The Tau - essential SET amplifier development

A fine example of a single ended triode (SET) amplifier was required to demonstrate the abilities of our speakers and so developed elements from decades of amplifier building were put together. The concept was then a distillation of learned essentials which were actually not at all conventional but rather imperially discovered though trial and error. These essentials were not an economisation but a focus toward the purpose of fidelity. To go where anaemic circuits with a 300B stuck on the end simply cannot. Conventional electronic design is motivated by finance and this quest was for music - damn the cost.

Once a tube spelunker pokes around enough, chasing the darkness of convention and reviews, which are essentially paid advertisements, one discovers that power sucks. Power sucks the nuance out of playback, yea and transmitter tube amps are dogs with high miller capacitance, huge transformer losses and unnecessary gain stages to drive them. High efficiency speakers and low power amplifiers can do what nothing else does - portray low level detail and the gestalt* of a musical event, to make music something special to hear and enjoy. Euphony. It takes courage to look away from the shadows of the cave wall, what commercial interests have spent decades and millions convincing us of, to emerge into the sunlit terrain. I am grateful I was brought up with live music - the bullet proof, bullshit detector* - which forced me to transcend electrolytic capacitors, untreated ground paths and useless watts. 

The 4P1L replaced the 45 and 46 tubes which got me into SETs decades ago. I intuitively bypassed the 2A3 and 300B. After competing with trophy collectors for tubes it was fun to find and use a tube that was equivalent of the 45/46 in linearity, low level detail and ambient aura that was overlooked, abundant and inexpensive. It is a direct heated Wehrmacht tube, a Russian copy of the RL2 / 4P6, on which there is no information in the German lexicons - they are listed there but with no information. Top Secret? I was curious how it would do in parallel single ended triode operation having used it for years in a Sakuma stye amp - single ended configuration. The 2P29L was chosen as initial drive tube keeping the eventual, almost century old, 226 globe tubes in reserve until after the amp was stable. This tube is also a direct heated pentode clone, CO-257, direct heated pentode, also in triode mode.

So what does this design do different? All DHT, low power = small transformers, two stages = low losses. An over abundance of inductance, some of which is in the ground leg a la CMC. Attenuators are not at the input but used later to increase current drive and slew rate. No cables between preamp and power section to drive which reduce transformer and cable losses in the critical section between stages. Schottky, choke loaded supplies for Rod Coleman's excellent cathode regulators. These were tried with traditional EI AC transformers before finding toroidal transformers sounded better with filament bias. There is overall a plethora of iron - more than any theorist/engineer would endorse. That said the rise time is measurable and is faster than convention endorses. 

This highlights the key element of testing with ears and trusting what they indicate. Music is a series of dynamics, shock waves, over a broad spectrum of frequencies and cannot be dictated by frequency tones or square waves. There was also no politics or economy in parts selection: if it works - its stays, if it don't - it goes, not matter the expense or theoretical/emotional attachment... Measurements bow to the human ear which is what music is for and from.

Many of these concepts were derived from the trifecta of collaboration over decades between Peter Fowler, Bill Jaensch and myself: we three geezers.

On to the build:

Parts collection initiated.

 

Shuffling parts around for layout design. As one can see here the henries of the PSU, Ale's SiC bias boards and Dave's most excellent attenuators.

The beginning of tie down.

Wiring initiated with ground bus.

Provisional power supply...

Clip leads are fun! Here beginning with Ale's gyrator plate load. 

Then on to amorphous core choke plate loads from NP ... and the next iteration of the PSU.

The beauty of breadboards is that a new device or circuit alteration can be changed in minutes.

I became rather fond of the little filament PSU sculptures almost naming them after movie robots... before going to choke loading for the filament supplies. 

This provisional supply used a high voltage Schottky bridge for the B+ which eventually became a mercury vapour rectifier tube half wave. The damper diode is in the ground leg replaces a SS slow start PCB - not only to provide slow ramp up of the B+ but it does more as per JC Morrison's inspirations.

After almost a year on the breadboard with many, many iterations this was the final configuration of the circuit that became preferred. All six supplies are now fully choke loaded with toroids and R cores and the B+ went to mercury vapour rectifier. Nothing, not even TV damper diodes come close to what a MVR sounds like. Yes, I have heard the nonsense that one cannot hear the rectifier in a properly designed PSU. That tidbit of wisdom is for people who do not have good ears IMHO. Hybrid rectification need not apply as they sound predominately like its weaker links. 

The driver section "evolved" to conventional interstage. The "natural motion" of a transformer plate load - "reactor" - as the old text books describe it - sounds the most natural to my ears all theorising aside. Also plastic PP caps and electrolytic have begun to be been minimised and the favourite GE97F series oil caps beginning to populate filtration.

Dennis Boyle's transformers (Ace Amplifer) and interstage are used up to this point with

M-6 Orthosil@ 29 gauge .014” thick laminations. He tested cores of Orthosil@ Super Orthosil@, Permasil@, 40% and 60% Nickel in .009”, 0.10” .014” and .185” lamination thicknesses in both power and audio signal transformers. His findings were that the M-6 .014” EI cores offered the lowest distortion with a wide bandwidth. This was measured and they sound surprisingly good.

That said, Finemet and Dave Slagle's 49% Nickel come in later and clearly do some things that M6 does not. More to follow as we discover some nice things about nickel that the ear and nervous system can easily and pleasantly discern. 

A few more posts on this amplifier will follow - including pictures with less spaghetti...

 * gestalt -

1. A physical, biological, psychological, or symbolic configuration or pattern of elements so unified as a whole that its properties cannot be derived from a simple summation of its parts.

1. A collection of physical, biological, psychological or symbolicentities that creates a unifiedconcept, configuration or pattern which is greater than the sum of its parts (of a character, personality, or being)

1. a shape, a form