Watt Sucking Fireball Series

[Roadbagger]

Scott, will you be at NAMM this year? Would like to meet up.

[carpenter]

Whew... that was quite a description, Roadbagger.

[scott wurcer]

Originally Posted by Roadbagger

Scott, will you be at NAMM this year? Would like to meet up.

The chance is nil, but I do turn up in San Jose on occasion. I will send advanced notice next time.

As another funny coincidence, I live 2 blocks from the former Teledyne Crystalonics JFET fab on Sherman St.

[kenpeter]

Not sure the purpose of your resistors in series with the
caps that bridge your split tail pairs in the 1st schematic?
I assume the intent was folding a cascode both ways to
make a theoretically perfect differential?

See you like the Wilson Mirror! Ever tried Blumlein's Garter
in that application? This was kinda my thought process how
to autobias a pair of tubes accurately enough for a toroid
OPT... I don't know how well it'd work under a JFET input
circuit like yours.

Drawing #1 is the pair of CCS shown in Drawing #2.

[Roadbagger]

Originally Posted by kenpeter

Not sure the purpose of your resistors in series with the
caps that bridge your split tail pairs in the 1st schematic?
I assume the intent was folding a cascode both ways to
make a theoretically perfect differential?

The story behind this design was pure inspired motivation. When someone suggested in late Sept. that I should look at this thing called a Burning Amp Festival, I decided to suggest building an amp, using all Linear Systems parts to demonstrate how good they sound in the linear gain chain of an ultra linear amp, to Paul Norton at Linear. So this amp became a reality from a clean piece of paper and a promise, to demo in 3 1/2 weeks. It started working at a demo level at 6AM of the show day. There were a number of choices that had to be made with no turning back because there was no time to make a change if something went wrong with some new idea. This design used a number of tried and true concepts from past designs, though never tried together. the input differential needed to have the lowest noise pair of current sources musterable. Matching of the LIS dual bipolars allows me to make the Wilson based dual source with fractional percent accuracy without thinking and with only 2 parts aside from the low noise current feeder made from a selected LSK170. Early in this series I said I didn't include any cross coupled anything like a Barrie Gilbert type preamp, (another long discussion) Because I wanted to keep the illustration of circuit concepts as simple as possible. The split resistor on the driver stage eliminates a resistive load imbalance that would come from the resistor biasing the source coupling caps. The 150 ohm total is there to be the degeneration/linearization of that stage, as all stages are controlled gain linear stages. The object in this design is to have the amp as linear as possible without feedback so the small amount of feedback employed will straighten but not interfere with transient IM performance. So, yes, theoretically perfect differential, but with the ability to independently steer the output bias condition without messing with the AC differential gain.

"See you like the Wilson Mirror! Ever tried Blumlein's Garter
in that application? This was kinda my thought process how
to autobias a pair of tubes accurately enough for a toroid
OPT... I don't know how well it'd work under a JFET input
circuit like yours."

Auto-biasing is an art that I developed in high school for putting Marshall Amp heads in line using a couple transistors in a differential balancing integrator for the grids. That way I could regulate the quiescent and use a mismatched Tungsol and a Sylvania, or whatever, without an imbalance in magnetizing current. That was how I financed an equipped basement amp shop in HS. I would do the same thing for a toroid output application, that is, to sense the cathode currents with as small a resistor pair as possible and use a differential integrator driving a pair of current sources to steer the grid bias into compliance. Most musicians in High School were as poor as church mice and getting matched tubes to achieve full output power from a Plexi-head was out of the question. Someday someone will find one of those old heads and wonder about that "hack" under the chassis. I have a PCB kit that I made for Guitar amp balancing that I used to sell. No more complicated than the one from 1969 but better.

K-wood


Drawing #1 is the pair of CCS shown in Drawing #2.

[Roadbagger]

Here's the circuit example for concept

[Roadbagger]

The Real Goal

In the grand scheme of things we strive to record and play back for all time to recall, a sound, like the smell of something that brings back a memory, reproduced the way we remember, the moment. There are certain phenomenon in audio reproduction that, when eyes are closed, will reveal that what is heard is real or reproduced. Much of that “false” image in reproduction is an intermodulation, fabricating non-original artifacts. What would be great is a holographic reproduction system. One that flawlessly reconstructs the entire field of acoustic influence from all angles, delays, environmentally influenced spectral convolutions, and most of all, true fidelity from copy to projection. Bringing this discussion out of the ethereal and making the bold assumption that a Louther, Manger, NHT, Magnaplane, Tanoy or the like will truly move air the way that it should be, an amplifier working in harmony with that load should be able to drive an electrical current with enough voltage compliance to satisfy uniform consumption of that energy but to do it without accelerating the load beyond it’s compliance limits. A perfect amplifier would have zero intermodulation distortion, constant output impedance, not zero or infinite, and if driven to a point of creating harmonic distortion, that it be compliant to the ear’s own asymmetric compliance limits and therefore acoustically invisible. A speaker system can be as guilty of hidden distortions as anything in the chain. Often unnoticed are non uniform speaker input impedances causing spectral notches in energy transfer through to the air. A speaker crossover should not only equalize the transfer function between radiating elements but also present a constant load to the amplifier allowing linear power transfer through its entire range. In radio frequency applications we call this broadband impedance matching, often not a trivial task. A filter will pass what it will and the remaining energy must go somewhere other than back down the throat of the amplifier. Though not efficient, adjacent filter dummy loads and attenuators are employed to feed off excess energy not transferred to the air and equalize the impedances throughout. When not, amplifiers having bipolar follower output stages will reflect an image of the load backward through beta demagnification to the base drive and so on, this is a loop gain shifter and it does have an effect. Reverse transfer capacitance, GM shifting drain to source V/I swings, Beta shift with collector current, Source and emitter resistance changes with current, all contribute to a series of complex interactive quantities that interfere with perfect linear transfer gain. As Time allows I’d like to illustrate some simple concepts toward that goal. An effort to keep this discussion on a conceptual basis first is to hold the “Concept” as primary, details of implementation may then have a stable framework for construction. Years ago at KRL’s Vudu Lab we observed subjectively that a perfectly clean audio source could be heard through a mix of other sources far better than the same source through a dirtier amp. When a clean source is heard, the perception is unmuddled by artifacts and can be said to have a high signal to noise ratio. The same source through a dirtier amp needs to be turned up in order to have the same perception of the source material being heard. In a constant noise field the clean source will be more easily perceived because all of its structure is a fixed power above the noise where the more distorted amp contributes noise to the listening field and must be turned up to compensate. I think this may explain why really distorted music is often played louder than should be humanly tolerated to say nothing about the attendant fatigue, and clean sources tend to be more comfortably level adjusted. So, the cleaner the source, the lower the volume and hearing damage. Not that I have an opinion, but it seems to me that poor tradeoffs have been made over time in favor of convenience and we may be loosing the perceptual ability to discern what is real and what is not. More later, after you cut this prattle to shreds…………………………
K-wood

[Roadbagger]

Originally Posted by kenpeter

This was kinda my thought process how
to autobias a pair of tubes accurately enough for a toroid
OPT.....

Transformer outputs are a wail of tradeoffs. If you want a Single ended drive the leakage reactance will be high so as to accommodate DC bias plus the differential swing about that point. In the P-P amp leakage reactance is managed depending on how much the tube quiescent current can be tolerated. Sloppier builds with more open window area in the core means the sloppier the tube balance can be. But little can be done with toroids accept balance, balance, balance. Low leakage, good coupling and high permeability tend to saturate toroid cores with little imbalance. It's a tight path to acoustic coupling efficiency as long as DC balances are corralled. There is a toroid tape winding technique available from Stangennes Industries in Palo Alto that spaces the permeable steel a bit with a dielectric coating. This provides some leakage in the core and makes a more forgiving toroid transformer. General Radio built a number of toroid transformers for audio, the 942-A comes to mind. it was used in the Gotham recording amplifiers some time back and driven with 811's I think. They used a winding technique called astatic winding which tended to reduce the imbalance issue by increasing some of the leakage and reducing distributed capacitance by this winding and isolated primaries. The isolated primaries were useful, allowing them to operate in parallel in a totem pole output stage developed by Donald Sinclair in 1952. This lowered the input impedance by 4 and effective distributed capacitance consequently upping the frequency range and uniformity of coupling coefficient to the speaker. The Sinclair P-P output stage is what I use almost exclusively in all the amps I do weather they are OTL or not. But then balance is always a good thing to do any way for other subtle influences in symmetry.

Prattling on....................K-wood

[Roadbagger]

Looks like this

[Roadbagger]

This is an example of a Sinclair-Peterson type push pull totem pole amplifier from the Oct. 1951 GR Experimenter publication. It puts both output tubes across the voltage rail to ground, cancels the magnetizing current as in a normal push pull and couples both primaries in parallel capacitively so the sending impedance is 1650 ohms instead of the usual 6600 ohm P-P. Of note is the fact that both output tubes are driven grid to cathode and the gain transfer function of both are identical as in the normal P-P configuration. Some stacked tube amps are driven so as to have the lower tube with transfer gain and the upper tube used as a unity gain follower with upper-lower drive correction through feedback. The Sinclair-Peterson architecture is a symetric drive impedance particularly when using triodes and requires little feedback.