|
|
[Review by EDWARD J. FOSTER, orig. from Audio magazine 100th anniv. issue, May 1997]
Is it a contradiction to speak of a design as both innovative and traditional? In this case, I would say not. The roots of Carver’s A-760x Magnified Current THX stereo power amplifier hark back to the past while the product itself looks forward to the future. The most obvious ties to the past are the twin, round analog “power” meters, which are calibrated in decibels and watts. They are the only deco ration on an otherwise stark front panel and, as on Carver’s previous upscale power amps, add an air of technical seriousness that belies their limited usefulness. Al though I’d usually pooh-pooh such meters, I must admit that Carver is quite frank about their strengths and weaknesses in its excellent owner’s manual. After describing their calibration and ballistics (average- reading with fixed overshoot), it concludes that “the best way to tell whether the amplifier is overloading is simply to listen.” I concur—but the warm glow of the meters does remind me of a simpler time.
The only controls on the front panel are a heavy-duty power rocker and two buttons for the meters. One of these controls the meter illumination; the other changes the meter range by a factor of 10, so that 0 dB corresponds to 38 watts with “Range” depressed and to 380 watts with the button released. The scale ex tends to +3 dB (78 or 780 watts) to leave room for overshoot. All power readings are calibrated assuming 8-ohm loads, however; the meters themselves read output voltage, not power directly.
The back panel is equally simple: one pair of binding posts per channel, gold- plated RCA input jacks for each, and separate left and right finger-adjusted controls to vary voltage gain. (When the controls are advanced fully, the A-760x adheres to Lucasfilm’s Home THX specifications.) There’s also a switch to bridge the two channels into one for mono operation, which almost quadruples the power available into a single 8-ohm load. In bridged monophonic operation, the left-channel RCA jack is used as the input and the two red binding posts serve as the output, with left-channel red being “positive.”
The A-760x’s binding posts accept bare wire and standard banana plugs but are spaced 1 inch apart instead of ¾ inch, so they can’t be used with dual-banana connectors. The manual says this is done “to comply with international safety standards” and has been Carver’s practice for some years now. Compatibility with international requirements being the concern, it’s a little strange to find the A-760x equipped with a fixed two-wire line cord for the United States rather than a detachable IEC power
In design and construction, the A-760x shares the tradition of the well-respected AV-705x. Signal paths have been kept as short as possible to minimize the impact of electromagnetic interference and the “cumulative reactance in the critical signal conductors,” and the amplifier is assembled from modules, which reduces manufacturing cost without impairing quality. In fact, one can make a strong argument that modularity improves quality, since circuitry can be tested initially on a subassembly basis and accepted or rejected at that level. In the A-760x (as in the AV 705x), input and output connectors are mounted directly on the circuit board and simply poke through holes in the back pan el, so an entire amplifier channel can be plugged into a test jig and evaluated prior to installation in the chassis.
= = = = SPECS = = = =
Rated Power: Stereo, 380 watts per channel into 8 ohms, 600 watts per channel into 4 ohms; bridged, 1,200 watts into 8 ohms.
Distortion: THD, less than 0.08%; SMPTE IM, less than 0.03%; CCIF IM, less than 0.0 1%.
Dimensions: With handles, 19 in. W x 5¾ in. H x 18¾ in. D (48.3 cm x 14.6 cm x 47.8 cm); without handles, 17 in. W x 17¾ in. D (43.2 cm x 45.2 cm).
Weight: 391bs. (17.7 kg). Price: $1,299.
Company Address: P.O. Box 1237, Lynnwood, Wash. 98046; 206/775- 1202.
= = =
Fig. 1—THD + N vs. frequency at four output levels in stereo mode into 8 ohms
(A), in stereo mode into 4 ohms (B), and in bridged mode into 8 ohms (C).
Of exceptional importance, to my way of thinking, is Carver’s Total Direct Coupling (TDC) output topology. It does away with the series inductor that’s usually interposed between the output transistors and the loudspeaker, to “protect” the output stage from a wayward load. Such output inductors are really nothing but crutches that en able an amplifier of limited open-loop bandwidth to use a lot of overall feedback to reduce distortion and still remain stable when driving a reactive load. The downside of using an output inductor is that the out put impedance rises with frequency, which can result in audible response aberrations when driving real loudspeakers.
Excuse my soapbox; I’ve been on this kick for some time because I think the bad rap that’s been given to global feedback results at least in part from the side effects of using output inductors—not to the feed back per se—and I was glad when Lucasfilm picked up the cudgel and limited the permissible high-frequency output impedance of THX certified amplifiers. (Unfortunately, many of them don’t meet the requirement on my test bench, but that’s another story.)
There’s no secret regarding what’s needed to design an amplifier that doesn’t require an output inductor.
It’s just good engineering: using very fast transistors to ensure a wide open-loop bandwidth (I’m told that of the A-760x extends to beyond 20 kHz), using local feedback to control distortion within the loop, and limiting the amount of global feedback to an amount adequate to lower output impedance and keep the overall distortion within reasonable bounds, without going hog wild and counting on global feedback to cure poor design elsewhere.
I don’t have a schematic for the A-760x, but I’ll bet dollars to donuts that the Carver elves have done something similar to what I’ve described. I do know that they took special care in selecting high-speed, low- noise transistors for the front end. Each channel’s output stage uses ten 150-watt Motorola triple-diffused planar bipolar transistors (five from the positive supply, five from the negative), yielding a combined safe operating capability of 1,500 watts per channel.
The power supplies in the A-760x bear a resemblance to those in Carver’s Lightstar amplifiers but differ in that the A-760x’s Magnified Current power source simply supplies two tiers of rail voltages, whereas Lightstar rail voltages follow the audio signal itself more or less continuously. According to the manual, the supply “uses two 160 amp peak current power MOS-FETs per channel, with nanosecond switching speed to assign twice the continuous voltage to the output stage when high voltages are required, or more than twice the continuous current when higher current is needed. This maximizes both the voltage and current (peak or continuous) available for any loudspeaker load, even those that have a substantial reactive component that is difficult for conventional amplifiers to drive.”
Indeed, the A-760x’s ratings are impressive, and the amplifier is almost indifferent to load impedance. It’s specified to deliver 380 watts per channel into 8 ohms (20 Hz to 20 kHz with less than 0.08% THD), 600 watts per channel into 4 ohms, and 1,200 watts into 8 ohms when bridged. This greatly exceeds Lucasfilm’s minimum requirements for Home THX certification. Carver says it employs “precision passive components. . .in all critical signal paths” (read, no electrolytics in the signal path “fully-complementary differential circuitry using low-noise, high-speed transistors throughout,” and a “double-stage ground isolation system [ prevents ground loops and RE interference.”
Measurements
The Carver A-760x was simply outstanding on my test bench. Rarely have I seen a product meet or exceed virtually every one of its specifications as handily as this. The only two specifications it missed—and then by hardly a smidgen—were A-weighted noise (which I measured at —94.7 dBW, worst case, compared with a —95 dBW specification) and THD at rated power into 8- ohm loads, which, worst case, hit 0.0863% at 20 kHz in the right channel compared with a 0.08% claim. Needless to say, I don’t consider these serious discrepancies.
As you can see in Fig. 1A, which plots total harmonic distortion plus noise (THD ± N) versus frequency in both channels at 1 watt, 10 watts, 100 watts, and rated power (380 watts) into 8 ohms, the Carver’s distortion stays under 0.03% to 10 kHz under any condition, and the left channel clears the 0.08% specification at rated power even at 20 kHz. These measurements (and all others) were made with the level controls fully advanced, yielding the THX-required gain and sensitivity.
Figure 1B shows THD + N versus frequency in the stereo mode with 4-ohm loads. Curves were again taken at 1, 10, and 100 watts and at rated output (600 watts per channel). Distortion remains under 0.05% at all frequencies and power levels up to 100 watts per channel. However, as you can see, it climbs to 0.161% at 20 kHz with the amp putting out 600 watts per channel. Let me tell you, that’s a heck of a lot of power, and 1 had the lights and every unnecessary piece of test equipment in the lab turned off to keep the line voltage up. Even then, it was impossible to maintain a 120-volt line with my 20-amp Variac, and I was forced to scale the numbers to account for the line voltage drop.
Figure 2—THD + N for three frequencies in stereo mode into 8 ohms (A), in
stereo mode into 4 ohms (B), and in bridged mode into 8 ohms (C).
I find Fig. 1C particularly interesting, especially when compared with Fig. 1B. Figure 1C shows distortion when operating the A-760x in bridged mode into an 8-ohm load. Technically, bridged operation into 8 ohms is similar to stereo operation into 4 ohms, in that the same total power is avail able and the output stages function with the same voltage and current swings. But performance usually is worse in bridged mode because both channels con tribute to distortion, noise, response irregularities, and so forth. For the most part, that wasn’t the case with the A-760x.
As you can tell from Fig. 1C, distortion at high output levels is actually better (lower) when operating
A the amplifier bridged than when operating it in stereo, and the A-760x delivers 1,200 watts with distortion levels under 0.02% to frequencies in excess of 10 kHz. Even at 20 kHz, distortion is less than 0.07%! I can’t re call seeing performance like this be fore, and I’m most impressed. I’m not certain why there is more distortion with stereo operation into 4 ohms than with bridged operation into 8, but I suspect it may have to do with the different way my 2-kilowatt load bank was configured for the two tests. When handling these kinds of currents at 20 kHz, it’s not unknown for connections themselves to cause measurable distortion. If this, in fact, occurred, the A-760x is even better than my tests indicate.
Figures 2A through 2C show THD + N versus output at 20 Hz, 1 kHz, and 20 kHz into 8- and 4-ohm loads in stereo and into an 8-ohm load in bridged operation. The stereo curves were taken with both channels driven and the left channel measured.
(Data taken on the right was the same.) The seeming discontinuities in the curves are typical of amplifiers that use switched power-supply rails (a.k.a. Class H), as this one does. What’s surprising about these curves is how well behaved the discontinuities are, their virtual absence in the 20-Hz curves, and the rather consistently low distortion right up to the clipping point. Based on these curves, the 1-khz clipping point (1% THD) occurs when the amp delivers 500 watts per channel into 8 ohms, 725 watts per channel into 4 ohms, and when bridged an incredible 1,500 watts into 8 ohms.
With the IHF dynamic-headroom tone burst, stereo output power climbed to 560 watts per channel (8 ohms), 935 watts per channel (4 ohms), and nearly 1,500 watts per channel into 2 ohms—well above Carver’s 2-ohm dynamic power rating of 1,150 watts per channel. Bridged for mono, the A 760x delivered nearly 2 kilowatts of dynamic power into 8 ohms!
= = = MEASURED DATA = = = =
Data taken on one channel is for the left except for THD + N, frequency response, S/N, and separation, which are for the worse channel. Unless noted, data was taken with 8-ohm loads in stereo, both channels driven. Data for output power at clip ping and for THD + N at rated power has been adjusted for sagging line voltage.
Output Power at clipping (1% YHD at 1 kHz): Stereo mode, 500 watts/channel (27 dBW) into 8 ohms and 725 watts/ channel (28.6 dBW) into 4 ohms; bridged mono, 1,500 watts (31.8 dBW) into 8 ohms.
Dynamic Output Power: Stereo mode, 560 watts/channel (27.5 dBW) into 8 ohms, 935 watts/channel (29.7 dBW) into 4 ohms, and 1,480 watts/channel (31.7 dBW) into 2 ohms; bridged mono, 1,900 watts (32.8 dBW) into 8 ohms.
Dynamic Headroom: Referred to 8-ohm stereo rating, +1.7 dB; re 4-ohm stereo rating, +1.9 dB; re 8-ohm bridged mono rating, +2 dB.
THD + N, 20 Hz to 20 kHz: 8-ohm stereo loads, less than 0.086% at rated output and less than 0.019% at 10 watts/channel out; 4-ohm stereo loads, less than 0.161% at rated output and less than 0.027% at 10 watts/channel out; bridged mono into 8-ohm load, less than 0.067% at rated output and less than 0.021% at 10 watts out.
Damping Factor re 8-Ohm Load: 750.
Output Impedance: 11.2 milliohms at 1 kHz, 12.1 milliohms at 5 kHz, 14 milliohms at 10 kHz, and 17.4 milliohms at 20kHz.
Frequency Response: 20 Hz to 20 kHz, +0, —0.09 dB (—3 dB below 10 Hz and at 142 kHz).
Sensitivity: 101 mV for 1 watt (0 dBW) output and 1.97 V for rated output.
A-Weighted Noise: Left channel, -94.7 dBW; right channel, —94.8 dBW.
Input Impedance: 50.8 kilohms.
Channel Separation, 100 Hz to 10 kHz: Greater than 57.3 dB.
Channel Balance: ±0.12 dB.
= = = =
Fig. 3—Frequency response.
Fig. 4—Channel separation.
Fig. 6—Noise spectra.
I measured output impedance and damping factor on the left channel in stereo mode. (There’s no reason to believe the right was any different, so in the bridged mode the output impedance should theoretically double and damping factor drop by half.)
I cannot remember ever measuring an amplifier with such a uniform out put impedance and high damping factor. (The two are related in that damping factor is calculated by dividing the output impedance into the nominal load impedance, i.e., 8 ohms.) In this respect, the A-760x is in a class by itself, and I expect it will reveal the true frequency response of every loudspeaker that it drives.
The amplifier’s own frequency response (shown with expanded scale in Fig. 3) is stellar. The —3 dB high- frequency limit extends to nearly 150 kHz, and the response in the audio band is within +0, —0.09 dB.
Even more outstanding is the infinitesimal difference in response between stereo and bridged operation. Even on the expanded scale of Fig. 3, the two curves overlie nearly perfectly; there's no difference whatsoever in the treble region (where there usually is), and the difference at 20 Hz amounts to only a few hundredths of a decibel.
Sensitivity in the stereo mode was right on the THX target (see "Measured Data"); when bridged, the amp had 6 dB more gain. Input impedance was high (many power amps come in at 10 kilohms or less in an attempt to reduce noise), and channel separation was adequate for all practical purposes. As the crosstalk curves of Fig. 4 indicate, channel separation at 1 kHz easily beats Carver's 70-dB specification, al though it gradually diminishes at higher frequencies.
Since Carver chose to specify in put/output phase difference, I decided to measure it in both the stereo and bridged modes. In stereo, the left and right channels were identical, so only the left is presented in Fig. 5. It handily beats Carver's tolerance of ± 10°. The noise spectrum analyses of Fig. 6, taken for both modes of operation, suggest that the noise floor is essentially "white," with rather small amounts of power line-related hum at 60 and 120 Hz in the stereo measurements. The hum components are virtually absent when the amp is operated in bridged mode, although the lie of the curve is some 10 dB higher. (Six of those 10 dB are ac counted for by the difference in gain.)
Use and Listening Tests
In the lab, I was surprised at how cool the A-760x ran. I mean, the top cover got toasty warm, but considering the amount of power I was dragging out of it, I would have expected a lot more heat. Furthermore, this amp is convection-cooled—no noisy fans, thank heavens. The A-760x’s cool operation is undoubtedly due to its Magnified Cur rent (Class-H) topology, which is much more efficient than conventional designs.
I also blessed Carver heartily when it came to moving the amp from the lab to the listening room. At 39 pounds, it’s really not heavy for a beast this powerful, and the handles make transporting it a breeze. With the handles, the front panel is 19 inches wide, but I’m not sure I’d try to rack-mount the package. The handles and wings to which they’re attached are removable; doing so narrows the front panel to 17 inches, so the amp will fit easily on a shelf with other standard-width equipment.
I admit to having mixed feelings about Class H. I admire the efficiency; but I some times find the sound rather fatiguing. I didn’t with the Carver A-760x. The sound was wonderful—transparent and effortless with anything I tossed at it and with the several different speakers I had at my disposal. Any way you look at it, this is a clean, powerful amplifier with virtually unlimited re sources; it never wimps out. And although the speakers I had in house at the time were not particularly difficult to drive, I do believe that the A-760x will prove able to drive almost any speaker, for good or ill, to its fullest potential.
I recall a number of years ago recording my friend and neighbor, Igor Kipnis, in his home studio playing his favorite harpsichord, “Big Red.” That was back in the early days of digital recording, and I wanted to see if I could hear a difference between dig ital and analog tapes recorded simultaneously using the same microphones and electronics. So did he, since he has an abiding interest in good sound.
Well, I made the recordings, we played them back, and we both agreed that the analog was much better—smoother, nicer, more “real.” Fine, except for one thing. When I asked Igor to play again and I stood where the microphones were and compared each recording with the actual sound of the harpsichord heard by the microphones, the digital recording was much more accurate—pleasant, no, but accurate, yes. My point is that an amplifier like the Carver A 760x may reveal things you’d rather keep hidden. But, I ask you, isn’t that what an amplifier should do?
Adapted from Audio magazine (1947-2000). Classic Audio and Audio Engineering magazine issues are available for free download at the Internet Archive (archive.org, aka The Wayback Machine)
