This review page is supported in part by the sponsors whose ad banners are displayed below

For me, the most eyebrow-raising spec of the Lux 91 was the 16-watt power rating at twice that of most other 300B SETs. Even your own LD91 monos are rated at only 7 – 9wpc. How was this almost doubling of power achieved? Does it push the valves to their limits?

"The relatively high power output of the Lux 91 was not really the main design goal. It’s more a side effect of achieving other design goals. Western Electrics themselves show a wide range of operating conditions in their 300B data sheet. They actually show class A1 power levels up to 17.8 watts. Of course the original WE 91 ran the 300B very conservative because these amplifiers where leased—not sold—to movie houses. WE remained responsible for their upkeep. We do run the 300B a little harder than the WE 91, at 32-watt anode dissipation over the original’s  28W. This is still quite conservative. The original WE charts promise us 12.5 watts in this application. That would normally do quite nicely. However, we wanted to achieve two other things.
"One was to remove the self-bias resistor/capacitor combo on the 300B cathode. Not only does this circuit color the sound, it reduces the current flowing into the 300B when the signal increases. That limits available power. Of course classic fixed bias has many drawbacks like the need to frequently readjust and a certain... shall we say volatility. Using the ABS active bias supply module from our DIY range of high-tech solutions, we are able to remove the need for adjustment and the trend to eat tubes while using fixed bias.

"The second thing we wanted for the Lux 91 was optimized overload behaviour. With typical tube amps, as power approaches maximum, the output tube grid starts drawing current. If its driver stage is high impedance as in the WE 91, the amplifier clips hard before ever reaching full power. If the driver is a low-impedance triode, the driver stage will supply the grid current up to a point. The current will charge any coupling capacitor such that it will again reduce the current in the output tube to again cause amplifier overload before ever reaching full power. Worse, these power-limiting effects have a so-called time constant. They last for a certain period after the musical high power peak has passed. We hear the amplifier change sonics as the music moves towards and past a crescendo.

"Combining capacitor coupling and self bias invariably introduces these problems. While such amplifiers may sound very good at low volume levels or with simple music, pushing them quickly reveals their limitations. Our regular design consultant Thorsten Loesch developed a circuit we call reactive interstage. It overcomes the problems introduced by the capacitive coupling and high impedance from a pentode driver - including the high-frequency limitations.

"This pushed the power output of our original Lux (introduced in late 2007) to 12.5 watts, from a 300B operated in the WE 91 manner and with a circuit that borrowed much from the original 91 except for the gain pentode. The original Lux used an EF86, not 310A. The reactive interstage combines a cathode follower with a high-inductance choke similar to the circuit seen in the 1950's Altec Model 1570 push/pull amplifier but here optimized for a single-ended 300B output stage. As a result the current drawn by the output tube grid under high signal levels can be supplied without shifting operating conditions. The original bandwidth limit is completely removed. Clipping when overloaded no longer occurs in the input/driver stage with that hard solid-state like unpleasantness. Instead only the output stage clips - but triodes don’t clip hard. They show a soft clipping/compression characteristic that needs a lot of overload to be subjectively annoying or objectionable. A small amount of overload is easily tolerated."

"The Lux 91 implements a 310A gain stage taken from the original WE 91 amplifier. We added the reactive interstage circuit from the original Lux, then added fixed bias as well as a quite special power supply to overcome the main sources of poor overload behavior in SE amps. Key is that all these ‘extra headroom’ techniques only kick in when needed. They neither degrade the justly legendary sound quality of the WE 91 at lower power levels nor do they stress the output tube in any way unless called upon to serve up short-term musical peaks with authority.

"The result, in summa thermionica, is an amplifier that can be said to go to 11 where common amplifiers using the same tube often do not even reach 10.

"This translates as the ability to drive speakers at the lower end of the ‘high-efficiency’ scale. Finding a good partnering speaker now is easier than it ever was with the original WE 91. The legendary WE 91 sound can reach a larger audience. Many more good sounding speakers exist with a real 93 - 94dB/W/m efficiency than you can find at 97 - 98dB/W/m. In many cases even a speaker with around 90 -91dB/W/m efficiency may be successfully combined with the Lux 91. Just don't put a full live-concert first-row experience of Mahler's Symphony of a Thousand on the books. Regardless, even with all that P:USH in the Lux 91, we still don't recommend that anyone use it with notoriously difficult power-hungry loads such as Dynaudio Mini monitors or a BBC-spec LS3/5."

What exactly is P:USH?

"P:USH is our shorthand for Power for Ultimate Signal Headroom. It sums up all the different techniques we applied to maximize headroom. The choke in the reactive interstage releases energy on signal peaks (which a resistors can't do) to produce the incredible non-clipped transients levels we can measure. Our big L Film Capacitor Universal Power Supply No Sag B+ rail keeps the PSU end up and stiff. Fixed bias via the active bias supply module brings it all home.

"Hence P:USH delivers where others fail. As we are talking about the power supply, it’s worth noting that the Lux 91 Max uses only film capacitors to eliminate the distortions, microphonics and sonic haze that seem to be the hallmark of electrolytics. We could go on about the various design features and parts quality but shall merely mention that the only coupling capacitor in the amplifier is an Obbligato tinfoil & Teflon type made for us and found in our DIY shop; that the resistors in sonically critical positions are Japanese Kiwame carbon types; and that all wiring is pure soft-annealed silver in Teflon dielectric. The tube sockets are very high-quality Teflon types except for the 310A driver and the transformers so critical to sound quality are custom made for us by the renowned Silk factory in Thailand."

"It may be equally worth mentioning that we employ three high-technology power supply modules for the high tension, heater and bias circuits. Those are at least as instrumental to the final sound quality as the magic WE 91 heritage. The same modules (minus some modifications specific to the Lux 91) are also available to DIY amplifier builders everywhere to upgrade existing amplifiers or for new builds. More information is available for the active bias supply module, the ultra low-noise filament supply and the film-capacitor universal power supply.

"Our published circuit schematic above omits one or two values in areas we consider proprietary but the rest is available for those who take an interest in such things."

How much farther can the WE91 circuit be taken?  Can it be further refined or have you reached the stage where only a completely new circuit design would bring significant improvements? Is it indeed possible to significantly improve upon the original WE91 circuit?

"For now we have most likely reached the limits of what, with some justification, may still be called a 91 and actually sounds like an amplifier based on the WE 91 design. Any amplifier which significantly improves on what we have assembled would indeed have to be a design that's largely redone from the ground up, likely also using different output and perhaps even driver and gain-stage tubes. But that would not be a 91 by any stretch of the imagination. As they say, it would be a whole nuther story. And amplifier." - Brian Cherry