 1W: 13 Hz-77,000Hz + 1dB
Frequency response @ full power: 30Hz – 63,000Hz + 1dB
Distortion @ 1W: <0.6%
S/N Ratio: >78dB
Dimensions: 13 inch x 10.5 inches x 9 inches (W, D, H)
Weight: 29 lbs.

Price: $1100.

North American Distributor:
Divergent Technologies TEL (519) 749 1565 FAX (519) 749 2863
Telephone: (519) 749 1565
Fax: (519) 749 2863