AXIOM DC300B AMP PROJECT


We started the Alliance Audio Project Program in 1999. We work with people building interesting designs that can offer "benchmark" performance. Our only requirement is that the design and the parts developed in this effort are made available to other audio constructors and completed units are made available to Audio Enthusiasts. The Axiom Direct Coupled 300B Monoblock Amplifier is the first project nearing completion.

What's also exciting is that this "alliance" has also resulted in some developing some excellent audio performance components in addition to the transformers. The Axiom Project has developed the MaxiCap Power Supply Film Capacitors, the UltraCap Hybrid High Slew Rate Audio Capacitors, and the Axial Series Shock Mounted Tube Sockets.

Axiom DC300B Design Information

The Axiom is part of an "Audio Alliance" project with Peter Chappell, a wonderful transformer designer in the UK, and Dr. Bob Hoekstra, an amplifier constructor. He is a professor at University of Central Florida. Dr. Bob is the same guy who builds the "Hoekstra" motors for the NASCAR and other Racing Teams. Prior to his motor career, Bob owned and operated recording studios for more than 20 years.

Bob is working on an interesting single ended amp design called the "Axiom" An axiom, by definition, is an unproven statement that in combination with other scientific principles leads one to new and correct conclusion.

Axiom I  Amplifier Tubes should be operated as variable voltage, constant current devices.

Axiom II  Each amplifier stage should have an independent power supply to eliminate phase cancellation and amplifier stage interactions in the power supply.

Axiom III  An un-bypassed tube diode in the cathode circuit reduces distortion

It's an interesting and innovative design with the following design elements:

Each of the three tube amplifier stages has its own independent tube rectified "quasi" choke input power supply. This provides the following benefits:

"Quasi" choke input supply is one with a low value input capacitor and the input capacitor has an extremely high slew rate. The low value and high slew rate combine to make the capacitor virtually invisible to the audio signal while substantially reducing the hum in the power supply. The choke input capacitors are 1 MFD with a slew rate of more than 500 V/uS.

Maximum decoupling of active amplifier stages to provide excellent isolation. The use of 50 Henry chokes in the 6Y6 and 300B power supplies and two split chokes in the 6J5 power supply act very similar to a constant current source feeding the plate load chokes and the output transformer.

The rectifier tube, choke and capacitor values can be optimized to provide a clean, fast, noise free energy reserve for each specific amplifier tube. Limited output swing problems due to DC coupling are resolved by independent power supplies per tube stage.

Separate power transformers for the high voltage and filament secondaries. This allows the tube filaments to be preheated prior to the application of high voltage to the plates of the tubes.

The use of multiple secondaries in a transformer, properly executed, provides balanced secondary voltages. Changes in the line "mains" voltage has little effect on the amplifier maintaining the required voltage ratios to operate properly.

You will notice the independent power supplies are "stacked". The high voltage "line of the 6J5 tube stage is the ground "plane" of the 6Y6 tube. The high voltage "line of the 6Y6 is the ground plane of the 300B tube. This concept is one key to maximizing the performance potential of the audio tubes.

The use of separate stacked supplies does away with the high offset voltages common in DC-Coupled amplifiers. The required cathode resistance value on the 6Y6 and 300B tube cathodes are reduced.

Using "stacked" power supplies permits the use of foil and film type power supply capacitors. Not only do they sound better than other types but also they are the most reliable.

The 6J5 power supply uses two split chokes in a "hum bucking" configuration to provide phase cancellation of hum and noise. By tuning the potentiometer on the 300B filament the noise can be "tuned" to below 1 mV RMS.

Plate choke loading on the first two stages offers an increase in gain when compared to a resistive plate load circuit. It also allows the 6Y6 and 300B to be operated with "Combination Bias", a ratio of fixed and self (cathode) bias.

The relatively low DC resistance of the plate load chokes provides transient performance and an effortless dynamic presentation that cannot be achieved with plate load resistors. The tubes operate at optimum gain, with very low distortion.

By selecting the correct DC Resistance of the Plate Load Chokes, the optimum ratio of fixed versus cathode bias can be selected. This Combination Bias provides the Axiom with a stability and tune-ability that is not possible in standard direct-coupled designs. The owner can use a variety of audio tube brands to "voice" the amplifier to achieve the best synergy with specific loudspeakers and music sources.

Plate Load Chokes function differently than Plate Load Resistors. A choke presents an apparent constant current source to the tube. A choke is a "current storage device". A voltage drop across a choke opposes a change in current thus a high inductance choke presents a high impedance load and an effective constant current source to the plate of a tube.

A resistor is a proportional device. A voltage drop across a resistor results in a proportional change in current. The plate load choke on the other hand is nonlinear

A plate load choke circuit is more linear than a resistor loaded circuit. This can be observed by re-graphing a tube curve with the current held constant.

The cathode bias resistor and bypass capacitor that were on the 6J5, 6Y6 and 300B have been replaced by tube diodes. A tube diode provides a dynamic rather than static resistance device to bias the cathode. One of its significant features is that while it provides an extremely stable DC bias voltage, its dynamic resistance is lower than its DC resistance.

For example, in the 6J5 circuit, the diode tube acts like a 175 ohm resistor providing a 1.7 Volt DC Bias voltage whereas its dynamic resistance is 33% less or roughly 125 ohms. The 6J5ðfs distortion is significantly reduced.

Testing establish that the "sound" performance of tube diodes varied significantly. The best performance was realized with TV damper diodes.

As part of the amp project Bob developed the differential equations for amplifier and entered all the mathematical terms, (there were a lot of them), into a Math software program designed to solve phase and gain equations in the form of a Bode plot analysis. It lets him analyze each tube stage, each tube stage with its power supply and the entire amp.

The software was instrumental in identifying the importance of the correct values for the bias resistance, the DC resistance and inductance of the plate load chokes, the capacitors and the chokes in the power supply and the inductance of the output transformer. The inductance value of the output transformer is critical for good performance; we are lucky that Peter designs have always provided excellent performance. The bandwidth of the Axiom using Peter’s output transformer is 20 Hz to 90 kHz +/- 1dB

The effectiveness of the mathematical transfer function analysis was verified by numerous listening evaluations and testing. Changes were evaluated by listening in Mono and comparing the change in one amp against an unmodified amp as a baseline.

Until you look at the schematic, the Axiom, with three amplifier stages, three separate tube rectified power supplies and three cathode bias tube diodes might seem a bit complicated. But, there are no power supply decoupling caps, grid resistors, bleed resistors, grid stoppers or plate load resistors, "just lots of iron". There are no regulation or constant current sources in the supplies and yet the amp acts like there are. You can swap tubes without worrying about bias...it is self-adjusting (like any well behaved amp should be). It is not unstable as other direct-coupled circuits because of the Combination Bias.

Listening evaluations have ranked this as an outstanding 300B design, breathtaking dynamics, a sparkling high end without harshness, a sweet rich midrange, and bass that is full, fast and accurate. Just as important is the reliability that will ensure it will provide this performance for a lifetime.

Axiom DC300B Data and Specifications

The development of the Axiom is complete or almost. We have made significant modifications since the last update. The Axiom is now sporting more diodes. The most significant change was the addition of a diode in the anode circuit of the 300B. In addition the cathode resistor in the 6Y6 circuit has also been replaced by a diode. There are no resistors or capacitors in the audio circuit only wire and tubes.

Along with the on going "ear" testing we have just completed weeks of analysis using a HP FFT Dynamic Signal Analyzer. The test results confirm what the listening tests were telling us.

Total Harmonic Distortion

Before reporting on the Axiom let me give you the numbers from a well-regarded traditional design 300B amp. This amp was constructed using highly regarded caps, resistors, and transformers and has received rave reviews from Central Florida audio community. The Total Harmonic Distortion (THD) for the amp was measured at 1 kHz.

1 Watt = THD 1%

2 Watts = THD 2%

5 Watts = THD 5%

6 Watts = THD 200%

The Axiom measured using the same measurement techniques

1 Watt = THD 0.042%

2 Watts = THD 0.110%

5 Watts = THD 0.512%

6 Watts = THD 1%

7 Watts = THD 5%

The low distortion measurements are reflected in the many listening test evaluations, the Axiom Amp has great dynamics and sounds exceptionally "clean". The THD distortion numbers are those that would be expected of an amp with significant global feedback. The Axiom, of course, is single ended without global feedback.

Another important difference between the two amps is the shape of the distortion curves. The Axiom had virtually no harmonics beyond the third. The higher order harmonics were buried in the noise floor. Whereas, the traditional amp had harmonics that were distinct and measurable out to the ninth harmonic and the odd order harmonics did not roll off evenly.

Another important comparison is how the amps respond at higher outputs. Since the harmonic distortion of the traditional amp changes significantly as the power increases the apparent timbre of the instruments change as the level is increased. The Axiom, on the other hand, maintains low distortion numbers throughout the power spectrum; thus the instruments maintain the same quality throughout the dynamic range.

The large jump in distortion that occurred in the traditional amp was the result of clipping. The traditional design begins clipping at approximately 5.25 watts. The Axiom clips at approximately 6.25 watts. When the Axiom clips however the clipping is more benign. During listening test the Axiom amp was pushed to approximately 8 watts without any noticeable distortion at the peaks. The headroom of this design is noticeable when listening to uncompressed orchestral CDs. The amp effortlessly reproduces the sound of single violin through the sound of the full orchestra at a triple fortissimo.

Following are some additional test results that will help put the Axiom's design into perspective.

All of the diodes were removed and equivalent resistors substituted.

1 Watt - THD 0.557%

Compared to the traditional design the Axiom did well even without the diodes. The following tests were done sequentially with resistors being removed and diodes inserted.

The resistor in the cathode circuit of the 300B was replaced with a diode.

1 Watt - THD 0.123%

The resistor in the cathode of the 6J5 was replaced with a diode.

1 Watt - THD 0.079%

The diode was added to the 300B's anode circuit (with the previous diodes still in the circuit)

1 Watt - THD 0.051%

The resistor in the cathode circuit of the 6Y6 was replaced with a diode.

1 Watt - THD 0.042%

Blind listening tests for the various iterations were as follows:

Diode in the Cathode of the 300B

Significant improvement in presence and low-level detail

Diode in the Cathode of the 6J5

Better dynamics and a slight improvement in low level detail

Diode in the Anode of the 300B

Improved open-ness and slight improvement in low-level detail

Diode in the Cathode of the 6Y6

Slight improvement in low-level detail

The THD test was also performed at low frequencies (60 Hz and 400 Hz) and at high frequency (5 kHz and 10 kHz) Although there was a slight increase in THD it was always less than 1% at all frequencies. This also is very unusual for single ended amps.

Generally single ended amps have significant increases in THD at both the low and high frequencies. It is not unusual to find distortion on the order of 2% or more at 1 Watt. The fact that distortion and the distortion ratios remains relatively constant throughout the frequency spectrum of the Axiom ensures that the timbre of instruments do not change throughout the range of the instrument. In listening test, this improvement was particularly noted on solo piano. The piano didn't take on a harshness or edginess that is often noted in the upper octaves.

Bandwidth

The bandwidth of the traditional amp was found to be 12Hz | 57 kHz +/-3dB

The bandwidth of the Axiom Amp is 20 Hz to 90 kHz +/- 1dB the bandwidth performance is not reduced as the power output goes up. In fact it gets wider!

The Axiom produces virtually distortion free signals down to 20Hz whereas a traditional amp has significant distortion in this range. The result is that the bass produced by the Axiom is cleaner and tighterh and therefore sounds gbetterh than the traditional amp. Much the same can be said for the high end.

In part, the reason for the extremely low distortion numbers across the entire frequency spectrum is the use of plate load chokes.  It was noted during listening tests that a change in THD of less than 0.03% was perceived by the listeners. Since distortion is so critical to how we perceive music, maintaining low distortion across the bandwidth is critical.

We feel distortion ratio curves are as important as distortion levels. The Axiom has a steep downward quadratic curve. Third Order is about 1/3 of Second Order and Fourth Order is about 1/3 of Third Order, Fifth and Higher Orders are below the noise floor of the HP Spectrum Analyzer. The Axiom ampfs distortion ratios do not change as power output goes up. In fact the amp was quite listenable at power peaks of 13 Watts.   

We must caution you that how effective using a tube diode is in a circuit depends on matching the correct diode to a specific amplifier tube. A diode tube has both static and dynamic resistance, see diode tube curves. It turned out that the same diode turned out to be the best for the 6Y6 and the 300B tube. The 6Y6 has a mu of 6-9 and a plate resistance around 900 ohms. However that tube diode did not perform as well as a different diode on the 6J5.

We have lost count on the number of revisions made and, like a tube diode on the anodes of the 6Y6 and 300B, why the revisions made a difference. Is the Tube Diode acting as a Cascode circuit with the top tube having a gain of zero? In that case is the reduction of Miller Effect responsible for reduced distortion and improved sound performance. Or is the additional tube diodefs dynamic curve combining with the amplifying tubefs curve to glinearizeh the tube and reduce distortion for improved sound performance.

Once we did the math and established the original circuit, we only used listening evaluations, for parts selection and to optimize and refine the circuit. Once we had reached the circuit on the published schematic, a very powerful HP spectrum Analyzer was used to verify what our ears told us to do. I firmly believe if we used the test equipment after every revision, we would have never reached the performance level the amp has.

Bear in mind there are a lot of design elements that are responsible for the Axiomfs performance; no resistors or capacitors in the signal path, plate load chokes for loading the tubes and establish bias voltage for the following tubes, separate stacked power supplies with high Henry chokes. The use of hum bucking interaction in the separate power supplies to reduce hum.    

300B Output Transformer Comparison

Dr. Bob Hoekstra - 08/13/01 - Bob's system is a Sony DVP 7700 DVD front end and monoblock single ended 300B amplifiers driving a pair of Hedlund Horns with Lowther DX-3 drivers. The 300B amp is his own design called the Axiom(An Audio Alliance Project). His interconnects and cables are Litz braided OCCC wire purchased from Chimera Laboratories.

The transformers that were tested are the 15 watt 3.5K Tango Permalloy output transformer and the Chimera 5K-300B transformer. The 8 Ohm Lowther driver was connected to the 6 ohm tap on the Tango yielding approximately a 4.7K load on the 300B. The listening test was conducted after the Tango was "broken in for approximately 30 hours before beginning the evaluation.

A wide range of music was included in the test. The music included modern jazz, classical jazz, symphonic, string quartet, female vocalist, piano and pipe organ. The most noticeable difference between the transformers was the "edginess or glare from the Tango. The edginess caused listening fatigue in as little as 30 minutes. In addition, the edginess emphasized and colored the tonal quality of some instruments. The instruments most affected by the edginess were the saxophone and the piano. The glare was also apparent on some trumpet pieces, the Irish "tin whistle and the 1" flute stop on the organ. The Chimera transformer, on the other hand, had none of this glare or edginess. The instrumental pieces were better balanced; individual instruments were not over emphasized. There was no listening fatigue even after extended listening sessions.

Generally, the Tango is regarded as an output transformer with superior frequency response. It was expected that there would be an improvement in the high-end frequency response. The Chimera transformer performed as well as the Tango and perhaps better for low level high-end response. The sound of brushes on a cymbal sparkled when listening to the Chimera, as they did with the Tango. There was a sense that perhaps the high end definition might be slightly better on the Chimera, but this difference, if it is real, is certainly small and arguable.

The Tango is rated as having a much higher inductance transformer than the Chimera. I expected to hear a difference in the bass response. There was no perceptible difference. The bass was tight and well defined for both transformers. Kick drums were well defined with no "bloat". The 16' and 32' pedal stops on the organ were rich and full but maintained excellent separation between individual chords.

The Tango is an excellent transformer, but the better overall tonal balance of the Chimera and its lack of glare make it the better choice. The final design of the Axiom 300B Amplifier will specify the Chimera 5K transformer.