Hitachi 2SK134 / 2SJ49: The Lateral MOSFETs That Gave the Luxman L-58A its voice.

There are certain components in vintage audio that stop being merely components.

The Hitachi 2SK134 and 2SJ49 are among them.

I encountered this complementary pair in a Japanese-market Luxman L-58A, and they are a large part of what makes that amplifier such an interesting machine. The L-58A isn't simply an old Luxman with MOSFETs bolted onto the output stage. It comes from that fascinating period when Japanese manufacturers were actively exploring what power MOSFETs could do differently—and perhaps better—than the bipolar output transistors that dominated solid-state audio.

And Hitachi's lateral MOSFETs were something genuinely special.

The Pair

The 2SK134 is the N-channel device and the 2SJ49 its complementary P-channel partner.

Hitachi designed the family specifically with low-frequency power amplification in mind. The original data describes the devices as offering high power gain, excellent frequency response, wide safe operating area, good complementary characteristics, and built-in gate protection. The 2SJ49 is rated at 140 volts drain-to-source and 7 amps of drain current, and was explicitly specified as complementary to the 2SK134.

These were not generic switching MOSFETs that audio designers happened to discover worked well in an amplifier.

They were audio devices.

During the late 1970s and early 1980s, MOSFET output stages carried a certain amount of mystique, but there was good engineering behind the enthusiasm. Lateral MOSFETs have operating characteristics that are particularly attractive in linear power amplifiers. Their thermal behavior is considerably friendlier than that of traditional bipolar output devices, and their high-frequency characteristics allow designers to construct extraordinarily fast output stages without necessarily creating an unstable monster.

Hitachi became one of the names most associated with that technology. And Luxman knew exactly what to do with it.

Enter the Luxman L-58A and the L-68A

The L-58A is a 100-watt-per-channel (L-68A is 110w) integrated amplifier into 8 ohms. Luxman's own archive identifies it simply and rather wonderfully as a “MOSFET Stereo Integrated Amplifier,” rated at 100 watts per channel with 0.015 percent distortion.

But those numbers don't really explain the amplifier. The interesting part is what Luxman was attempting with the output stage.

Period documentation describes the L-58A's 2SK134/2SJ49 output MOSFETs as extremely fast devices and notes a standing bias current of more than 300 mA. Combined with a high-speed driver stage, Luxman referred to the resulting topology as “Notchless Class A.” The design was intended to avoid the conventional switching behavior around the crossover region rather than simply running the entire amplifier as a brute-force Class-A heater. Luxman also quoted a remarkable 300 V/µs slew rate for the power amplifier section.

Those were serious numbers. Especially for 1979 and 1980.

There is something very Japanese about the engineering philosophy here. Instead of pursuing one specification in isolation, Luxman attacked the problem from several directions: fast devices, substantial bias, local feedback, reduced open-loop gain, improved open-loop bandwidth, and careful attention to distortion before global feedback was applied.

The MOSFET wasn't the marketing story pasted onto the front panel. It was central to the architecture.

Why Lateral MOSFETs Became Legendary

A power transistor in an audio amplifier has an unenviable job.

It must handle substantial voltage and current while behaving linearly enough that a tiny musical signal can ultimately control a loudspeaker without the output stage imprinting too much of itself onto the result.

Bipolar transistors can do this extraordinarily well. Some of the greatest amplifiers ever built use them.

But lateral MOSFETs behave differently. One important characteristic is their relatively benign thermal behavior at higher currents. They love to run hot, by nature. With conventional bipolar output stages, increasing junction temperature can increase conduction, which increases temperature, which can increase conduction again. Designers therefore need careful bias compensation and thermal coupling to keep the output stage from wandering—or in a badly designed circuit, destroying itself.

Lateral MOSFETs have characteristics that make thermal stabilization considerably easier in the region where audio amplifiers operate.

They are also voltage-controlled devices. The driver circuitry isn't required to supply the continuous base current demanded by bipolar output transistors. Then there is speed.

The Hitachi devices were startlingly fast for power semiconductors of the period. Luxman's literature surrounding the L-58A claimed an effective device transition frequency around 500 MHz, using that speed as part of its argument for dramatically improving the amplifier's open-loop high-frequency performance. These were absurd for the time period.

That doesn't mean an L-58A is reproducing 500 MHz audio. It means the devices themselves weren't becoming the dominant limiting factor anywhere remotely near the audio band. That gave designers room to work. And work, they did.

And Then You Listen to One

This is where specifications stop being enough. My experience with the 2SK134/2SJ49 came through a Japanese-market Luxman L-58A, and the amplifier immediately reminded me why people remain fascinated with this generation of MOSFET equipment decades later.

There is an ease to it. Not softness exactly. Not the exaggerated warmth that people sometimes project onto anything old and Japanese...or tubes. The L-58A can be remarkably quick and articulate. But it doesn't sound mechanically hard. That distinction is important.

There is plenty of transient speed, yet cymbals don't become sheets of metallic glare. No sibilance to be found anywhere. At all! Midrange textures arrive intact. Voices have body without becoming syrupy. Guitar has bite without turning into an outline of itself. And perhaps most importantly, the amplifier seems reluctant to draw attention to the transition between delicate and powerful. It simply scales.

That characteristic makes more sense once you understand what Luxman was trying to accomplish with the heavily biased MOSFET output stage. “Notchless Class A” sounds like wonderful late-1970s Japanese hi-fi marketing—and it certainly is—but there was a real engineering objective behind the name. The output devices were being kept well away from the crude on/off handoff traditionally associated with Class-B operation. You can hear why Luxman cared about it.

Not a Tube Amplifier in Disguise

Vintage MOSFET amplifiers are frequently described as sounding “tube-like.” I've never found that description particularly satisfying. A good lateral MOSFET amplifier doesn't necessarily sound like tubes. My Luxman SQ-38FD is a tube amplifier. It does what tubes do, and there is no mistaking one for the other. The L-58A is still a solid-state amplifier.

What the MOSFET output stage seems to avoid is some of the hardness associated with lesser solid-state designs of the same period.

It retains control. It retains speed. Bass remains governed by a low-impedance solid-state output stage. Transients can be extremely fast. Yet there's a wonderfully unforced quality to the presentation. Rather than calling it “tube-like,” I think fluid is closer. And that word comes up repeatedly when people discuss the better lateral-MOSFET amplifiers.

The Problem With Owning One Today

There is, unfortunately, a dark side to all of this. The original Hitachi devices are long obsolete. I mean...almost "unobtanium".

That turns every surviving 2SK134 and 2SJ49 into something worth protecting. Search for replacements today and you quickly encounter discussions from L-58A owners trying to determine what can legitimately substitute for them. Modern lateral MOSFETs exist, and devices from later Exicon/Semelab families are often considered when repairing vintage lateral-MOSFET amplifiers. But replacing output devices isn't simply a matter of finding something with vaguely similar voltage and current ratings. Capacitances, ransconductance, bias behavior, circuit compensation and stability all matter. And, pairs must be matched to really shine.

There is also the counterfeit problem. Many, if not all, counterfeits are embloazoned with the original Hitachi symbol instead of the "Hitachi" name.

Obsolete Japanese power semiconductors are exactly the sort of components for which an attractively printed part number on an online marketplace should inspire suspicion rather than confidence.

L-58A repair discussions going back many years are full of owners trying to locate genuine 2SK134/2SJ49 devices or determine whether a modern substitute will behave correctly in the original circuit.

If you have an L-58A with its genuine Hitachi outputs intact, you have something worth preserving.

More Than a Vintage Part Number

What makes the 2SK134 and 2SJ49 interesting isn't rarity alone. Rare components are everywhere in vintage electronics. These devices represent a particular period when semiconductor technology opened a door and ambitious audio engineers immediately ran through it.

Hitachi produced a genuinely unusual power transistor optimized for linear amplification.

Luxman took it and built an amplifier around the things it did well. The result was the L-58A.

One hundred watts per channel, substantial standing bias, an extremely fast output stage, low open-loop distortion, enormous slew rate by contemporary standards, and that wonderfully understated Luxman exterior hiding an amplifier that was technologically quite adventurous. More than forty years later, the engineering still matters and so does the sound. God, that sound.

There are newer MOSFETs. More powerful MOSFETs. More efficient amplifiers. Amplifiers with vanishingly low measured distortion and specifications that would have seemed absurd when the L-58A was built.

None of that diminishes what Hitachi and Luxman accomplished here in the late 70s. Sometimes an old semiconductor isn't interesting because technology has surpassed it. It's interesting because it represents a moment when the technology, the circuit designer, and the intended application lined up almost perfectly.

The Hitachi 2SK134 and 2SJ49 were such devices. And hearing them in a properly functioning Luxman L-58A makes a convincing argument that they were much more than another pair of transistor numbers stamped onto a TO-3 form factor.

THE BRITISH CONNECTION

The story of the lateral audio MOSFET did not end in Japan. Britain developed its own strong lineage of these devices, and for vintage-audio people that lineage became increasingly important as the original Hitachi parts disappeared.

Semelab was one of the key names. Its BUZ900 and BUZ905 complementary lateral MOSFETs became familiar to amplifier builders and repair technicians looking for TO-3 devices in the same general tradition as the early Hitachi parts. These were not ordinary vertical switching MOSFETs repurposed for audio. Like the Hitachis, they belonged to the comparatively rare family of lateral power MOSFETs intended for linear operation.

That matters because the lateral construction is really the heart of this entire story.

The electrical behavior that made the Hitachi devices so attractive to Luxman and other amplifier designers did not automatically appear in every transistor carrying the word MOSFET. The British devices preserved the important characteristics: wide linear operating capability, comparatively benign thermal behavior, ease of paralleling, low drive requirements, and freedom from the secondary-breakdown problems associated with bipolar output devices.

Later came Exicon.

Introduced by British specialist distributor and semiconductor company Profusion in 1993, Exicon continued the lateral-MOSFET idea after the Japanese parts became difficult or impossible to obtain. Exicon's own history explicitly places its devices in the lineage of the famous Hitachi audio MOSFETs.

That makes the British devices more interesting than simple substitutes.

They represent a continuation of the technology. Exicon continues to describe its current lateral MOSFETs as devices designed specifically for high-power linear use, with audio amplification as a primary application. The company's modern ECX10N20-W6 and ECX10P20-W6 complementary pair is rated at 200 volts and 8 amps, while retaining the characteristics that made lateral MOSFETs attractive in the first place: excellent frequency behavior, resistance to thermal runaway, simple drive requirements, and suitability for high-quality linear amplifiers.

The old metal-can versions are especially interesting to vintage enthusiasts.

For years, Exicon produced TO-3 lateral devices that visually and mechanically carried forward the classic power-transistor format used by Hitachi, Semelab, and countless amplifiers of the 1970s and 1980s. Exicon eventually ended production of its TO-3 versions as manufacturing costs increased and demand declined, describing the decision itself as the end of an era. That description feels appropriate. A Hitachi 2SK134 or 2SJ49 inside a Luxman L-58A represents the Japanese beginning of this story.

A Semelab BUZ900/BUZ905 pair represents the British continuation. And Exicon represents the remarkable fact that the basic lateral audio MOSFET concept survived long after most of the semiconductor industry moved toward devices optimized primarily for switching power supplies, motor controls, and other applications.

There is something satisfying about that.

The original Hitachis became legendary because they were designed around the needs of linear amplification at a moment when amplifier designers were eager to exploit what MOSFETs could offer.

The British manufacturers looked at the same set of virtues and decided the idea was worth keeping alive.

They were right.