This review page is supported in part by the sponsor whose ad is displayed above
The underlying premise of Tenor's original 75i OTL -- and by extension all OTLs -- is the direct connection from tube to speaker. Any intermediary device, be it transistor, transformer or MOSFET, could compromise the sound with a loss of that 3D magic and transparency. Yet François believes that it is not the use of a solid-state output that causes problems but how the output is configured. If a solid-state output stage is used for gain, the tube magic can be lost. Yet in a more passive configuration as Tenor uses theirs for conversion of voltage to current, things are far more benign.


"Some competitors might use one or two tubes as part of the gain stage but none have all the gain in the tubes. We believe ours to be very unique. Of course we cannot say with absolute certainty since there are so many small amplifier companies. But for an amp with the power and refinement of the 350M, we are confident that this is unique to Tenor. 95% of the sonic signature comes from the first three tubes. The rest account for only 5%. The [solid state] driver stages are so neutral that they do not interfere with the overall neutrality of the amp. The sonic signature of the driver stage is truly insignificant. Although the output stage does have its own distortion spectrum, the way Michel designed it, the nature of the distortions can be quantified and minimized through phase cancellation. Provide the opposite phase signal which is basically a distortion signal and it will cancel the distortion signal inherent in the output stage. Although this is a gross simplification, Michel uses a similar technique to minimize the effect of distortion in the output circuitry."


François is not discussing the phase of the audio signal but "the phase relationship of a harmonic signature". By manipulating that, you can diminish the signature of the transistors and retain only that of the tubes. Because everything is fully regulated, there is no 'tubey' signature either, just pure transparency.


The 350M design at its simplest is a two-stage tube OTL voltage amplifier with a six-tube output buffer. The tubes form a functional mini amplifier within an amplifier, producing 13 watts at 500 ohms, which could actually power an appropriately rated speaker. However, since there are not many 500-ohm speakers on the market, there were three design choices:

  1. A traditional pure tube OTL amp. Michel did the calculations. To scale up the original 75-watt OTL to 300 watts required drawing over 2,000 watts from the wall - not a practical solution.
  2. An output transformer - a fatally flawed solution from Tenor's perspective. Yes, tube lovers always have that love-hate relationship with transformers. Looking for that big tight bass, you will need a big heavy transformer. But you also get lots of wire, stray capacitance, interference, nonlinearities and phase shift. Everything in life is a compromise so while top-of-the-line output transformers are good, they're also expensive, large and require very careful engineering.
  3. Finally, while MOSFETs are not perfect, they do not suffer from frequency and phase-related problems. In the 350Ms they supply no voltage gain. Their sole purpose is to provide current and convert 500 ohms to a more reasonable 8 ohms. Apply the corresponding mathematics, convert the power from the tube stage and you have essentially the output power of the 350Ms. This amplifier really is in many ways an OTL amp driving two large buffers.


The first gain stage uses two Tung-Sol ECC803/12AX7, the second gain stage a single JJ ECC99 followed by six GE 7044 buffer tubes. Although the 7044 is not currently in production, Tenor has been stockpiling them for over three years and claims to have a very large supply. Michel uses these tubes for what they do well - linear voltage gain as a driver.


"Every high-grade hybrid amplifier has tubes but they only share a small part of the gain. I do not know of any hybrid that has all of the gain and drive in the tube stage. This is our unique design. Although it has been greatly refined, it is the same general structure of the 75i OTL." The tubes sit on boards mounted on a fairly sophisticated dampening system. Although the tubes were originally chosen for low microphonics, the suspension cancels any residual vibrational effects. Tenor's testing showed that output from the speakers caused microphonic interference with the low-level audio signal. Their solution is a suspension tuned to audio band frequencies. The MOSFET output stage's level is too high to be influenced by audio band vibrations and therefore suspension is not necessary but at the input stage it is vital. While not a custom product, the suspension is uniquely sourced for this amp and far more advanced than a simple spring.


One of the joys (or frustrations) with tube equipment are the endless possibilities of tube rolling. Change tubes and you will change the sound. Is it an infinite flexibility or a frustratingly expensive tone control? Your call but there's no arguing that tubes do alter the sound. If you're an obsessive tube twiddler, maybe the 350Ms are not for you. As with everything about the 350Ms, their tube selection is nothing short of fanatical. Tenor sources high-quality Tung-Sols, GEs and JJs. Using a proprietary machine, they burn in each under an exact operating load and temperature for more than 100 hours, then test and hand select. Once installed in the amp, everything is burned in and adjusted using custom-made trimmers. If any minor variations are detected, the process is started all over. Once the unit is calibrated, the amp is fixed bias. Even with time and temperature fluctuations, Tenor claims absolutely no drift in values.


When I asked François about tube rolling, he really grimaced. "We honestly believe that this will not be useful especially for the driver. The entire circuit has been optimized for these specific tubes. The spectrum of these tubes is known and the circuit calibrated. Someone with no knowledge or possibly a preconception might want to try and you might get a different sound. But we do not believe you will get a better sound. We go to extraordinary lengths to select the tubes and match the pairs. This ensures that both amplifiers in the mono pair produce the same sound. We've heard too many instances of monoblock tube amplifiers with different sound characteristics because of improper tube matching."


Published tube life is 10,000 hours or five years. Tenor's prototype amplifier, run continuously and stressed to higher than normal levels, has already exceeded 7,500 hours with absolutely no drift, not even 1/10mv. François credits this to Michel's unconventional circuitry, producing very high gain with very very low current. The GE 7044 buffer tube, also built to last 10,000 hours, was originally designed for computer use.


"We use them at much lower current than they were designed for. We also use coolers which have been shown to double the life of the tube. Should we claim 20,000 hours? I don't know. 10,000 hours should be conservative. The Tung-Sols are the finest tubes of their kind available - extremely low noise floor and used by us within 10 to 15% of their design current. They just are not 'working'. The 12AX7 runs at a nominal current of a few hundred microamps!"


Although I am not an engineer, there appear to be some unusual design elements here, with François hinting at some very unorthodox circuitry. When questioned about the differential input circuits or the ability to get large tube gain while drawing minimal current, he confirmed Michel's designs to be quite unconventional. "A knowledgeable circuit designer looking at the amplifying circuit might conclude that this should not work as specified. Yet it does quite well in fact." As you read on, you will see that it does indeed work quite well.


MOSFETS
The hybrid moniker of course implies an amp with at least one tube stage and one solid-state stage. But from there it gets a little fuzzier. Putting a solid-state gain stage following a tube gain stage might have little more effect than mating a tube preamp with a solid-state amp. Sure it's hybrid but is it the tubes or transistors which impart the final sonic character? Tenor has a rather strong opinion about this. They want the tubes to completely control the sound of the amp, with the solid-state side remaining neutral, providing impedance and current conversion and then just getting out of the way, sonically speaking. Apparently that's a lot easier said than done.


I sense the pride of Michel's design is his new MOSFET output stage. It's that defining element which provides virtually unlimited power yet maintains neutrality, allowing for ultimate transparency so that the amp retains the character of its OTL tube design. Of course tube input stages are certainly not unique. Every designer understands that tubes are superb voltage amplifiers. They're linear, operate with simple circuitry and a well-designed version can be made without negative feedback. Likewise, the inherent nature of the MOSFET is very tube-like. Its high impedance allows it to be directly driven by tubes without the need for coupling stages. Tenor uses the MOSFETs essentially as followers whose only function is to convert voltage to current. The key is that they provide no voltage amplification, with actually a minor loss of gain.


These MOSFETs provide the same role as the large paralleled output tubes in the original 75i OTL. But can they make the sonic magic that was the 75i? As François says, "If the 75i was magic, the 350M is hyper-magic." It all begins with what Michel calls Harmonic Structural Integrity (HSI). Michel starts essentially with a mathematical model of what he is trying to achieve. Whether it's transistor or tube, it really does not matter to him. The only proof is the sound. "It is not our goal to reduce distortion to near zero like Halcro for example, which is a fantastic scientific achievement. But, when you compare our two designs... let's just say what is important to us is non-perceivable distortion."


Yet when even minute distortions impact the sound, he'll fiddle with the design. The entire output stage design was tweaked up to RF frequencies. "Even with a 40MHz spectrum analysis, you will not find any tendency towards oscillations. In testing Michel found a very small oscillation at 30MHz and spent a lot of time eliminating it. He considers it extremely important and although high in frequency, it can interfere with the audio band."


Now let's take a walk down Tenor's MOSFET path for just a small taste of the 350Ms' design and build fanaticism. Then take this MOSFET discussion and extrapolate it to every element of the 350Ms down to the nuts and bolts and you'll get a hint as to what the Tenor team has developed.


Tenor uses a special series of MOSFETs with unique characteristics required for the amp. Tenor believes that a typical failing of any device with solid-state outputs is the improper matching of the devices which cause one or a few to provide greater output than the others. This phenomenon causes tons of ills including thermal problems, improper biasing, shortened life and a host of detrimental sonic issues. To assure precise matching and control, Tenor developed a number of proprietary devices and techniques. Their MOSFET testing machine is one such device. Although François asked that photos of the device remain confidential, he did offer some insight into its operation.


"For us it is essential that the devices be matched at their working temperatures. The device essentially heats 36 of these transistors simultaneously to their let's say 60°C working temperature and allows very precise measurements. The transistors are tested in the exact same conditions found in the amplifier for voltage and temperature. The outputs are monitored on a precision scope. Given a specific voltage at the gate, each matched device produces the exact same current at the output of the transistor. Every transistor is matched to a maximum 1.5% variance. After installation into the amp, we go through another series of tests for reliability. If they were matched to, say 10% for example, you would not hear the difference in the sound. But we want to be sure that none of the output groups would try to take the current by itself and overheat. This level of precision causes a tremendous amount of rejects and costs a fortune but unfortunately, there is no free lunch. When you have transistors perfectly driving the exact same current, you get a tremendous impact and purity. This impact is strong and even - on both sides of the cycles of the music."


Continuing down the fanatical path of obsessive detail, François described the MOSFET mounting techniques in excruciating detail with a dissertation on the design, quality control and shape of a washer - yes, an 'ordinary' washer. But the interesting thing is that once you see how the little pieces fit together as a coherent Tenor Philosophy, these overly obsessive individual details fall into place and begin to make sense. Back to the mounting process of the MOSFETs. Remember, the goal is to make all output MOSFETS react in unity with total precision and with an ability to respond almost instantaneously. The bias control system maintains the precise operating parameters and their biasing relies on accurate temperature measurements. With this system, the thermal transfer between MOSFETS and heat sink must be uniform and stable for the life of the amps - regardless of any thermal expansion or contraction.