A Comparative Survey of Power Amplifiers: Part III (Winter 1977/78)

A Comparative Survey of Power Amplifiers: Part III ---Audionics CC-2; GAS 'Grandson'; Mark Levinson ML-2

-----

Our final installment (at least for a while) of this running survey is short but exceedingly sweet, as it presents both the best buy to date and the best, period. Plus some modestly sweeping generalizations about design criteria.

In the preambles to both Part I and Part II of this survey, we rashly held out the hope of finding more precise correlations between the measurable and audible characteristics of power amplifiers. We must confess that we didn't get nearly as far as we would have liked to, when suddenly and providentially help arrived in the form of an outstanding engineering paper by Eero Leinonen and Matti Otala, under the just-what-the-doctor-ordered title of "Correlation Audio Distortion Measurements" (Journal of The Audio Engineering Society, vol. 26, pp. 12-19). The paper accomplishes at least half of what we had in mind-and, we must hasten to add, better than we could possibly have done that half: it isolates five basic distortion mechanisms in amplifiers (three static and two dynamic types) and correlates five different distortion measurement methods (including several of ours) to evaluate the sensitivity of each method to each type of distortion. We couldn't have asked for anything more germane to our needs.

The other half of the problem-the one that the paper doesn't deal with-is just what these distortions sound like, singly and in combination, and just how much of each represents the threshold of audibility. Some insights along these lines may emerge from our future amplifier tests, which will be at least partly based on the Leinonen/Otala findings (published too late to influence this survey); further more, Dr. Otala and his team have already done some interesting research on the audibility of TIM, which will have been published by the time you read this. Andy Rappaport is also hard at work on sonic correlations in amplifier design (see his letter to the Editor in this issue), as are some good people at Mark Levinson and a few other places, so that things are definitely looking up.

Our new amplifier measurement procedures will go into effect beginning with the next issue and will be fully explained there;

meanwhile you may assume the same laboratory and listening tests to be behind the reviews below as were discussed in Parts I and II, except that the exquisitely transparent Tangent RS2 was added to the range of speakers used for listening and the GAS 'Sleeping Beauty' Shibata cartridge replaced the Denon as our reference cartridge in the few in stances when we were not using master tapes.

What the best power amps have that others don't.

Our difficulties in fine-tuning the exact sonic correlations of our laboratory tests haven't prevented us from coming to some general conclusions regarding the specific circuit characteristics of good-sounding power amplifiers. It seems fairly obvious to us, after having measured and listened to all these different units, that the best ones have certain design principles in common.

Preeminent among these is not too much negative feedback. As we've explained before, zero feedback could be argued to represent the theoretical ideal, and a number of amplifier designers are now working on making it feasible. Whatever benefits negative feedback may provide (and no one can deny those benefits) fade into insignificance compared to the nastiness of excessive feedback. How much is excessive? Not very much. The ‘optimum maximum' can be mathematically derived for any given circuit and is generally in the 12 dB to 20 dB range. (The majority of amplifier designers appear to be innocent of these computations.) Thus a power amplifier with 50 or 60 dB of negative feedback is virtually guaranteed to be a screaming horror and unworthy of a serious reviewer's time and effort.

(Look for ridiculously low static distortion figures at very high power, damping factors in the upper hundreds, and very low slew rates. Then start walking rapidly in the opposite direction.)

An inescapable prerequisite of not too much feedback, and therefore another feature of all good power amplifiers, is an open loop of sufficient bandwidth and linearity; in other words, the amplifier must be a pretty good one even before the feedback is applied by closing the loop. In the old vacuum-tube days, this was considered elementary; designers agonized over the performance of each individual stage of the circuit and then very gingerly applied a few dB of feedback to make everything a little better still. The solid-state revolution encouraged a cavalier attitude toward the open loop; any cruddy class B circuit could now be slapped together by a novice and then drenched in feedback as a cure-all, with high power output and magnificently low THD and IM figures to show for it on the spec sheet. The sound quality was another story, but who listens? The remaining universal but frequently ignored criterion of a good power amplifier is a stiff power supply, capable of supplying the current required to make the amplifier a true voltage source regardless of the load, which in the real world may be resistive, capacitive or inductive-and most likely a combination of the three. Somehow the design of the power supply gets left out of the arguments among cultists about this kind of amplifier circuit versus that kind, but the incontrovertible fact is that no amplifier can be better than the stuff that comes down its power supply rails. What's more, the higher the maximum output-signal voltage the amplifier is intended for, the more critical the power supply. The Bryston 4B, for example, wouldn't be able to put 40 volts per channel into just about any load if it didn't have quite a bit of money sunk into its power supply. The some what skimpier RAM 512 begins to run into problems when you ask for 38 volts, whereas the Audionics CC-2 will happily give you 19 volts (and even a little more) into very nasty loads with its well-designed but relatively modest power supply. The Electrocompaniet amplifier, on the other hand, is in trouble with 14 volts at 4 ohms, even though it's a marvelous amplifier in many other ways.

A recipe for power amp design.

Singling out the above three criteria may seem a bit simplistic, since many other considerations obviously enter into the design of a good power amplifier. We haven't even touched upon such matters as protection circuits (cur rent limiting is a potential source of sonic problems!) or the best method to deal with the very real distortion produced by out-of band program components (not many designers are dealing with it at all, but we plan to delve into it in future issues). As we've pointed out before, however, The Audio Critic isn't an engineering school, and we must simplify matters, sometimes even flamboyantly, for the sake of effective consumer education. As long as our priorities are correctly ordered, we don't mind harping on the main issues one-sidedly or over-emphatically. In that didactic spirit, we offer the following recipe for correct power amplifier design as the distilled experience of this survey:

(1) Make the amplifier as linear as possible before the application of feedback. (2) Apply feedback very sparingly and never mind the resulting lack of double-oh distortion figures. (3) Make the power supply as stiff as the budget permits-then spend a little extra and make it even stiffer.

Our very first review is a case in point.

Audionics CC-2

Audionics, Inc., Suite 160, 10950 SW 5th, Beaverton, OR 97005. CC-2 Amplifier, $399. Three-year warranty. Tested #04001, on loan from manufacturer.

This very new unit, representing a later stage in Audionics' thinking than the higher powered and costlier PZ3-1I1, is the big surprise of this survey. It's the kind of amplifier the leading Japanese manufacturers have always advertised but never delivered: moderate in price and superb in sound quality. It seems that they know a few tricks in Oregon that are still unfamiliar in the land of Origami.

To our greatest amazement, the Audionics CC-2 blew away every power amplifier in our listening tests except the Mark Levinson ML-2.

Does it sound better than the Bryston 4B? Yup. Does it sound better than the Electrocompaniet? Yes, indeed. Does it sound better than the Futterman? Well, we didn't have the very latest version of the H-3 available, but an earlier version didn't stomp the Bryston 4B and the Electrocompaniet the way the CC-2 did. It's enough to make the most seasoned audio purist walk away shaking his head and muttering to himself.

Let's specify what we mean by sounding better. The Audionics CC-2 equals in transparency and transient detail the Bryston/Electro level of performance and is at the same time smoother and sweeter on top. We never thought of either the Bryston or the Electro as hard or zippy-on the contrary, we found them far superior to others in that respect- but the CC-2 makes them appear that way.

Make no mistake about it: we aren't talking about the thick, whipped-cream kind of smoothness without clarity. The CC-2 is both clear and smooth. Only the $1800-per-channel Mark Levinson ML-2 beats it on clarity, definition, inner detail, and total freedom from fuzz.

It mustn't be assumed, of course, that the Audionics can drive with ease every load the Bryston can. The 4B is capable of 40 volts per channel into any load down to 3 ohms or so; the CC-2 will deliver at least 19 volts, even into 2 ohms, and as much as 25 volts into 8 ohms. (Its official rating is 70/70 watts at 8 ohms, i.e., not quite 24/24 volts.) Thus it's far from a super amplifier in sheer output capability-but wait. There's a toggle switch in the back that bridges the left and right channels into a single mono amplifier rated at 225 watts into 8 ohms (that's more than 42 volts). What's more, the slew rate of 36 V/uS in the stereo mode is automatically doubled in the bridged mono mode. Now that begins to look like a super amplifier-and still for only $798 the pair. Unfortunately we didn't have two CC-2's available to listen to in the bridged mode as a stereo pair; we've been promised a second one for a follow-up report in the next issue.

Now comes the interesting part: why does the Audionics CC-2 sound so good? You've probably guessed it already. It's a perfectly simple, straightforward circuit with not too much negative feedback, so that it produces very little dynamic distortion. Its static distortions are far from spectacularly low; none of our THD, SMPTE-IM and CCIF-IM measurements were in the double-oh or even low single-oh range. High single-oh or point-one something are typical at 20-plus volts out into a variety of impedances. Nothing to brag about on a Japanese spec sheet. Live and learn.

We're told that Audionics worked very closely with Fairchild, the maker of the out put transistors, in developing the CC-2 circuit.

That may explain part of its success; engineers who design solid-state devices generally know more about their application than circuit de signers who look up the devices in a catalog.

In any event, we're both impressed and de lighted. Since the low-profile CC-2 matches the almost equally excellent BT-2 preamplifiers both in appearance and, within a few dollars, in price, Audionics appears to have a very good thing going for a large number of audiophiles.

GAS 'Grandson'

The Great American Sound Co., Inc., 20940 Lassen Street, Chatsworth, CA 91311. Model A-701M 'Grand son' Servo-Loop Amplifier, $329 (with meters). Five year warranty; customer pays all freight. Tested #A4-701356, on loan from manufacturer.

"Hard as nails," commented one of our auditioners, and we concur. This is not a nice sounding amplifier. Very hard, yes, but also peculiar in balance, nasal and sizzly. "Thorns on the cymbals" was another comment. The listening sessions were short, since the sound was extremely fatiguing.

We can't even blame it on Jim Bongiorno, who had left the company before this bottom of-the-line power amplifier was finalized, leaving only the fading mark of his funky nomenclature on it. (You know whose Grandson it is, don't you? Get the laser guns, professor . ..) Perhaps a genuinely accurate amplifier is too much to expect at this price, but we can't refrain from hoping, especially in view of the extraordinary quality of the Audionics CC-2 for not too many dollars more.

The Gas 'Grandson' is designed as an 18 volt amplifier (40/40 watts into 8 ohms, 80/80 watts into 4 ohms), and near its maximum output its distortion figures are far from impressive. If we still had our former faith in CCIF-IM, for example, it would be easy to say "Aha!" because the 13 kHz sideband alone is of the order of 0.2%, but the Leinonen/Otala findings tell us not to trust this excellent test in all cases and for all diagnoses. If forced to take a guess at the root of the sonic problem, we'd say it was the power supply, since the amplifier distorts several times more into 4 ohms than into 8 ohms, indicating in sufficient current capability. We also noticed some peculiar cross-coupling effects between the two channels through the power supply, as well as an unhealthy amount of ringing with capacitive loads. So there seem to be enough things not to like about the Grandson on the test bench. We must admit, though, that we didn't dig very deep after those uninviting listening tests.

We hope that this model, or at least our sample of it, is untypical of the direction the GAS Company is now taking. Otherwise the successor to Bongiorno may turn out to be Addio.

Mark Levinson ML-2

Mark Levinson Audio Systems, 55 Circular Avenue, Hamden, CT 06514. ML-2 Class A Power Amplifier, $1800 (per mono chassis). Five-year warranty, customer pays all freight. Tested #1196 and #1199, on loan from manufacturer.

Here's the chance of a lifetime for the avenging consumerist: a 25-watt mono amplifier priced at $1800-and no discounts. The head of Louis XVI couldn't have been a more tempting object to Robespierre than this arrogantly elitist piece of hardware is to the long suffering audio reviewer with the slightest taste for journalistic retribution.

Well, sorry to disappoint you, but in our considered opinion this is the best power amplifier in the world-at least the world we're aware of. Nothing we've ever heard equals its absolutely focused, pristinely delineated, rock solid and totally un-fuzzed sound. Nothing.

Other amplifiers, even very good ones, that are A-B-ed against it invariably create the impression of slight distortions, smears and colorations. It's unfair but there ain't no Santa Claus; money talks and everybody else ends up with second best. (No one held a gun against your head to make you become an audio perfectionist, right?) A word about that 25-watt rating. That's what the ML-2 will deliver into the official 8-ohm load, at any frequency. In other words, it puts out a little over 14 volts. But-a big but-it will put 14-plus volts into any load, including the kitchen sink or your big toe; for example, it will give you 100 watts rms into 2 ohms, which is just over 7 amperes rms or 10 amperes peak. And you can't call any amplifier with a 10-ampere peak capability a little amplifier. The ML-2 is a big amplifier (it weighs 65 pounds with its giant heat sinks and carrying handles front and back), but it's big on current rather than voltage. That's one reason why it sounds as good as it does. Of course, it will clip if you push it beyond 15 volts or so;

we've done it. The solution is to bridge two of them; an inverting second input is provided for that very purpose and you've got yourself a 28-volt (i.e., 100-watt, at 8 ohms) amplifier for only $7200 the stereo pair. (Take their names, comrade. They'll be the first to be liquidated.) The circuit of the ML-2 is the epitome of the 3-point design philosophy we discussed above: the open loop couldn't be more linear since it's pure, unqualified class A (in fact the amplifier is designed to maintain class A operation over its full output range even with a 2-ohm load-probably a world's record); the total amount of feedback used is extremely low; and the power supply is as stiff as you'll ever see. Quite frankly, the one thing we can criticize about the amplifier is that possibly inaudible design trade-offs were obviously not even investigated in its development; 'if it's theoretically better, do it" was the guiding principle. Who knows, maybe an equally good sounding unit could have been built for quite a few hundred dollars less. Certainly, at least $200 could have been chopped off the price if the front panel were less magnificently sculptured, but then a Rolls Royce owner wants that Rolls Royce grille up front.

Our measurements of the Mark Levinson ML-2 (admittedly BLO-Before Leinonen/ Otala) could reveal no sins whatsoever. All the numbers were beautiful: THD, SMPTE IM, CCIF-IM, propagation delay, recovery time, the works. If there's something wrong with this amplifier, we don't have the test for it yet. Oh yes, for slew rate collectors: it's of the order of 100 V/uS and double that in the bridged mode with two units.

We can't conclude this review before coming back to the sound of the ML-2, al though it's hard to talk descriptively about something that aims to be totally neutral and actually succeeds. We feel that the ML-2 lets through all the sound that's fed into it; we heard it retaining tiny ambience details, for example, that got lost in the soup through other amplifiers. And, always, we must keep coming back to that complete lack of fuzz.

Next to the ML-2, the sound of any other amplifier we know of has hair on it. Also, the ML-2 controls the bottom end of a marginally damped woofer better than anything else we've tried.

Enough. After all, if the amplifier were less good than it is, it would be an outrageous rip-off. That it's still a highly plausible purchase is a credit to both Mark Levinson and Tom Colangelo, the engineer behind the design.

Recommendations

Keep in mind that the following is based on our tests of only 18 amplifiers and our working knowledge of a couple of dozen others.

That's not the total power amp universe; surprises may be forthcoming in individual tests reported in future issues.

Best power amplifier, regardless of price: Mark Levinson ML-2 (or bridged ML-2's for more power).

Second best power amplifier (and only incidentally the best sound per dollar): Audionics CC-2 (or bridged CC-2's for more power).

-----

[adapted from TAC, Vol.1, No.5 ]

---------

Also see:

Sophisticated Speaker Systems, Large and Small: A Comparative Survey

Various audio and high-fidelity magazines