deHavilland Audio KE-50A Monoblock Power Amplifier
Reviewed by Dick Olsher

This is more than just a review, it is the story of the re-birth of an American classic. The Fisher 50A’s genesis took place in the early 50s. In an age before the arrival of ultra-linear, 6550s and KT88s, the sonic battle raged between pentode and triode output stages. Pentodes definitely had the power edge, though discerning listeners preferred the bass control and purity of triode sound. The engineering conundrum was this: how to maximize the power output of a triode push-pull power amp? One obvious approach was to run a beam power tube with significant plate dissipation, say a 6L6/1614 in triode mode with the screen tied to the plate. But in Class AB1, where most power amps operate (no grid current), power output is still only a fraction of what is obtainable in pentode mode. The Fisher engineers decided to push the envelope and operate the output stage in Class AB2. The difference between Classes AB2 and AB1 is the presence of grid current at high power levels. The payoff is a dramatic boost in available power as plate voltage is allowed to swing over a much wider range. The idea may have originated in the commercial sector, but clearly the Fisher Corporation deserves kudos for blazing a new trail in the consumer market with such an design.

Of course, this is easier said than done. As soon as grid current starts to flow, the power tube’s input impedance drops significantly and RC coupling from the driver stage becomes ineffective. A viable class AB2 output stage requires a low-impedance cathode follower stage that can sink and source current, i.e., deliver some power to the grid circuit to cope with positive grid swings. The 50A uses a pair of triode-connected 6CL6 pentodes as a push-pull cathode follower. The cathode followers drive the output stage through an interstage transformer (IT); a Lundahl 1629 (1.125:1 step down) is used in the KE-50A. Very neat, as the IT primary's resistance provides the bias voltage drop for the 6CL6s and fixed grid bias voltage for the output stage is applied through the IT’s secondary winding. The Williamson amplifier’s influence is very evident in the first couple of stages. A triode voltage gain stage (one section of a 12AU7) is DC coupled to a cathodyne phase splitter (a second 12AU7 section). This should be a familiar topology to thousands of Dynaco power amplifier owners. The driver stage is a 6CG7 dual triode, essentially a 9-pin version of the 6SN7. It is AC coupled to the cathode followers using 0.1µF Jensen paper-in-oil caps.

For the output stage, Kara settled on the gorgeous Gold Lion Genalex KT88 reissue tubes. These are operated conservatively at about 2/3 of their maximum ratings, and in combination with a slow warm-up, the payoff is long life expectancy and minimal power tube problems. Power output is about 40 wpc of pure triode power into 8 Ohm, of which about 20 watts are obtained in Class AB1 and the rest in AB2. At small signal drive the amp operates in class A with both tubes conducting. As the drive signal increases, one output tube cuts off and the amp transitions smoothly to Class B operation. They say that beauty is in the eye of the beholder, and I have to confess to taking a liking to the KE-50A. I found it to be a good looker, certainly far superior to the original in terms of fit and finish, and with just the right complement of elegance and geek. A bias meter is provided on the top deck which is calibrated specifically for KT88s. Once set, the bias was rock solid over a period of several months. Two thumbs up for using Cardas binding posts – the best in the business. No amp should be without them! The review samples were configured for 8 Ohm service. However, internal 4 Ohm transformer taps are also available.

Kara’s Story

It’s time to turn over the podium to Kara Chaffee. It’s her story of how she became enamored of the 50A and was moved by its siren song to embark on the re-issue project. “I had heard in reputation about the 50A from some long time audio friends. Then a friend of mine sent me an old tired pair about 2 -1/2 years ago and wanted me to refurbish them, which I did. When you first see the original amplifier, it looks like a creature from the 1940s . The speaker output was taken from a funny 3-prong plug , and you had to plug an ammeter directly into the circuit to obtain the bias , via a 1/4" phone jack. All very awkward. And with the very industrial utility chassis in the original, it was not an elegant design visually. In fact, the original Sams Photofact  on 50A was published in 1954. So be it. After studying the circuit and becoming enthralled in the sound , I realized that the critical design features of the 50A were not antique at all. The design choices were examples of doing it the right way, and not cutting corners for cost savings reasons. Some amplifiers have used these elements , but 50A is among the very few designs to use them all. So let's look at them.”

“The power supply uses a choke-input configuration. This is a passive  and simple form of voltage regulation providing a stable
voltage supply for the tubes. It is rarely used because it requires a power transformer to be 30% larger to obtain the same voltage as a conventionally configured  (capacitor-input) power supply. And it requires a high quality inductor to be used with the larger transformer. These were, and are major cost items in a tube amplifier. Next, the final audio signal driver for the KT88 tubes is a wideband interstage transformer.  It costs nearly as much as the power transformer, and was almost never designed into an audio power amplifier. The benefits of this design element are several. The transformer drive allows the pair of KT88 tubes to be run in triode mode, and still produce nearly 50 watts of power--and all this at a very conservative plate voltage. Normally one could expect about 25 watts from a pair of KT88s when run as triodes. The transformer drive also has a unique sonic quality to my ear.  It is so evenhanded from top to bottom, that I feel that the sound is ‘all of a piece’ Many amplifiers emphasize the mids and upper mids. I call this a ‘mail slot’ kind of presentation. The 50A does not emphasize a part of the audio band. And it has remarkable clarity in the lower midrange and upper bass. Take a jazz recording where there is piano doing chords on the left hand and bass and drums.  It does not blur all of this and you can easily make musical sense of the rhythm . So many amplifiers lose this part of the sound.”

“Next, the 50A uses tube rectification. In 1954 , this was the only option for supplying the DC voltage required by the tubes. The original amp used a pair of 5V4G tubes, which was a heavy duty configuration.  Many audio listeners prefer the smoothness and tone of tube rectification, and a huge side benefit of the 5V4G tubes was that they warmed up slowly. This offers a very gentle startup for the amplifier  and contributes greatly to reliability. Most power tubes fail on startup because the power supply surges very high voltage, before the amplifier catches up with the power supply. In the 50A , it comes on like a little lamb. Combining the gentle startup, with the conservative  power supply voltage,  power tube detonations should be really rare. The last feature of note is the tube-regulated bias supply. The bias voltage is stabilized by a simple shunt-type tube voltage regulator. It is a sophisticated feature that was rarely incorporated in a home audio amplifier.”

Original Vs. New

I asked Kara Chaffee to detail for our readers how the KE-50A compares to the original. “I have incorporated all of the essential features of the original Fisher design, in my KE50A. The KE50A uses dual tube rectifiers. I chose to use a pair of heavy duty rectifiers, known as "damper diodes" from the tube television era. The advantage to the damper type rectifier, is a very low voltage drop across the tube itself, and a very slow warmup.  Much in the spirit of the original. The KE50A uses a very high quality interstage driver transformer. On the oscilloscope , it has better bandwidth and a prettier waveform at high frequency than the original.  I think it is at least as good. The original Fisher 50A had a ton of gain--too much really. Approximately 0.2V would drive the amplifier to full power.  I kept the basic topology of the original circuit, but I eliminated the very first gain stage. My amplifier requires 1.6V to drive it to full power, which is not at all a problem for any preamplifier.  The benefit was that I eliminated two coupling capacitors, a gain trimmer potentiometer and one whole stage of amplification. I get a more transparent sound with these changes. This streamlining  allowed me to free up one-half of a dual triode, which was no longer needed for audio amplification. I doubled this stage up with the existing voltage regulator stage, to provide a more robust regulator, which is even more stable.”

“The original 50A used 12AU7 tubes for all the audio voltage amplification. I kept the first stage as a 12AU7, and the Amperex 12AU7s really do a beautiful job here. For the second audio voltage gain stage, I selected the 6GC7 tube. It can produce the high amplitude voltage swings needed in the 50A, at lower distortion than any 12AU7. I felt it to be the best tube for this job. Regarding the tubes , I also had greater liberty in design, than the original designers. In the original 50A , the only resistors used in the audio circuit were the common 1/2 watt size. This necessarily limited the operating  current  for each tube. I was able to double or triple these current values without even coming close to the operating limits of the tubes.  In my experience running a beefier current often sounds richer and smoother. If one looks at the tube performance curves, it’s easy to see that the higher current operating point is in a more linear region of the tube's characteristic. The original 50A was fitted with  metal envelope type 1614 power tubes. They were basically an industrial version of a 6L6.  I designed the KE50A with the new Gold Lion Genalex KT88 reissue tubes in mind. They are wonderful tubes and have the advantage of being made today and easily obtained. Of course , my amplifier will accept any 6L6, 1614, 7027A, KT90, KT88 or KT66 tube. I incorporated an onboard bias meter, which was absent in the original 50A. It is calibrated for KT88s, but smaller tubes can be biased using a pair of jacks provided on the top deck of the amplifier. I also brought in creature comforts in the form of Cardas speaker binding posts and an on-off switch. You turned the original 50A on, by plugging it in. That was it!”

The Sound

Choice of matching loudspeaker is always a key consideration for low and moderately powered tube amps. The 50A is most comfortable with nominal loads in the range of 4 to 16 Ohm, and ideally, would appreciate a speaker sensitivity of at least 90dB. It was happiest in the company of my Basszilla Platinum Edition Mk.2 DIY speaker, a 96dB sensitive load. Hence, much of my time was spent auditioning that coupling. But the 50A also had the opportunity to drive the Salk SongTower and the DALI Helicon 400 Mk.2, both being 4 Ohm nominal loads rated at a sensitivity of 88dB, as well as the QUAD-57 ESL.

If you value music’s inner beauty, then most likely the 50A will instantly push your emotional buttons. My first impression was of a totally relaxing presentation imbued with rich, dense, and liquid harmonic textures. However, there was more to the presentation that just a beautiful veneer. Emotions were given full scope of expression as pitch, volume and rhythmic modulations were reproduced fully intact. In general, midrange textures displayed stupendous levels of sweetness and purity of tone. Tonally, the 50A sounded full bodied, a perception I’ve come to associate with tube-rectified designs. The lower midrange, the orchestral center of gravity projected a proper tonal weight in the best tradition of classic tube sound. Rotating the 50A into the system after spending considerable time with both Ultra-Linear and pentode beam power push-pull tube amps highlighted its essential lack of electronic glaze and freedom from upper octave brightness, grain, and glare. Ah - the joy of triode sound!

Tonal colors were solidly saturated, vivid, and remarkably realistic. It’s important to note that the 50A accomplished all this without conceding either low-level detail or spatial resolution. One of its endearing strengths was the finesse with which it unraveled layers of orchestral harmony. Overdubs, massed voices, and transient decay presented absolutely no problem. The soundstage excelled in conveying depth perspective and “painting” image outlines with precise brush strokes. I should point out that Kara really takes the time to voice the amplifier and that her choices in tubes types are therefore near perfect. It turns out that the 12AU7 that makes up the input and phase splitter has the greatest impact on sound quality. I experimented with a couple of substitutions, including a Mullard CV4003 with box anode that has done really well in other applications. But in every instance the stock Philips Miniwatt 12AU7 gave the best tonal colors. You could do a lot worse and most likely no better than the stock tube complement.

It so happened that the Berning ZH-230 ZOTL amplifier (36 wpc into 4 or 8 Ohm) was also in house for review and these two amps at times were auditioned one after the other. It made for interesting sonic contrasts. What I found instantly startling about the ZH-230 was its superlative transient speed at the point of attack. By contrast, conventional transformer-coupled tube amps appear to round off the leading edge making transients appear softer and more liquid than the real thing. The difference in speed between the two amps was noticeable. Also significant was the ZH-230’s edge in immediacy, a refreshing lack of veiling allowing the listener to access the inner recesses of the soundstage. The Berning, on the other hand, was unable to quite match the 50A in terms of harmonic color saturation and textural density. At the end of the day, it boiled down to a personal preference between the comforting yin of classic tube sound versus the considerable yang of ZOTL technology.

In the context of the Salk SongTower, the 50-A did not match the Audio by Van Alstine Ultravalve’s bass line delineation. However, driving the basszilla, bass impact and definition were a pleasant surprise. And not only that, it negotiated the transition from soft to very loud without any complaints, with plenty of boogie factor, and with dynamic conviction I did not expect from a 40-watt amplifier.

Conclusion

What blows my mind is that this Fisher circuit is some 60 years old. Yet the wisdom of those original Fisher engineers shines through the mists of time. Refined and updated by Kara Chafee, the KE-50A represents the closest approach I’ve heard from a push-pull power amp to single-ended triode magic, but with sufficient dynamic headroom to drive a significant subset of real-world loudspeakers. It is the sonic equivalent of a vintage Port wine. Kick back, take a sip, and enjoy the music! Best in class and a must audition for anyone serious about reproduced music.

Specifications
Power Output: 40 Watts RMS (4 and 8 Ohm)

THD: 0.2% at 1W, 20 Hz – 20 kHz; at 20W: 1% at 20Hz, 0.13% at 1 kHz, 3% at 20 kHz; at 40W: 0.7% at 1 kHz
Input Sensitivity: 1.6 V

Dimensions: 17-1/8” (L) x 9-3/8”(W) x 9-1/2”(H)
Weight: 42 lbs.
Price: $9,995/pr

Company Information
deHavilland Electric Amplifier Co.
2401 NE 148th Court  
Vancouver, WA 98684
Voice: 360-891-6570
Website: www.dehavillandhifi.com
E-mail: kara@tubewizard.com