ELECTROSTATIC AMPLIFIER Paul Seydor, reviewer
Last Updated: 09 March 2014 09 March 2014
Created: 01 March 2009 01 March 2009
The Absolute Sound, February 2001, Paul Seydor, reviewer
It's still relatively rare that a solid-state amplifier sends our editor scurrying for superlatives, so when his late-summer Editors Choice last year (Issue 125) featured a four-star rating for InnerSound's new amplifier, driving Quad 63 ESLs, together with some tantalizing comments about unparalleled dynamic range (from the 63s) and unexampled purity, he had my attention. Since I'm the resident Quad-man on staff and IAG America had not yet reclaimed the new 989 electrostatics (review, Issue 126), HP ordered me to clear my schedule for an early review. Not that an order was necessary; anything that can get even better performance from Quads automatically gets front-burner status in my kitchen.
Formally designated the InnerSound ESL Amplifier, it announces itself with specs and performance claims, not to mention good looks of solid fit and impressive finish, that are a little unbelievable in view of its welter-weight size a compact 17 x 5.5 x 16-inch chassis that comes in at a mere 41 pounds and $2,995 price; 2000 volt-amps/channel to electrostatic loads; 300 watts/channel/8 ohms, 600 watts/channel/4 ohms, a momentary 1,000 watts/channel/2 ohms to conventional loads; essentially dual-mono topography (apart from the transformer core); high efficiency, with less than 0.08 percent distortion before clipping; high-quality parts; minimal wiring of silver-plated copper; gold-plated connectors, balanced and unbalanced.
For the InnerSound's designer, Roger Sanders, a longtime tube aficionado, the twin nemesis of transistors are clipping and overzealous transistor circuits (see his white paper at www.innersound.net). Essentially it comes down to a simple premise: transistors don't sound harsh if they are not overloaded Of course, this begets a design of considerable ingenuity and sophistication (see sidebar). Conventional dynamic speakers present a largely resistive load that provided no ultra-low impedances are involved, solid-state amplifiers can handle with ease. But since electrostatics are, in effect, large capacitors, they present highly reactive loads that can easily drive transistors into unsafe operating areas, with a consequent increase in distortion until the protection circuit is tripped. In Sanders' view, these protective circuits have become so sophisticated that they are frequently triggered long before the amplifier distorts or shuts down. The sonic results are a subtle impression of grunge, grain, and hash; a distinct loss of dynamics and openness; an overly damped, closed-down character; and a general lack of ease and relaxation. Ironically, the better the protection, the more insidious these effects.
Sanders' solution? No protection circuitry at all. To safeguard against hazardous conditions like a dead short, the speaker outputs are fitted with fuses that, he says, work better than relays and don't work when they shouldn't. With 18 rugged, high-speed transistors per channel, rated at 250 watts each, the ESL amp could theoretically put out a total of 9,000 watts in both channels. But since Sanders uses a power supply that will deliver only 2,000 watts total, the transistors are never driven even remotely close to their full power rating or, by extension, their unsafe operating range. No in-field failures have been reported. It is a testament to the stability of Sanders' design that when I deliberately drove the 989s loud enough to activate their protection devices which, basically, just crowbars a dead short across the input terminals when the level gets too high the InnerSound didn't kick a fuse. It simply went silent for a moment, then started back up again once the transient had passed. Impressive.
Before taking up the sound of the ESL amp, I should say that it has been a long, long time since I've heard an audiophile-grade amplifier exhibit any of the hash, the grain, the brittle, prickly highs, and all the other evils of transistoritis from the early days of solid state technology. This string of tin cans has long since been shed by this particular tail. In the past year alone, I've reviewed or otherwise commented upon several amplifiers, basic and integrated, in several price ranges that span the whole Yin-to-Yang spectrum, sounding sweet, sharp, warm, brilliant, veiled, transparent, open, dark you name it, you'll find the flavor. No, they don't sound like tubes; but the best of them are capable of a musicality that easily rivals tubes and relegates the cliché’s of transistor sound to history.
Sanders says that an amplifier optimized for electrostatics will also perform better with dynamic loudspeakers. As this review is being written under a severe time constraint, the remarks that follow apply mostly to the Quad 989s (often augmented with REL's potent new Strata III subwoofer). However, see my review, this issue, of the Sehring System 603 loudspeaker; the last round of evaluations was conducted using the InnerSound, with superb results, dynamic range in particular was quite stupendous.
Smoothness, purity, and dynamics triangulate the aural landscape of the ESL Amp. Following a long break-in, first impressions brought a number of welcome negatives: no colorations to speak of; no silvery highs, liquid midrange, or tight bass; none of that excessive articulation that etches the tune like a scalpel or beats-out-that-rhy-thm-like-this. When amplifiers get into trouble driving electrostatic loads, it's usually way up high, so one of the first things I put on was Vadim Repin's violin recital Tutta bravura (Erato 3984-25487-2), which emerged without the slightest electronic haze or overlay, so clean that I wondered if something wasn't being taken away, until further listening confirmed that this was reproduction of rare purity. I played a number of my usual popular-vocal albums from Sinatra, Fitzgerald, Clooney, and Janintha, finishing with an old LP of Doris Day doing Over the Rainbow one of the more vivid recordings of a popular singer. It's so closely miked that she overloads it from time to time, but it never becomes harsh or distorted. The InnerSound/Quad combination reproduced this more suavely than I had heard it before. Indeed, when it comes to vocals, beauty is the operative word everything that comes out of it tends to be lovely. Throughout the listening sessions, never once could I make it reproduce an untoward sound that I could attribute to the amplifier itself. Recordings that I know to be unpleasant (much early digital, for example) don't sound particularly nice; but they definitely had less of the nasties than I've been accustomed to hearing.
You needn't worry, however, that the InnerSound prettifies sounds that are meant to be edgy, piercing, or disturbing. Those muted trumpets that announce the funeral march near the beginning of the Mahler Third symphony, cut to the quick and chilled the bone. Nor does the InnerSound achieve its listenability at the expense of smoothed-over dynamics. I was able to drive this symphony to louder levels on the 989s than ever. The return to tempo primo near the end of the First Movement brings a cataclysmic climax that managed to trip the Quad's protection circuits, but up to that (extremely loud) point, the reproduction was immaculate. Cymbal crashes were particularly impressive. Much of the time these can sound rather splashy, which must indicate some sort of spurious electronic artifacts. But these emerged like real cymbals with a natural decay and no electronic after-spray.
As for the low end, in the Naxos recording of Vaughan Williams' Sinfonia Antarctica, the organ is full, deep, and pervasive, each note audible as a distinct pitch, without becoming detached from the overall sound picture and calling attention to itself. Bass like this may not knock your socks off, but it certainly sounds natural.
The same goes for imaging. When the choirs advance from the rear and move to their lofts, singing Once in Royal David's City in the Kings' College Christmas Eve service, you're clearly aware of the distance they've traveled in one of the most richly reverberant acoustic settings in the world. Like so much else about this amplifier, the imaging never draws attention to itself, except, perhaps, in one way, which I found satisfying: The ESL Amp seemed to make vocalists stand somewhat more realistically higher through the 989s than I've heard from other amplifiers.
There is no question that the ESL amp forms an extraordinary synergism with the ESL speaker, the principal effect being to make the speaker sound smoother and more linear from the midrange on up. All of my reservations about the new Quad's altered high-frequency profile vis-…-vis that of its predecessor virtually disappeared. Even Robert E. Greene concurred that the InnerSound went some distance toward quieting several of his far more serious reservations about the 989's tonal balance. If the 989 interests you and it should be on the short list of anyone contemplating spending eight grand on speakers then you really must hear it with the ESL Amp.
I have only one concern I can't even call it a reservation about the InnerSound's performance, a concern that leaves me in a conundrum. Like other forms of criticism, audio reviewing tends to focus on positive characteristics, that is, on what we at least think we can hear and thus readily identify and describe. We speak of an amplifier's ability to control difficult sources, of its prowess in unraveling and clarifying demanding program material; of its grabbing us by the labels and making us take notice; its ability to retrieve detail and lift veils, of its resolving capability of its slam its resolution, its sweetness, hardness, lightness, darkness, Yin, and Yang. But how do we describe what isn't there; how do we write about what it doesn't do? And while we all agree that a theoretically perfect amplifier merely amplifiers the signal, adding or detracting nothing our old friend the straight wire with gain is it possible that we might actually prefer amplifiers that do pleasant or exciting things to our music over those that do not, let alone those few that do almost nothing at all, however greater their accuracy? Or is that preference merely transitory until we learn to appreciate the less obvious but ultimately more rewarding benefits of a higher degree of fidelity? And if so, how do we recognize it sooner rather than later? For aren't qualities such as neutrality or what I prefer to call absence precisely the sorts of things we're not aware we haven't achieved until a new component comes along that raises our awareness?
The InnerSound has started these and related questions to kicking around in my brain because I've been unable to resolve whether there is something about its presentation that precludes a full involvement with the music or, to put it more accurately, a full involvement along conventional audiophile lines. This is not a matter of dynamic range, coloration, control, stability, or any form of distortion I can readily identify. Nor do transparency and detail appear to be the issue. There may be a slight loss in the former as compared to a few other amplifiers I really like (such as the Marsh), but if so, it is slight. And though I wouldn't call it the most detailed amplifier around, those details in all the sources I regularly use to check for this sort of thing are clearly audible. Which must surely be a more accurate indicator than having details leap forward. If I seem less than decisive here, it is because a slight rise in the very high frequencies can make a component sound both more transparent and more detailed than it really is, or than a flatter, more honest component is.
What I'm trying to get at is an elusive, difficult-to-define lack of what, for want of better words, I'll call a little something more in the way of texture or perhaps just plain garden-variety flavor to the sound, a lack that from time to time seems to keep the music from fully engaging me. I am treating this so gingerly because I need more time to investigate it and because, though it is a recurring impression, it is by no means a consistent one. Nor am I at all certain it even is a shortcoming. The InnerSound is so free from typical kinds of distortion including a number of euphonic ones so familiar we may no longer recognize them as colorations that it is no doubt disadvantaged, not only in comparative listening but in our critical vocabulary. For one thing, it has managed to catch out some pretty subtle colorations in other amplifiers I thought were essentially neutral (like the Marsh, way up high). For another, the kind of texture I'm talking about may itself be an artifact. Of this I am certain: If the InnerSound ESL Amp is not the most completely nonfatiguing amplifier I've ever auditioned, I don't know of another that is its superior, and only the Quad 909 amp (see review in this issue) is a possible peer. A few years from now, we may look back upon these, and a couple of others I've heard reliable reports about, as they force us to reevaluate the very terms by which we now describe electronic components.
In the meantime, allow me to recount an experience from late in the evaluation process. After listening to the entire first movement (over 30 minutes) of Salonen's Mahler Third, I wanted to hear how it compared to Bernstein's DG recording. I listened to both at really loud levels the Salonen a powerful but natural recording, the Bernstein a good, though rather brightly lit one. Yet after more than a hour's sustained onslaught of this Juggernaut of a movement played twice, I felt not even the faintest tingling of listener fatigue. This is an acid test that few components of any stripe are able to pass. The InnerSound/Quad 989 combination passed it not just faultlessly, but with a sustained musicality that made me want to keep on listening to this magnificent music. Experiences like this suggest that I have only begun to grasp the full measure of this remarkable amplifier.
Dr. Robert Greene Comments: Electrostatic amplifiers often present a demanding load to amplifiers. That the strong capacitive component of their load can cause amplifiers to deviate seriously in frequency response was pointed out long ago by Tom Holman, with measurements, in Issue 26. I measured differences as large as 5 dB between different solid-state amplifiers into InnerSound's own Eros electrostatic speaker when I reviewed it a couple of years ago. And these were amplifiers with nominally low output impedance, the response of which was not supposed to vary, to any appreciable extent, with the load attached.
The InnerSound Amplifier is specifically engineered to perform well into the kind of load an electrostatic speaker presents. I did not have the opportunity to do response measurements on the Quad 989 speakers with the InnerSound vs. other amplifiers. But both PS and I did observe that the InnerSound made the top end of the 989 more agreeable and, I suppose, more accurate. With most other amplifiers, the 989 seemed somewhat glare-ridden. With the InnerSound, this effect was much reduced. It seems likely that the InnerSound does not develop the top-end rise that many amplifiers do when confronted with an electrostatic-speaker load. Under these circumstances, it seems obvious that an owner or potential purchaser of an electrostatic speaker ought to audition the InnerSound amp. This is a most emphatic recommendation.
Listening to the InnerSound with a dynamic speaker with sensible impedance behavior produced less decisive results. It certainly presented a relaxed and clean sound. It has enormous power reserves and never sounded even remotely strained or in any way disagreeable. I had the impression of a little less top-end energy than with, say, the Marsh. I don't really quite agree, however, with the idea that PS expresses that the InnerSound is showing up a coloration in the Marsh in this set-up, although I would bet on the InnerSound into electrostatic loads against all comers.
Amplifiers differ somewhat in their top-end behavior into real-world loads, even into comparatively non-troublesome ones. The output networks involved flip the top up or down a bit, often enough, even when the amplifier is nearly the proverbial ruler-flat into 8 ohms resistive. With a low-output-impedance, solid-state amplifier, this effect is likely to be small. But deciding by listening alone exactly what response at the top is correct, seems to me essentially impossible.
First, the effect may vary from speaker to speaker; second, one does not know how the speaker itself is without driving it with an amplifier. After all, the tweeters were tested and the speaker designed by listening or measuring with some amplifier or another. To then use a speaker to make anything but a comparative judgment about amplifier response is somewhat circular.
I agree with PS that the Marsh gives signs of more extroversion (my word) in the top than the InnerSound. Indeed I mentioned this quality of the Marsh in my review (Issue 125). One may prefer one sound over the other, but I don't see it as simple to decide via listening which is correct. At best, one is talking about some averaging process in comparison with other amplifiers, not any kind of absolute. And certainly I would think three times about calling the behavior of the Marsh a coloration. (There is a good reason why I try not to review amplifiers very often too many undecidable questions. The differences are there to be heard, but the question which is correct is far less easy to decide strictly by listening than most audio reviews suggest.)
It would seem almost obvious in this situation, that amplifiers really ought to have adjustable output networks to minimize the difference between input and output, like Hafler's XL-280 and its associated null test: The output network could be adjusted to be correct for one's particular speaker to get maximum agreement between input to the amplifier and (attenuated) output to the speaker. As HP commented long ago, correct adjustment eliminated tizz in the high end. This was a good idea that unfortunately vanished, and HP's remarks (Issue 46) certainly point up that the effect I am discussing indeed does have audible significance.
None of this is supposed to suggest that frequency response is the only effect involved in amplifier sound, but it is certainly one of the effects, and it interacts with the perception of the others.
In summary, the InnerSound is a clear contender and something of a revelation for electrostatic speakers. As an amplifier for dynamic speakers with reasonable impedance behavior, it is certainly a competent and pleasing product, one that does a clean, honest job of playing music, another worthy addition to the list of inexpensive but powerful amplifiers that bring high-quality performance to the moderate price range.
I enjoyed very much its calm, un-aggressive sound that seemed like the Quad 909 amp in its freedom from electronic character. The Quad had a slightly different character, however, with a more forward perceived sound.
But as to choosing among the various solid-state new contenders there lies a tricky business, indeed. The little variations in how the basically well-engineered amplifiers sound will have different significance with different speakers, and even for different people. If you have speakers that you wish had a bit more top-end sparkle, an amplifier with a little extra top may be the ticket. To tame a nasty, rising tweeter, a little darker one. A forward speaker is likely to make one favor the InnerSound, a recessed speaker, the Quad and so on and on. A really neutral speaker is likely to make any of the amps acceptable. It is when you are about to go over the line to some sort of pain that little things count.
Fortunately for our sanity, these changes tend to be small for speakers presenting standard loads. But for electrostats, special care is needed, and the InnerSound has surely done something exceptional there.