They say that no new technology can last forever, that subsequent innovation eventually overtakes it. Well, the one clear exception is vacuum tube audio. As glass audio marches past its first centennial, it shows no sign of imminent demise when it comes to musical instrument and audio applications. I used to think up until recently, that with over 100 years of exploration, all significant tube amplifier design ideas have been exhausted. For simple circuit topologies such as SET, where there are only a few parameters to tweak, it's hard to imagine significant originality in any new designs. Of course, the design art, the sonic magic if you will, resides in the choice of power tubes and driver stage, quality of the matching output transformer, and the voicing the amplifier via passive part selection. When it comes to push-pull designs, especially Class AB, the optimization process is far more complicated. As a consequence, the path of least resistance has been to imitate successful vintage designs. You should not be surprised to learn that many modern tube amplifiers are essentially copies of vintage tube circuits with upgraded passive parts and power supplies. With the VM60, JE Audio's designer, John Lam, has taken a fresh design approach proving decisively that you can still put new wine into old bottles,.
The output stage uses two paralleled pairs of KT88 beam power tubes per mono block in an ultra-linear connection. Each pair of push-pull power tubes is biased via an auto bias circuit. There are no user bias adjustments. In this respect, the VM60 is pretty much plug and play. The stock power tubes are Russian reissue Gold Lion. These are said to be biased at a nominal quiescent current of 65 mA. Using a Compu-Bias Meter I measured each KT88's cathode-to-plate voltage and quiescent (plate + screen) current after about a two minute warm-up. The plate voltage was steady at 424 VDC. The quiescent current was within 5% of 65 mA for six of the KT88's, while one pair was a bit of an outlier at 60 and 70 mA. The nominal KT88 plate dissipation is 27 Watt. Since each mono block is rated at 60 wpc, each KT88 contributes about 15 watts to the total output, implying a pretty darn rich Class A bias for a Class AB amplifier.
Another bit of innovation is represented by JE Audio's dual balanced feedback topology (DBFT) which has also been awarded a US patent (#7304535). Not only is feedback applied in balanced fashion but it is distributed into two pairs of feedback loops to provide a higher degree of freedom in optimizing control over amplification bandwidth and distortion products. In the VM60, one feedback loop is applied between the input and driver stages, taken from the plates of the driver stage. The other feedback loop applies global feedback over the entire amplifier, taken from the output transformer's secondary winding. The amount of feedback used in each feedback loop is considered proprietary and is not disclosed by JE Audio. I'm told that a large amount of time was spent in optimizing settings for the two feedback loops. Sufficient global feedback is used to maintain a damping factor of about 12 referenced to an 8-Ohm load.
Feedback applied intelligently makes for meaningful improvements in amplifier performance. Although in the case of triodes in push-pull, or even single-ended triode (SET) output stages, it's possible to make do without global feedback I happen to think that pentodes and beam power tubes need a modicum of global feedback to sound their best. Sadly, some audiophiles are afflicted with feedback phobia. But there's no need for that, the record speaks for itself. There's a host of vintage Williamson type amplifiers using up to 20 dB of global feedback that sound terrific even today and serve as a testimonial to the wisdom of sensible global feedback. Another prime example is the much venerated Harman Kardon Citation II amplifier, introduced circa 1959. It used 32 dB of global feedback and featured no less than three feedback loops.
It is worth discussing how the VM60 handles a single-ended input. Rather than use an input transformer or op-amp for conversion of the single-ended input to a balanced signal, the input stage is used as a long-tail pair phase splitter. The negative output signal is generated by amplifying the single-ended input through one tube. The positive output signal is generated by amplifying the input signal through two tubes. As a consequence, the positive and negative going outputs differ slightly as far as gain and phase shift. According to John Lam, these differences affect the amplifier's overall sound quality and are the reasons why the VM60 performs better in balanced mode than unbalanced mode. He adds that in his experience, a true balanced push-pull power amplifier performs better than an unbalanced push-pull power amplifier when the same type of power tubes is used. I performed preliminary auditions of the VM60 in both single-ended and balanced modes, and can report that I preferred balanced mode, which also sounded a bit quieter. Most subsequent listening sessions were performed with the Pass Labs XP-30 preamp providing balanced inputs to the VM60.
In the lingo of mastering engineer Bob Katz, the music should be able to breathe freely. Complex musical signals have crest factors of as much as 10 to 15 dB, that's a peak to RMS value. Sadly, we're in the middle of a "loudness war," a systematic ongoing increase in the loudness of pop recordings though the use of aggressive dynamic compression and limiting. It is based on the premise that louder attracts a greater radio audience and in general sells better. To quote Katz, "we're making popular music recordings that have no more dynamic range than a 1909 Edison Cylinder." Specifically, the comparison was between Metallica's 2008 Death Magnetic CD ("My Apocalypse" track) with a loudness range of 1.9 dB and an Edison Cylinder recording of "Down Where the Big Bananas Grow" which yielded a loudness range value of 9 dB. Thankfully, classical music recordings are generally free of pop music mastering excesses and make for more reliable program material for evaluation purposes. I'm happy to report that the VM60 was able to negotiate an orchestral crescendo, from soft to loud, without congestion and with sufficient startle factor. In addition, the power range of the orchestra was fleshed out with conviction facilitating a believable concert hall experience.
Sticking for the time being with the stock 12AX7 tubes, Sovtek 12AX7LPS, it didn't take me long to realize that the VM60 while easy to listen to, exhibited several notable sonic issues that got in the way. Subjectively, the presence region was laid back yielding a tonality that was darker than the real thing. It was the sort of balance you might experience well back from an orchestra, say in the middle of a hall. But with closely mic'ed recordings, the timbre shift was unmistakable: female voice being dulled through the upper mids, trumpet being pushed back in the mix, and violin overtones lacking sufficient sheen. Its presentation was fundamentally different than that of the Carver Cherry 180. The Carver offered more bandwidth extension while sounding every bit like a Class AB amplifier. By contrast, the VM60 sounded Class A, but closed-in in the treble, with richer and more liquid textures. There were also concessions in terms of soundstage transparency, overall clarity, and detail resolution. For example, the stock 12AX7 was unable to place a complex passage under the microscope and fully resolve individual melodic lines.
I began to suspect that the stock Sovtek 12AX7LPS tubes, which I'm not a fan of, were seriously handicapping the sonic potential of the VM60. Could they have been responsible for overly thickened textures, fuzzed over detail, and rolled off highs? SPOILER ALERT: THE ANSWER IS A RESOUNDING YES! As I'm fortunate to own a quad of Sylvania Gold Pin 5751s, I decided to roll them in for a quick listen. OK, the Mu is only 70 versus 100 for a true 12AX7 resulting in lower gain, but the sonic improvement was dramatic, amazing to be exact: increased treble extension, enhanced soundstage transparency, and improved microdynamics. At this point I contacted JE Audio to investigate new production front-end tube options. John Lam agreed that the Sovtek 12AX7LPS isn't the best sounding tube for this application. I was informed that JE Audio does offer Russian Mullard and Gold Lion reissue 12AX7 tubes as alternatives for customers (at added cost), and that unless requested to do so by dealers, the amp ships with the Sovtek stock tubes. I asked for and received two pair of Russian Gold Lion 12AX7 for further testing.
Out with the Sovteks and in with the Gold Lions! Since the stock 6922 is already a Gold Lion reissue, my VM60 samples were now fully Gold Lionized. Textures retained their original sweetness and purity, but the tonal balance manifested far greater neutrality with realistic tonal colors, and good high-frequency extension. The soundstage opened up making it a snap to localize image outlines within its inner recesses. Depth perspective and resolution of spatial outlines were now first class. There was an enhanced sense of speed and transient control that propelled musical lines forward with conviction.
The VM60 played an integral part in voicing the latest member of the Basszilla DIY speaker family, the Feastrex Edition, which uses the Feastrex D-5nf as a wide range driver above 200 Hz. It was involved in the crossover design process and proved to be an invaluable reference tool by allowing me to differentiate the sound of various crossover network tweaks. The Feastrex's calling cards are a huge sense of coherence and ability to retrieve plenty of dramatic nuances. The VM60 acquitted itself very nicely permitting the D-5nf's siren song to shine right through, and in particular, its passionate rendering of the vocal range over the bandwidth from 200 Hz to 4000 Hz was very much in evidence. Bass lines were lucid with excellent pitch definition, and the lower midrange, the orchestral power range was communicated with superb clarity. Needless to say, the VM60 took a liking to the Feastrex Basszilla and vice versa.
Next in store for the VM60 was a date with the Acoustic Zen Crescendo loudspeaker, a superlative transmission line design, as well as further head-to-head competition with the Carver Cherry 180 and Triode Corporation's M845 monoblocks. The M845 is an ultra high-end SET amplifier ($28,000 per pair) that delivers 50 watts from a paralleled pair of 845 transmitting tubes. It is known to be a synergistic match for the Crescendo, making this a stern test for the VM60. To its credit, the VM60 delivered perhaps 80% of the M845's magic. The missing 20% were in the areas of spaciousness, tonal color saturation, and microdynamics. The Crescendo, being a very tube friendly speaker, brought out the best in the Cherry 180. In what was a classic confrontation between pentode and UL designs, the edge went to the VM60 in terms of textural smoothness and bass control. However, the carver gained the upper hand by offering a more immersive, you-are-there spatial experience. And in the context of what is an 89 to 90 dB sensitive speaker, the Carver with its 180 watt reserve demonstrated greater ease in scaling the dynamic range from loud to very loud.
-- John Lam