Everyone, I'm sure, has by now heard that the innovative and dare I say legend in his own time, Bob Carver, is back! Carver LLC is about tube amplifiers, and soon loudspeakers. It is really the tale of two Bobs: Bob Carver the designer, and Bob Farinelli the business and marketing hub. Two tube monoblock amps are currently shipping: the Cherry 180 and Black Beauty 305. Both share the same chassis, but the 180 is optimized for a KT88 output stage while the 305 is designed specifically for the KT120. Although promoted as a nominal 180 watt amp, Carver tells me that the Cherry 180 will deliver about 270 watts. Although on the surface they may appear to be conventional push-pull Class AB power amps, there's much more here than meets the eye. They should be understood as Formula 1 racing car designs, optimized for high-power performance.
Throughout this review I've taken the liberty of
quoting from Bob Carver responses to my many questions. Bob says that he's a
slow typist, but still he was kind enough to exchange many emails on the topics
I raised. I've never met a man more passionate about amplifier design. By his
own admission, he could stay up all night discussing amplifier design nuances.
Few Technical Details
Just how cool running is the Cherry 180? The front panel bias meter displays the combined idle currents (plate plus screen) for all six KT88s. To obtain the idle current for a single KT88, the meter reading needs to be divided by six. Bob sent me measurements of the plate and screen currents for one of the KT88s. At idle, the plate current measured 11.6 mA at 715V, while the screen current measured 0.3 mA at 375V for a total dissipation of about 8.4 watts. To quote Bob: "Cool enough so we can't injure ourselves if we hold on to a tube. Still, and just the same, I would not tempt fate by holding on for an extended time. I'm sure it would feel quite uncomfortable." The idle current may be adjusted via the bias pot to a maximum of about 150 mA.
And what about tube reliability at high plate
voltages? Bob reminded me that all push-pull designs approach twice their plate
voltage during dynamic conditions due to the coupling between the two halves of
the transformer's primary winding. For example, a 200V drop in one half of the
Cherry 180's output stage (from 725V to 525V) would give rise to a voltage of
1250V in the other half. Furthermore, Bob says that "the only worry would be
arc-over inside the tube. I did some experiments to find out what voltage on the
plate was required to make it arc-over inside. No problem at three thousand
volts." Note that the tubes are warranted for an unprecedented period of one
Fixed bias is used rather than cathode bias, which is the right choice for overdrive conditions. To begin with, cathode bias is wasteful of B+ voltage. An even more important advantage is that since fixed bias is not affected by large plate current fluctuations, the operating point maybe set closer to Class B operation without the penalty of increased distortion.
(U-L) vs. Pentode
"I met Stu Hegeman when he went to CES to show his new Happi One preamp. He told me that ultra linear was the worst fad of all time, that he only designed one that way, that it was the biggest blunder of his career, and he would never design an ultra linear output stage ever again. And he didn't. He designed lots of amps for Sidney Harman and for Lafayette, and only one was designed to be ultra-linear, the Citation II. Sidney, his boss, made him run the KT88s at 0.100 amps per tube with 500 (earlier) volts on the plate at idle (see Sam's photofact). 50 watts for a tube rated at 35 watts! Sidney wanted to have low distortion to compete with Marantz. Stu objected, but Sidney made him. He relented after too many tubes blew up, frying the screen resistors and the Citation division of Harman Kardon had to fix them all."
"The other problem with ultra-linear is that the screen grid is pushed too low when the plate pulls close to ground at high output. This limits the maximum output current to a rather wimpy value and makes the amplifier's power response very peaky, falling off rapidly as we go away from the ideal loading. Speakers do that. Stu taught me all this one enchanting day at CES. Before that I believed everything I read about ultralinear as being the cat's meow and the hot set-up. Then I noticed that the tube amps I loved the most were pure pentode, not ultra-linear (Citation V, Lafayette 550, Mesa Baron, I could go on and on). My heroes are Einstein, Gandhi, Hegeman. Hegeman got it right. So much for Ultra Linear!"
The restorer is simple yet ingenious. It uses a
6AL5 dual diode. Each diode section services one bank of power tubes. The 6AL5
cathodes are connected to the control grids of the push-pull power tubes which
are held at a nominal fixed bias of -47VDC. The diode plates are at an
acquisition threshold of -56VDC. Bob admits that some aspects of the DC restorer
operation are somewhat mysterious to him, as they are to me as well. My take is
that the circuit aids significantly in recovery from hard cutoff conditions.
Under those conditions the KT88 grids act as rectifiers and shift the effective
DC level below -56V, to the point of causing the 6AL5 to conduct momentarily and
pull the DC bias back to its nominal value. Bob estimates the lifetime of the
6AL5 as about 50 years. "I know that seems wrong, but 6AL5's are ubiquitous in
tuners, and 50-year old tuners always have these tubes and they still check as
new. There are so many of them in this world that a guy on eBay sells a string
of them for three cents (each tube) to be used as Christmas
I expressed my doubts to Carver about the audibility of this effect, noting that the speaker microphone effect couldn't amount to more than a few millivolts. Carver's response was as follows: "You are right - about 3.1 mV at 90 dB SPL at our listening position, then multiplied by the gain of the amp, 32 X = 0.0992 volts rms. With a high quality speaker about 89dB or so sensitive at 14 feet away we have about 8.5 forward volts at the speaker, and 0.0992 "speaker microphone" volts. Finally, 0.00992/8.5 = -38.6 dB, just barely noticeable, but quite significant after we get used to listening to it."
"As low as it is. All tubes amps do this a little bit by their nature, transistor amps don't do it at all, and this amp does it a lot by intentional design. It's easy to hear the "oom signal if we hook up a small speaker on the end of about 150 feet of wire and go outside. Hook it across the main speaker in your living room with the amp turned on, but no input. Then have a friend come into your living room and clap his or her hands, stomp around, make noise and sing happy birthday. While you are 150 feet away outdoors with the door closed so you can't hear your friend from the room, hold the small test speaker in your hand and listen to it. You will hear the sound of the room. With a solid state amp, you will not hear a thing, just silence. All tube amps that have a non-zero output source impedance do it, at least a little bit as you surmised. Transistor amps don't do it at all because the almost zero source impedance shorts out any "speaker microphone" signal. Yes, my amps do it a lot by virtue of the current feedback loop. With this amp, lower idle current (as read on the meter), makes the amp do it more. Higher idle current makes it do it less."
Let me first start off with Carver's views on the
subject. "What I'm going to say next will probably fly in the face of
all the stuff we have been taught over the years about amplifier-speaker
interface effects. I was taught that a low-Z output impedance was a figure
of merit for tube amps, and that the best sound followed from that. The
lower the better. It is a long held belief, and it is a false belief. Just
ask David Manley, a contemporary tube-amp designer, and a
champion of about an ohm; I like it even higher when coupled with current
feedback (for a variety of technical reasons).
The damping factor is important, but only as it emerges naturally when we try to make the amp sound better with speakers and our listening room. Forget everything your mother taught you about moderation listen to the room, and let 'er rip take what sounds best and don't think about damping factor!"
"What goes on inside that amp and the experience we have with the sound is too complex to understand in terms of damping factor. Man, this business of tube amp source impedance has occupied me for years and years. I've listened and listened to values ranging from almost zero to a high of about 10 Ohms. I have slowly, painfully come to believe that 1.7 Ohms is the best overall, and is my current favorite. I have listened to new tube amps with low output Z, and don't like them nearly as much as units with about two ohms. Every time I listen to a tube amp I find that I like it best when it can interact with the speaker however it wants to. I know that sounds unscientific, and it is, but by allowing the spectral energy density delivered by the amp to track the speaker impedance, I hear the "classic" tube sound of mid-range glow, soft, sweet highs, and especially the warm bass. And with almost all speakers to boot. I believe it's for two reasons: one, the spectral energy tracking mentioned above, and two, the fact that it can listen to the room better with a high output Z. And I could be so wrong in all of this! This discussion seems kind of goofy because of all we've been taught for a long time about damping factor. Still, and just the same, these beliefs of mine are the results of a series of converging experiments over the years. It's these results that have taught me the way and have led to the design decisions evidenced inside my amp."
My concern with high source impedance is not driven by bass control considerations. Some years ago, during CES, I asked several tube gear designers about what they considered to be an optimal damping factor. Most of them considered 1 to 2 Ohm as a reasonable source impedance design value for a tube amplifier. This translates to a damping factor of 4 to 8 @ 8 Ohm load. Tim de Paravicini, as you might expect, had a much different opinion. He felt that a high damping factor was not a necessary figure of merit. When pressed to name an acceptable damping factor, his answer was one! I agree with Tim to the extent that high damping factors are not a panacea for bass control. It's the source impedance/speaker impedance interaction that worries me as it can result in significant adulteration of the speaker's frequency response.
Driving the MartinLogan Summit X hybrid ESL, the
Cherry 180 shelved the treble by 2 to 3 dB. Some tube amps do much worse. You
would expect that from a capacitive load approaching 1 Ohm impedance at 20 kHz.
Actually, it's not a bad EQ effect for the Summit X as it tends to sound too
bright in my listening room. Granted, that's an extreme example, but all
bass-reflex loaded box speakers possess large impedance peaks on either side of
the box tuning frequency. A power amp with a high source impedance will
emphasize the bass range and any frequency band associated with an impedance
peak (e.g., near a crossover frequency). Granted, such tonal changes may not be
subjectively objectionable, and considering the impact of listening room modes,
the amp may well by chance smooth out the in-room response. The important point
to remember is that just because a high-source impedance amp sounds good with a
particular speaker is no guarantee of success with other speakers. Such an amp
can sound terrific in one context and terrible in another. Let me make it
perfectly clear that my definition of excessively high source impedance is
greater than 4 Ohm. The Cherry 180's 1.6 Ohm source impedance is perfectly fine
for any tube amp, and actually terrific for a pentode output stage. Bob don't
change a thing!
Feedback is essential to the success of any
push-pull power amp. The Cherry 180 uses three feedback loops. Current and
voltage feedback loops are taken from the output transformer secondary to the
input stage. Another short loop from the transformer's primary to the driver
stage provides high-frequency compensation, a scheme Bob says he copied this
from the Marantz model Nine.
It became obvious pretty damn quickly that the
stock Chinese 12AX7 had to go. Its sound could be best described as texturally
grainy and a bit screechy through the upper mids. A pretty miserable combination
if you ask me. Substitution of a Siemens ECC83 made for a night and day
difference. Raytheon black plates sounded very nice as well, richer and tonally
a bit darker than the Siemens. Bob Carver's feedback on the subject of 12AX7
tubes was as follows: "Changing the 12AX7 tube changes the sound.
Telefunkens are my favorite less noise and less modulation noise. To a large
extent, we can change the inside loop gain by changing the bias and sort of
simulating the sound of different tubes. Also changing the feedback switch, or
both. All change the fine details of how the amp responds to the room and to the
speakers. It is soooo much fun to change tubes with a friend during a listening
session far into the night!"
Out came my stash of Telefunken ECC83s. The
Telefunken clearly bettered the Siemens in my listening tests, being smoother
and better balanced tonally. Therefore, I'm in agreement with Bob that the
Telefunkens are as good as it gets in this context. However, toward the end of
the review I popped in the Russian Vaskhod (sunrise) 12AX7. This new production
tube sells on Amazon for $11. I didn't expect much, as previous Russian and
Chinese new production samples left me unimpressed. Much to my surprise, the
Vaskhod gave the Telefunken a run for its money no kidding! Substitution of
vintage 12AT7 types for the stock Chinese tube did not make for a dramatic
difference in sound quality, though my final choice, a Sylvania gold brand,
resulted in an enhanced depth perspective and more palpable image outlines.
Don't expect the Cherry 180 to generate a
euphonic camouflage for bad recordings. In an ideal world with perfectly matched
tubes, and totally balanced phase splitter and output transformer, even order
distortion products would be entirely absent. And although that's never the
case, a push-pull amp's distortion spectrum is dominated by odd order distortion
products, mostly third order. By contrast, singled-ended designs tend to
generate primarily second order distortion products which are quite consonant to
the ear lying exactly an octave above the fundamental frequency. I thought it
would be instructive to pit the 180 against a high-quality SET amplifier. It
just so happened that the Coincident Speaker Frankenstein II monoblocks were in
the house for review, which gave me the opportunity to compare their respective
first watts in the context of a high-sensitivity loudspeaker the Zu Audio
Omen Def (98 dB/W/m). To be sure, these amps offer far different sonic
perspectives. But the catch is that the Omen Def is slightly deficient in upper
midrange energy, which suggests that the amplifier richer in second order
distortion would provide a more synergistic match. And that (no surprise here)
turned out to be the Frankenstein monoblocks, offering far sweeter sounding
violin overtones. But then no high-power Class AB push-pull design would be my
go to power amplifier in this context.
The Cherry 180's serious test was to be the
Analysis Audio Omega ribbon/planar magnetic loudspeaker, a load known to be
power hungry. Since the 180's 4 and 8 Ohm impedance taps are tied together,
there are only two choices available. I tried both and noted significant tonal
balance shifts. The clear winner was the 4/8 Ohm tap as it provided the more
realistic tonal colors. My Omega planars have previously partnered a variety of
solid-state amplifiers, including the Lamm Audio M1.2 Reference monoblocks,
Spectron Musician III Mk2, and Arion Mk 1000. This was to be its maiden voyage
with a compatible tube amplifier.
It turned out to be a stunning combination. Power
together with tube liquidity and dimensionality. In fact, the most dimensional
reproduction of any amplifier driving the Omega. The soundstage was painted with
bold strokes, conjuring an exquisitely nuanced depth perspective. Image outlines
were fully 3-D with plenty of breathing space between performers. All this
combined to produce an immersive, you-are-there, spatial experience. There was
also plenty of dynamic fire and rhythmic drive to stoke the music's emotional
content. I must have heard the Hammond B organ on the Jazz
Me cut [Lesley Olsher: Jazz Me, BDM78003] a million times before.
David Manley had coupled the Hammond to a pair of Klipsch K-Horns, and they were
played loud to generate adequate SPL at the microphone. Remarkably, the Cherry
180 was able to generate vivid colors and dynamic shadings approaching the live
mic-feed studio experience.
Bass extension did not quite match that of the
solid-state competition, but I had nothing to complain about when it came to
bass control and pitch definition. Tube amps tend to trade textural liquidity
for transient speed, and that was the case here, though that's a trade I'm
always willing to make. There was plenty of low-level detail to resolve, though
it remained organic to the musical tapestry, never in your face. The Omega never
sounded any better. What I liked the most was the effortless gear change from
soft to loud and its gift for resolving complex orchestral passages with nary a
hint of congestion.
History repeats itself; it has happened again with the 12AX7 tubes used in this amplifier! Except this time I'm the boss who blundered and put the wrong tubes in!
If you have one of my amps with a new 12AX7 tube in the front-end, call Bob at (859) 258-9794 and he will send you an old (vintage) one. Or get that $11 unit from Amazon. It's on us.
Thanks again for a beautiful and thought provoking review.