Carver Silver Seven-t monoblock power amplifier New Measurements

Sidebar 2: New Measurements

Both in order to ascertain whether the samples of the Silver Seven-t that J. Gordon Holt auditioned were ostensibly the same as when Robert Harley had reviewed them, and to see if there were any measured reasons for this disparity between our writers' opinions, I carried out a set of measurements similar to those that Bob had done to support his original review.

The measurements were performed with Stereophile's Audio Precision System One, using high-power resistive loads built by Robert Harley. The distortion vs frequency graphs were made with the SS-t and Audio Precision grounding arranged to give the lowest measured THD and noise. A Heath/Zenith 8-bit storage oscilloscope was used to capture the square waveforms, and DRA Labs' MLSSA system was used as a distortion analyzer for the 50Hz and 19+20kHz spectrum analyses.

The amplifier tested (SN 00428) was preconditioned by driving 100W of 1kHz into an 8 ohm load for 30 minutes. The chassis had gotten very hot by the end of this time, but monitoring the distortion didn't show any change through this period after the first minute, implying that the amp was working correctly.

Though the FTC-recommended preconditioning level is one-third full power, which thermally stresses a class-B amplifier to the maximum, I arbitrarily decided upon a 100W level for preconditioning as this sample of the Silver Seven-t turned off due to overheating after about 25 minutes at the one-third power level. (A later test with the other sample, SN 00490, showed that it would drive one-third power, 191.67W, into 8 ohms for the FTC-specified 60 minutes, though the chassis became far too hot to touch.)

Though the SS-t is rated at 575W into 8 ohms, its use of a sophisticated, switched-voltage output stage means that it would be unwise to expect the amplifier to be able to deliver this power into an 8 ohm load on a continuous basis. Carver's design philosophy, which I believe first appeared in its diminutive but powerful "Cube" in the early '80s and which I have also seen in amplifiers from NAD, Proton, Hitachi, and Soundcraftsmen, is to give the amplifier's output stage a dual personality. The amplifier appears to be conventional at low volume levels, its output stage powered by relatively low-voltage rails. However, when the musical signal (which can often have transient voltage peaks very much higher than its average value) requires a brief high-power burst which would otherwise drive the output stage into clipping, the output stage switches automatically to higher-voltage rails in order to pass these peaks without waveform distortion. The amplifier's heatsinking is not adequate for maintaining the higher output power indefinitely, however, and is actually more appropriate for a lower-powered design. Nevertheless, with typical music program, the effect is of a much more powerful amplifier than its size and heatsinking would suggest (footnote 1).

And the SS-t does indeed appear to be a very powerful amplifier. On a short-term basis with a 1kHz tone, the clipping point into 8 ohms (defined as the output power where the THD+noise reached 1%) was 619.5W (27.9dBW, footnote 2), while into 4 ohms it raised 910.5W (26.6dBW). Into 2 ohms, I measured an astonishing 1087W (footnote 3) at 1% THD (24.3dBW). Though these clipping power figures are better than the amplifier's specification, which describes the SS-t's maximum power as 575W into 8 ohms (27.6dBW), 900W into 4 ohms (26.5dBW), and 1000W into 2 ohms (24dBW), they must be regarded as approximate rather than absolute, as, not surprisingly, the AC line voltage from the wall drooped at these very high power levels. Measuring 117.7V RMS with the SS-t not handling any signal, the line voltage dropped to 114.3V RMS with the amp clipping into 4 ohms and 107.2V RMS with it clipping into 2 ohms. For reference, RH had measured 608.4W (27.8dBW) and 878W (26.4dBW) into 8 and 4 ohms, respectively, for the same 1% THD point.

A final test I performed after the testing had been finished was to see how long the SS-t could sustain its specified power. It managed to drive 575W at 1kHz into 8 ohms with the distortion remaining below 0.068% for almost nine minutes before shutting down. This is excellent performance for this kind of amplifier design, implying that it will be practically burstproof under normal conditions of use.

Fig.1 shows the frequency response of SN 00428 measured at a 1W level into an 8 ohm resistive load. This is identical in shape to that mentioned in Robert Harley's review, a slight rise being apparent in the high treble, followed by a rolloff above 20kHz. This response was identical at 100W into 8 ohms and very similar into 4 ohms at the same voltage (ie, a 2W power level), though with very slightly more ultrasonic content. At higher levels into 4 ohms, however, a slight kink appeared at ultrasonic frequencies. Fig.2 shows the response at 200W into 4 ohms, plotted up to 100kHz, while fig.3 shows the response at 200W into 2 ohms. This is plotted only up to 50kHz, due to what I believe to be the amplifier's protection circuitry cutting off the output above this frequency. I suspect that at these frequencies and levels, I was approaching the edge of the SS-t's output-stage design envelope which, as implied earlier, appears to be optimized for the demands of real-world music program rather than for steady-state test tones.

899CSSTFIG01

Fig.1 Carver Silver Seven-t, frequency response at 1W into 8 ohms (0.5dB/vertical div.).

899CSSTFIG02

Fig.2 Carver Silver Seven-t, frequency response at 200W into 4 ohms (0.5dB/vertical div.).

899CSSTFIG03

Fig.3 Carver Silver Seven-t, frequency response at 200W into 2 ohms (0.5dB/vertical div.).

Robert Harley found that the SS-t's output impedance was very high for a solid-state design, measuring 1.24 ohms at 20Hz, dropping to 1.16 ohms at 20kHz. My measurements of output impedance were identical, within experimental error, measured at an approximate 2.83V, 1W/8 ohms level. (The 1kHz figure of 1.2 ohms at this 2.83V level was maintained with the amplifier delivering a hefty 28.3V RMS.) My experience suggests that this high impedance will correlate with both JGH's and RH's opinion that the SS-t lacks definition in the bass. Whether it will prove musically problematic to other listeners will depend on the speakers they use. In my opinion, it will be less noticeable with well-damped sealed-box loudspeakers, for example, than with rather underdamped reflex designs, though Larry Archibald did note in last December's "Final Word" that he "was dissatisfied with it [the SS-t] on the bottom [of the Infinity Betas]."

These output power, output impedance, and frequency-response measurements suggest to me that the amplifier as auditioned by JGH was substantially the same as when RH had listened to it for his review.

Were there any other measurements which would correlate with any of the listeners' descriptions of sound quality? Looking at distortion and noise levels, fig.4 shows the values measured between 20Hz and 20kHz at 1W and 100W into 8 ohms. The curves are nominally identical above 500Hz, the apparently higher level of distortion at the lower level below this frequency being due, in my opinion, to the measured distortion sinking below the amplifier's noise floor. Note that above 2kHz or so, the THD rises above 0.1%, reaching 0.6% at the top of the audio band, which is a hair above the specified figure of 0.5%. The distortion level also rose with the amplifier driving lower-impedance loads, fig.5 showing the THD+noise at 2W and 200W into 4 ohms, and fig.6 that with the amplifier delivering 4W and 200W into 2 ohms. My interpretation of these graphs is that the Silver Seven-t does have a slight problem maintaining linear behavior at high frequencies into low-impedance loads. The levels of distortion above 2kHz or so might become audible with some kinds of music with the amplifier driving loudspeakers that feature very low impedances in this frequency range—the MartinLogan Sequel II used by Robert Harley, for example, which drops below 4 ohms above 5kHz.

899CSSTFIG04

Fig.4 Carver Silver Seven-t, THD+noise (%) vs frequency at 1W into 8 ohms (top) and 100W into 8 ohms (bottom).

899CSSTFIG05

Fig.5 Carver Silver Seven-t, THD+noise (%) vs frequency at 2W into 4 ohms (top) and 200W into 4 ohms (bottom).

899CSSTFIG06

Fig.6 Carver Silver Seven-t, THD+noise (%) vs frequency at 4W into 2 ohms (top) and 200W into 2 ohms (bottom).

The superbly low distortion levels at low frequencies and high levels implied by the Audio Precision measurements were confirmed by looking at the spectrum of distortion products with the amplifier delivering a low-frequency (50Hz) tone at high levels. Even into 2 ohms at an 800W level, the absence of distortion products was excellent, as can be seen from fig.7. To examine the high-frequency linearity in more detail, I used a 1:1 mixture of of 19kHz and 20kHz tones, sourced from CD, using the Esoteric D-2 D/A processor which produces negligible levels of downband intermodulation products with this signal—see fig.20 in RH's review elsewhere in this issue.

899CSSTfig07

Fig.7 Carver Silver Seven-t, spectrum of 50Hz sinewave, 10Hz–2kHz, at 800W into 2 ohms.

Fig.8 shows the spectrum from 300Hz to 30kHz with the amplifier reproducing the two tones into 8 ohms at a 1W level. Any difference product at 1kHz due to the SS-t is below the resolution floor of the measurement system, while the sidebands at 18kHz and 20kHz are also vestigial. This is excellent performance. Increasing the level to 200W into 8 ohms, however, brings up both the 1kHz difference product and the sidebands, but not to any level that I would consider problematic. At high levels into lower impedances—fig.9 shows the spectrum with the amplifier driving 400W into 2 ohms—rather more sidebands appear, the highest of which measure –55dB with respect to the 19kHz level. The cursor shows the level of the 1kHz difference product at approximately –61dB, which is still low, however.

899CSSTfig08

Fig.8 Carver Silver Seven-t, HF intermodulation spectrum, 300H–30kHz, 19+20kHz at 1W into 8 ohms.

899CSSTfig09

Fig.9 Carver Silver Seven-t, HF intermodulation spectrum, 300H–30kHz, 19+20kHz at 400W into 2 ohms.

Finally, I examined the squarewave waveforms at a number of levels and loads. (No anti-aliasing filter was used for these figures, any ringing or overshoot being therefore due to the amplifier and not to the storage 'scope. The sampling rate was 250kHz.) Fig.10 shows a 1kHz squarewave at a 1W level into 8 ohms—a perfect shape, with a slight rounding to the leading edge due to the SS-t's ultrasonic rolloff. Increasing the power into 8 ohms (fig.11) introduces a slight overshoot, as does lowering the load (figs.12 & 13).

899CSSTfig10

Fig.10 Carver Silver Seven-t, 1kHz squarewave at 1W intro 8 ohms.

899CSSTfig11

Fig.11 Carver Silver Seven-t, 1kHz squarewave at 200W intro 8 ohms.

899CSSTfig12

Fig.12 Carver Silver Seven-t, 1kHz squarewave at 400W intro 4 ohms.

To sum up these measurements, I agree with JGH that Carver's Silver Seven-t offers an astonishing number of watts for its $2000/pair price. My enthusiasm must be tempered, however, by the fact that I feel that whether the amplifier's user will be able to exploit all that power will depend on a number of other factors, most important of which will be the kind of music he or she chooses to play. With spikey, transient-rich classical music or rock music with high-level drum tracks, the SS-t will deliver what its specification promises—all day long. But with music that more nearly approaches the steady-state—synthesizer or organ, for example—asking for all that power for more than a few minutes will lead to a rapid increase in the heatsink and case temperature, with the result that the amplifier will eventually shut down to protect itself. This will be presumably more likely if the user pairs the SS-t with loudspeakers that are both insensitive and feature a low impedance.

Regarding its sound character, the SS-t's somewhat loose, not very well-defined bass can undoubtedly be attributed to its very high output impedance, in my opinion. I also suspect that the interaction between this high source impedance and the low load represented by the midrange/treble IRS Beta towers gives a voltage-divider action that usefully depresses the speaker's treble. This might help explain why LA was so impressed by the combination. On the other hand, the Sequel IIs also have a low impedance in the same region, yet RH was still bothered by the amplifier's sound in the treble. (I don't have any data on the impedance of JGH's Sound-Lab speakers.)

Whether the SS-t has a treble character or not is perhaps due both to its high output impedance and to what appear to be increasing levels of high-frequency harmonic distortion when it is asked to drive low impedances. Whether this distortion will be audible or not will depend on the listener's preferred listening levels, on the loudspeakers used, and on the degree of masking offered by his or her favorite kinds of music.

I suspect, therefore, that the Silver Seven-t's sound will be somewhat system-dependent; prospective purchasers should audition the amplifier carefully with their own loudspeakers and their own favorite music before making a final decision. In view of Gordon's finding that the SS-t takes a long time to reach its optimum performance, they should also ensure that it is fully warmed up by the dealer before any auditioning takes place.

It is possible that this follow-up review may be moot, however. According to a report in the industry newsletter Inside Track (Vol.VI No.21), Carver introduced a revised version of the Silver Seven-t at the June 1990 CES in Chicago. The Mk.II SS-t monoblock was said to feature a "modified transfer function" which, presumably, would affect the amplifier's sound quality, thus the relevance of both RH's and JGH's findings.—John Atkinson



Footnote 1: See my "As We See It" in August 1989, Vol.12 No.8, for further discussion of this subject.

Footnote 2: Stereophile's convention for rating amplifier power in dBW is to subtract 3dB from the calculated figure when the load impedance halves. A "perfect" amplifier would then give the same dBW figure into any load, an easy paradigm to understand.

Footnote 3: The main 12A fuse was replaced with a 15A fuse for this measurement, as cautiously recommended in the owner's manual when the SS-t is to be used with very-low-impedance loudspeakers.

COMPANY INFO
Carver Corporation
Company no longer in existence (2019)
ARTICLE CONTENTS

COMMENTS
ok's picture

such explicit audio reviews nowadays..

Graham Luke's picture

As appealing and utilitarian-looking as a Soviet-era tractor.

grigorianvlad's picture

I disagree, the Soviet-era tractors had much better imaging and sound stage than these

Graham Luke's picture

Ah, the joys of collective agriculture!

Bogolu Haranath's picture

May be Stereophile could review the new version of Carver Silver Seven tube mono amps ($32,000/pair) ....... They are rated 700/900 WPC ......... or, Stereophile could review the other less expensive Carver tube amps :-) .......

shstrang98's picture

I thought Stereophile wasn’t allowed to mention the brand name Carver let alone review their products.

Or has that passed?

X