[SoundStage!]Home Audio
Equipment Review

February 2002

Atma-Sphere MA-1 Mk II.2 Mono Amplifiers

by Marc Mickelson

Click to view measurements of this product


Review Summary
Sound "An utterly clear and uncongested view into the music" -- "these amps breathed life into vocals"; "but this clarity and immediacy could be a double-edged sword" with less-than-perfect recordings; bass "certainly lacks solid-state weight and slam," but it "has punch."
Features Class-A, zero-feedback, fully balanced OTL circuitry from someone with "over 25 years of experience designing and building OTL amplifiers under his belt."
Use Not for a speaker that requires "solid-state amplification to sound its best"; consume 500 watts each and produce a fair amount of heat.
Value "What the MA-1 Mk II.2 does, no other amp [Marc has] heard so far duplicates."

In his February "Audio Hell" installment, Bill Brooks relays a story about an audiophile who doesn't like tubes because he "likes accuracy." This underscores the broad beliefs we audiophiles have about tubes and solid state, even though none of them are universally true anymore. I would love to own separate systems that utilize each, but not because either is more accurate than the other. In general, I enjoy the sound of tubes more than solid state, but for reviewing, having a system that makes best use of solid-state amplification would be a wonderful tool.

But there are factions within each of the larger categories, and I won't be wishing to have systems that conform to each of them. In terms of tube amplifiers, the OTL camp is vocal and seems to be growing, as makers such as Tenor Audio and Naked Truth Audio have emerged to take up ranks along with Joule Electra, Transcendent Audio and Atma-Sphere. With over 25 years of experience designing and building OTL amplifiers under his belt, Ralph Karsten of Atma-Sphere is the dean of current OTL makers, and his designs are both old and new. First, Atma-Sphere OTLs are based on the circlotron output circuit, which is the basis of other OTL designs, but Atma-Sphere amps are also fully class A and differentially balanced. The latter, as Doug Schneider and I found out when we visited Atma-Sphere, is the part of his design that Karsten considers his real achievement. Atma-Sphere OTL amps also use no feedback, which many designers of tube and solid-state equipment feel to be at the heart of the best possible sound.

The MA-1 Mk II.2 is Atma-Sphere's middle child among its monoblocks. At $9800 USD per pair, it costs more per pair than the well-known MA-60 Mk II.2 amp ($4650 per pair), which is a SoundStage! Reviewers' Choice Hall of Fame award winner, but less than the current top of the line, the MA-2 Mk II.2 ($27,200 per pair; a gargantuan two-chassis MA-3 is in the works), which looks like a tube tester from around back given the number of tubes it uses. The look of all three amps is decidedly retro, with the MA-1 Mk II.2 taking the award for the best looks because of its chrome chassis and open-air tube placement. The amp is manufactured in mirror pairs too, which is a nice touch and one that requires that Atma-Sphere stock different chassis assemblies for each channel. The internal workmanship of these amps is gorgeous. No wonder the guys from Atma-Sphere sign their names to it.

After you install the 14 6AS7G output and four 6SN7GT driver tubes per amplifier, you need to adjust tube bias and DC offset, which is a matter of holding down toggle switches on the front edge of the amplifier, making adjustments to the trim pots with a screwdriver, and reading the results on the front-mounted meter. Single-ended RCA and balanced XLR connectors are on the front edge of the amp, so your interconnects may need to be a half meter longer to reach. If you use the RCA connectors, you'll need to insert the included shorting plug between XLR pins 1 and 3 or you'll hear buzzing. The all-copper speaker terminals are around back. The ones on the review sample were not color coded, but they were marked with a small + and - on the plastic skirt beneath, requiring a flashlight to determine which was which. An IEC power-cord receptacle rounds out the connections.

The MA-2 Mk II.2 is a large amp, at least two times longer than it is wide, but at 40 pounds, it isn't as heavy as a non-OTL amp of similar size would be. Atma-Sphere rates its power output as 140 watts into 8 ohms, 135 watts into 4. Its frequency response is rated at a very impressive 1Hz-200kHz within .5dB (-3dB at 1MHz). Each amp uses 500 watts of power, but if you live in the great white north of the Midwest, as Ralph Karsten and I do, you'll be happy about the heat that comes from these amps for all the power they consume -- in the winter, that is. They raised the temperature in my 12' x 24' listening room five degrees.

I had heard that there were upgrades for the stock MA-1 Mk II.2, and I asked Ralph Karsten about these. It's easier to quote him than paraphrase:

"There are two upgrades for the MA-1. The first is the Caddock resistor package ($800), which replaces resistors in the driver circuit. These are the same type of parts used in the MA-2. The Caddock resistors take away an electronic veil that might not otherwise be heard; they add a sense of speed without increasing brightness. The second upgrade is the larger filter caps that we have custom built for the MA-2 Mk II.2 and can be used in the MA-1. The upgrade calls for the power switches and power rectifiers of the MA-2 as well, as the normal ones won't survive the turn-on surge. That upgrade costs $2000. Helps with low-end authority and smoother presentation at high volumes."

As is Atma-Sphere's way, older MA-1 amplifiers can also be brought up to current Mk II.2 status. You'll want to check with the company for pricing. The review amps were current and stock with no upgrades.

Other equipment

I used the MA-1 Mk II.2 amps along with a glut of other equipment, all of which would be considered logical mates with the amps. Speakers were from Wilson Audio, the mighty WATT/ Puppy 6 and Sophia, as well as the Mirage OM-5, all of which worked well with the amps. Preamps were the Audio Research Reference Two Mk II and Lamm L2 Reference. I used a Mark Levinson No.39 as both a standalone CD player and transport, in which case it fed a Bel Canto DAC1.1. Interconnects and speaker cables were either Nordost SPM Reference and Quattro-Fil, Ensemble Dynaflux and Megaflux, or Acoustic Zen Silver Reference and Hologram. Power cords were a variety of models from Shunyata Research, whose Hydra distributed power to the components that needed it. Amps for comparison were primarily my reference Lamm ML2 monoblocks, which is a SET design -- another tube-amp sub-category -- and to a lesser extent the Audio Research VTM200 monos.

Completely dismissing the myth that OTLs are unstable, Ralph Karsten's amps are so reliable that removing tubes and even shorting the outputs doesn't cause them fits. I don't recommend that you test this, but having seen it done, I can convey my own sense of security.

Sound

The Atma-Sphere MA-1 Mk II.2 offers an utterly clear and uncongested view into the music. Transients have snap, percussion moves at breakneck speed, and cymbals have a lifelike, steely sheen. All of this imparts an immediate quality that will have tube nay-sayers who "like accuracy" scratching their heads. Wynton Kelly's Piano [Victor VICJ-60259] and Cannonball Adderley's Know What I Mean? [Victor VICJ-60243] on JVC XRCD rang with vibrancy and life, as did the Jacques Loussier Trio's Satie [Telarc CD-83431] on Telarc. The piano on all three discs (Bill Evans plays on Know What I Mean?) displayed the kind of attack and decay we often attribute to the sound of quick drum shots or plucked strings. The space and ambience on Piano in particular were startling in their physicality. I felt like I was in some room other than my own.

But this clarity and immediacy could be a double-edged sword. While the various XRCD and Telarc discs I played sounded beyond terrific, some CDs took on a slightly hot, shouty quality, lending an overall forwardness that made discs like Fountains of Wayne's Utopia Parkway [Atlantic 83177-2] and Roseanne Cash's 10 Song Demo [Capitol 112364] hard to listen to at my normal levels. My reference system does not lean toward mellowness, so the amps may simply be giving more of the truth than I was accustomed to hearing.

The midrange of the Atma-Sphere MA-1 Mk II.2 is as pure as I've heard from any amp. The detail of the voices was consummate and caused me to pull out vocal CDs I hadn't heard in a long while. Tom Waits, in particular, benefited from the MA-2 Mk II.2's high resolving powers and way with vocals. You may know Waits from his later, more experimental works, but Small Change [Asylum 1078-2] from the mid-'70s is the disc I put on to hear Waits at his most diverse. Texture is something Waits' voice has in abundance, and it was even more evident through the MA-1 Mk II.2 amps, as was the silkiness of Doreen Smith on A Tribute to Julie London [Fidelio FACD006]. This laid-back and spacious disc profited greatly from the MA-1 Mk II.2's clarity. These amps breathed life into vocals that I hadn't noticed previously.

While the bass of the MA-1 Mk II.2 amps has punch, perhaps due to a small perceived bump in the midbass with the speakers I had here for use, it certainly lacks solid-state weight and slam. Test discs for bass like Harry Connick, Jr.'s She [Columbia CK 64376] and Suzanne Vega's Nine Objects of Desire [A&M 31454 0583 2] didn't have the low-end growl they've had with other amps, including the Lamm ML2s, but there was still plenty of impact and pace. A disc that I've been playing for months and from which I included cuts on my CES demo disc, is the remastered version of Dire Straits' self-titled first album [Warner Bros. 9 47769-2]. I used to own this on vinyl, and I then replaced it with the CD, which was one of the first CDs I purchased. The recording is darned good even by today's standards, and the remastered CD shows just how good. When I played "Water of Love" with the MA-1 Mk II.2 in use, the initial wood-block strikes had decay that went on forever, and even the low frequencies were nothing to sneeze at. Then I heard the same track with the 200W Audio Research VTM200 monoblocks, and the bass took on greater substance, as though a subwoofer were integrated more optimally. I know from listening to the less-powerful M-60 Mk II.2 amps driving big Classic Audio Reproductions speakers that OTLs can do bass. The two Wilson speakers I had for use are almost certainly tougher loads, so I'll admit here that your mileage may vary depending on your speakers.

In terms of system matching, I had the best luck using the Atma-Sphere amps with the Audio Research Reference Two Mk II preamp, whose slight warmth filled out the presentation to a welcome degree. Like the MA-1 Mk II.2 amps, the Reference Two is also fully balanced. And while I prefer the WATT/Puppy 6es over any speaker I've heard in my listening room, the Sophias made beautifully detailed music with the MA-1 Mk II.2 amps, and the price of the combination comes in at about what the WATT/Puppy 6es cost. But the most intriguing combination was the MA-1 Mk II.2 and the Mirage OM-5. The Atma-Sphere amps brought these speakers to life like no other amps I tried them with, and the OM-5's powered bass section proved its full worth. Audiophiles may scoff at spending almost $10,000 on amplifiers for $3500 speakers, but this combination proved that there is more than one way to build a terrific-sounding system.

SET vs. OTL

For audio consumers, the main difference between OTL and SET designs may be power. While a good number of OTL amps are rated to deliver enough watts to fill a room with just about any speaker, SET designs need help from a speaker in the form of higher sensitivity. Both require attention to the load a speaker presents, preferably one that's 8 ohms nominal or above and with no steep angles in its impedance plot. I have achieved very satisfying results pairing my Lamm ML2 SET amps with a good number of speakers, including the Wilson WATT/Puppy 6 and, to a lesser degree, Sophia. The same can be said for these speakers and the Atma-Sphere MA-1 Mk II.2.

But the similarities end there, as the overall presentations of these two amplifiers depart rather dramatically. The Atma-Sphere MA-1 Mk II.2 amps put forward their clarity and ability to unravel the elements of the music, while nuance, gesture and, for me at least, that illusive sense of involvement constitute the Lamm ML2s' strengths. It's safe to say that both amps have special qualities, and which combination of these makes more musical sense is up to the listener. However, the difference in price here is immense, so a more appropriate comparison would be between the ML2 and Atma-Sphere's MA-2 Mk II.2, about which I've heard nothing but great things.

Conclusion

I like to play the speculative "Who is this product for?" game in reviews to aid in discerning a product's niche in the marketplace. An Atma-Sphere MA-1 Mk II.2 owner has refined, even esoteric, tastes, which he can afford to fulfill with an expensive amplifier. He will have a speaker that doesn't require solid-state amplification to sound its best. Merlin, Silverline, even Wilson Audio are reasonable choices; Revel and Thiel are almost certainly not (although I now brace for e-mail proving me wrong). Most of all, the owner of the MA-1 Mk II.2s will be someone taken with directness of expression, midrange purity, high resolution, and the characteristic uncongested quality of the MA-1 Mk II.2. He will also need some power for his speakers to fill his listening space, but perhaps not the 220 watts that the larger Atma-Sphere MA-2 Mk II.2 amps provide. He will enjoy the look and reliability of these amps and not be too bothered by the heat they create. He may even be wary of OTLs in general but be drawn to the reliability the Atma-Sphere models have.

Have I described you? If so, give an Atma-Sphere OTL amp a listen, preferably with your chosen speakers. What the MA-1 Mk II.2 does, no other amp I've heard so far duplicates.

...Marc Mickelson
marc@soundstage.com

Atma-Sphere MA-1 Mk II.2 Mono Amplifiers
Price:
$9800 USD per pair.
Warranty: Three years parts and labor.

Atma-Sphere Music Systems
160 South Wheeler
St. Paul, MN 55105
Phone: (651) 690-2246
Fax: (651) 699-1175

E-mail: ralph@atma-sphere.com
Website: www.atma-sphere.com


Atma-Sphere responds:

We thank the staff at SoundStage! for the consideration they have given this amplifier and especially us during the process of this review.

There is a debate in the audio world that has intensified in the last 25 years. It began in the latter part of the 70s, with The Absolute Sound and Stereophile (under Gordon Holt) discussing the "sound quality" of various amplifiers they reviewed. What emerged is now a well-known phenomenon: some equipment that sounds very good may not measure well, and some equipment that measures well doesn't always sound that good. Add to that the phenomenon of every manufacturer in the world claiming to make the best audio gear. We now have a situation where audiophiles, unless they really trust a particular reviewer, must audition the equipment in their own home, regardless of what has been written about it.

Why is this? The simplest explanation, probably the correct one, according to the principle of scholastic philosophy known as Occam's Razor, is that common measurements are not relevant to what we hear. Indeed, it is now well known that total harmonic distortion (THD) measurements usually don't show good correlation to amplifier sound. Back in the 1960s, General Electric demonstrated that people are relatively insensitive to even-ordered harmonics (and will easily tolerate up to 40%) yet are very, very sensitive to odd-ordered harmonics, particularly those of a higher register such as the 9th, 11th and beyond. Yet to this day there has been little done to quantify the odd-ordered harmonic distortions in audio equipment. Forty years later we are still for the most part measuring and reporting only THD.

This discrepancy of measured vs. audible performance has its roots in ancient audio history. Decades ago, well before the advent of solid state, engineers began defining all audio performance in terms of measured voltages. This practice is still very much in use today. One of the consequences of this approach has been the extensive use of negative feedback, as negative feedback is well known to reduce THD and improve voltage response uniformity of power amplifiers. When the transistor was introduced, the linear voltage response of amplifiers was emphasized, as transistor circuits with massive feedback have a voltage response in power amplifiers that is nearly insensitive to load (in other words, a typical transistor amp will produce the same voltage regardless of speaker load). Tube amplifier manufacturers, despite the more linear devices they were using, were hard-pressed to produce an amplifier that had the linear constant voltage response of a transistor amp. In fact, this feat has not been accomplished in over 80 years of tube technology.

Despite the apparent linearity of transistor amps, over the years tubes have demonstrated audible performance characteristics that transistor amps still dream of accomplishing. This is well known by many audiophiles worldwide and should come as no surprise to anyone reading this text! This would seem to beg the question: is linear voltage response regardless of load a desirable thing?

In the last decade, we have witnessed a resurgence of some very old ideas in audio, for example the use of single-ended-triode (SET) amplifiers and the rejection of the idea of negative feedback. In 1990, Atma-Sphere introduced a new amplifier called the MA-2 that had switchable negative feedback. We found that no one used the feedback setting on the amp, as it was easy to demonstrate how the 8dB of negative feedback it engaged was detrimental to the sound on nearly every speaker. Of course, we had tried negative feedback in many of our prototypes, which is why many of our amplifiers, like the MA-1 Mk II.2, do not use it. The amplifier is really intended for ultimate audible performance rather than the best bench specs, as most audiophiles are likely to be listening to music rather than test tones.

Over the last 25 years we at Atma-Sphere have developed not just a philosophy about negative feedback (in a nutshell, we don't like it), but also ideas about how to best quantify the performance of an amplifier without feedback. As mentioned above, the traditional methods, known for poor correlation to audible performance, have used voltage tests to predict performance. However, prior to the advent of constant-voltage transistor amps, all speaker designers used vacuum tube amplifiers for their measurements, so there was less trouble correlating things then. Tube amplifiers have something closer to what one might call a constant power, rather than constant voltage, characteristic. For most practical speakers, the tube amplifier will produce nearly constant power at all frequencies. On the face of it, this might seem to be what you want, but in the last 20 years, with transistor amps becoming the accepted standard, for speakers designed with transistor amps in mind the constant power characteristic might not work so well.

The speaker sensitivity measurement is related to this issue, but in a different way. Let's take the example of a speaker that has a sensitivity of 87dB for 2.83V/1 meter. If it is an 8-ohm speaker, this is the same as 1 watt/1 meter (efficiency test). However, if the speaker is 4 ohms, now it works out that it is 2 watts/1 meter. In fact, the 1 watt/1 meter test for the 4-ohm speaker would show 84dB, which is not such a great number. This subtle confusion is how some speaker manufacturers "pad" their numbers. The sensitivity measurement came into being to support constant-voltage transistor amps, which double power as the impedance is cut in half. This makes the 4-ohm version of the speaker seem more efficient than it really is. The 1 watt/1 meter test (using power rather than voltage) is more apt for listening satisfaction.

There are other examples of how the "voltage school" measurements fail to correlate with listening satisfaction. One is the difficulty transistor amps (with large amounts of feedback) have in getting full-range electrostatic speakers like the Quad or Sound Lab to play bass. The reason is that electrostatics have very high impedance in the bass region, preventing transistor amps from making more than a small fraction of their 8-ohm-rated power. Because of the voltage feedback they have, they are unable to compensate for the rising impedance, so transistor amps driving electrostatic speakers will be bass-shy and overly bright in the highs.

Tube amps on the other hand have a nearly constant power characteristic, and can make the electrostatic speakers play bass without also being bright (OTL amps are particularly well known for this). The speaker is expecting constant power at all frequencies, not constant voltage, despite the measurement traditions in place.

We believe therefore that all amplifiers (and it is easily demonstrated), tube or transistor, are sensitive to load if we regard power rather than voltage response. Transistor amps are more, rather than less, sensitive. We also believe that tube amp designs should strive for a constant power characteristic, rather than the constant voltage character exhibited by transistor amps with massive feedback.

The MA-1 has such a character, and will play with nearly constant power over a very wide range of impedances. If we refer to figure 1 (left), we see a copy of Chart 1 of Mr. King's measurements, where this is demonstrated with an additional curve. This curve represents the power that the amplifier is producing, based on the voltages measured by Mr. King. What we see is that the maximum deviation at any frequency is only 1.3dB. Now if the speaker were designed with a tube amp in mind, this is probably OK. But if the speaker were designed for transistor equipment, it may not sound as good, as it expects the constant voltage character of transistor amps. I have identified points of interest on the chart as A through E. I have listed the impedance of the NHT test load at these points, and the powers thus produced by the MA-1 and a typical transistor amplifier.

If one seeks explanations for why one hears differences between tube and transistor amplifiers, and why measurements don't always correlate with what one hears, this is certainly an important reason. Many tube-amplifier manufacturers have added feedback to their designs, with limited success at improving the voltage response of their amps. But this is done at the sacrifice of image palpability, soundstage depth and the increase of the higher odd-ordered harmonic content (the kind GE showed 40 years ago that humans don't like). Of course, transistor designs suffer these same faults. It would seem that the use of feedback has resulted in improving voltage measurements at the sacrifice of satisfaction in what we hear.

So far I've not mentioned the power output measurements Mr. King made. There are two variables that have to be considered. The first is the tubes themselves. It turns out that Mr. King did not test them before setting up the test. They had been in operation for seven months prior (we ran the amps for two months prior to giving them to Marc Mickelson) and also had been shipped by UPS. Our experience has been that the time after shipment is the most critical to the tubes. The jangling they can get from shipment means that the first two weeks, and usually within the first day, of operation after setting up the tubes again is the most likely time for them to fail. The second variable is the line voltage itself. We spec it at 120 volts, which helps to prevent damage to power-supply components in places like New York, where line voltages can run in excess of 125 volts quite easily. Like any power amp, the MA-1 is sensitive to line voltage, and a 5-volt drop in line voltage can cut the power in half (note that this is true of many amps, not just ours). It turns out that 1 to 2 volts can easily be dropped across the line cord (the MA-1 is after all a Class A2 amplifier and draws A LOT of current at full power). If you've ever wondered how a power cord can make a difference in the sound, here is a measurable reason why. As a result, we remove the bottom cover and measure the line voltage on the terminals of the AC input connector. Mr. King did not do this.

Setting up the test with these things in mind, we get 143 watts at clipping fairly consistently (in other words, on several different amplifier samples) into 8 ohms.

This amplifier and its brethren have all received much praise in the press (as well as awards), over many amps that would have measured better using the traditional voltage measurements. We feel that because some speakers are designed using transistor amps and others using tube amplifiers, there is a pronounced divergence in the field of testing. We agree very much with Marc that care must be taken when matching this amplifier (and any amplifier) to the speaker. More attention to the issue of power response, as opposed to voltage response, is needed; we are all aware of the differences we hear.

Ralph Karsten
Atma-Sphere Music Systems

[SoundStage!]All Contents
Copyright 2002 SoundStage!
All Rights Reserved