This review page is supported in part by the sponsor whose ad is displayed above
The Composite 10 Design
The Linear A amplifier is full of creative & innovative ideas that you might not fully appreciate from a quick perusal of its specifications or a simple visual inspection. One of the more interesting aspects that ought to intrigue SET devotes is that the Linear A is a single-ended, Class A, ultralinear device. Four single-ended EL84s per side are linked in parallel to create a composite tube with characteristics that are "almost identical to the legendary Japanese 10 tube" according to Evans. This composite approach to tube circuits reminds me of the Conrad-Johnson ART preamplifier and its descendants but to my knowledge, nothing like it has yet been attempted with a power amplifier in quite this way. The claim is that the Linear A amplifier measures as an almost perfect single-ended triode while delivering higher power output and much more extended frequency extremes - along with the musical magic normally associated with SETS. [The series-connected super-ultralinear KT88s of Audiopax too endeavor to 'clone' the perfect triode, in their instance a combination of 2A3/300B THD behavior with the power output of a 211 or 845. - Ed.]

Tom explained that "the most complicated design problem we had was finding a tube that first and most importantly sounded great, ideally was a triode, didn't run off lethal voltages and could be driven by an op-amp to produce over 20 watts of output power!" As many 6moons toob heads will recognize, that combination of traits doesn't exist in a single tube. "We knew that we could, from both the op-amp & output transformer point of view, assemble a composite valve by paralleling two or more tubes. What we needed was a tube that was both a pentode and a triode, had high gain and was easy to drive and sounded wonderful .. oh blast deary me!"

What they chose was the much used and well-liked EL84 tube. "The data sheet shows it has an output power of 6 watts but boy, it's also got 10% distortion! More importantly, it only requires 5 volts RMS to achieve this, well within the capabilities of an op-amp. By using four EL84s, we can get 24 watts. If we use the suppressor grid (the other input connection) and configure it in ultralinear, we get a small increase in power and a huge drop in distortion due to the more linear tube characteristics giving us a pseudo triode."

The EL84 hasn't been on most peope's screen as one of the great tubes because it has been primarily used in entry-level implementations because of its low cost. If you're lucky enough to have listened to the budget Almarro A205A SEP EL84 integrated amplifier, you know that the EL84 is one great tube. But what if the EL84 were implemented with the most cutting-edge design possible? Then what would happen?

"Those of you versed in triode lore will blanch at our choice. However, one of the best-sounding triodes is the 10 tube and its derivatives. Their problem is low power output, low gain, their input capacitance and highish anode impedance. I hear you say that the 10's high anode impedance means it has high output impedance. Actually, its high anode impedance means it's easier to drive and kinder on the output transformer. With any given transformer, change the tube it sees and it changes the sound it produces. In one test, we compared a 10 with a 2A3, using the same single-ended output transformer with a resonant frequency at 86kHz, its core gap adjusted for best inductance. The 2A3 achieved -1dB figures of 18Hz to 12 kHz at 1-watt output. The 10 measured -1dB at 28 Hz to 20 kHz for the same output! Using our composite EL84 as a "super 10 tube", we have effectively produced a triode from the transformer's perspective and one whose electrical parameters nearly match a 10!"

Tom designed the Linear A to operate in Class A because he feels it offers the best sound quality, is on all the time so it doesn't pulse the mains and doesn't stress the amplifier's parts like class B or D designs. Class A also works better with differing power level demands and speaker impedances, an important consideration as you'll learn in a moment. Tom and Peter also wanted a design that would produce 20 watts or more make it compatible with more speaker designs than most SETs and would incorporate Tom's and Peter's favored topologies of op-amps with transformer-coupled valves.

Op-amp Input Drivers
If it wasn't innovative enough to use a composite single-ended tube approach, Tom and Peter went a step further to do something that to the best of my knowledge has never been done before: They use op-amp technology to drive their composite single-ended tube. Tom giggled, anticipating my reaction. "Look, they're using those new-fangled op-amps with valves. Oh dear! Go back to the drawing board and better luck next time, boys! But hang on a minute. Op-amps and valves used by themselves work well as you know. We have found through our research that using them together raises the combination to a new level of performance that they can't match alone. Op-amp technology offers the strengths of low distortion, compact topology, high gain and wide bandwidth but can't do large output voltage swings. On the other hand, tube technology offers large output voltage swings and some tubes offer high gain - but as for the rest, they are challenged. Combine the two devices and you get the best characteristics of both - if you do it right."

A major design goal for the Linear A was to react to the smallest changes of input voltage to maximize reproduction of the subtlest details. "This is the reason I have always preferred op-amps. The tube and transformer act as one. The tube amplifies the voltage and the transformer changes this voltage into a current. The EL84 has a current gain of 11 milliamps per volt of input and a voltage gain of 32. Putting four in parallel, we get 44 milliamps per volt of input and the output transformer converts the voltage gain of 32 into a current gain. Hence, the composite output stage converts 1 volt from the op-amp driver stage to 1.72 amps for the loudspeaker. This gives 20 watts at 4 ohms and only requires the op-amp to produce 3.88 volts. Couple this to the global feedback being enclosed in the feedback of the input op-amp and this makes the input circuit a virtual high current gain op-amp."

The Power Supplies
Another innovative aspect of the Linear A is the use of two separate power supplies to power the input and output circuits. "The power supply on conventional power amplifiers uses the same HT voltage to drive both the output stage and its driver stage. Herein we faced a problem. We couldn't run our op-amps and tubes from the same voltage. Therefore we built two separate power supply rails. One runs the Lithos board for the op-amp driver stage, the other provides a near ripple-free supply to run the tubes. Rather than try and come up with a fancy regulator circuit to reduce the pulsing of the mains from overstressed power supply diodes, we went for dual-mono 3-stage filtering using choke smoothing (two per channel) to accomplish a >150dB ripple rejection from a 60Hz supply. This approach also has the benefit of eliminating common mode distortion. This occurs because although you would assume that the output voltage of the device is measured with reference to ground, the amplifier also produces signal between output and HT rail(s), the effect being that the signal appears again in the next amplification stage via the HT to add distortion. In the Linear A, we've overcome this ill and avoided hooking up the output devices to a common HT rail. We split the HT in half after the first choke filter, into a further capacitor/choke/capacitor filter for each channel. The ultra low noise/distortion driver stage and active bias circuit are powered by a Lithos 7 power supply variant."