Vc Canceling Vacuum Tube Amplifier Topology




Free Counter


May be Last Significant Development of Vacuum Tube Amplifier Topology

All of the Following is Hereby Declared to be in the Public Domain. Therefore, the Concepts presented Herein Cannot Be Patented by Outside Parties or by Me.

Due to the Vulnerability of a small patentee, a patent can be proven invalid at his expense, or a slight change in concept can lead to a new Patent by a hostile business competitor. It is my desire that others pick up the ball on this technology, and that the technology described herein be developed in the Public Domain and hopefully continued instead of lost.

Once I started referring to these as the "god-amps" to Dave. He became deeply disturbed at this and wanted to be sure to give the glories or credit of any accomplishments or discoveries to God, not to his credit.

Dave C. - 2000

I am the third child of Dave's.

Early Dave Years - Shrader Sound in the 1940s in DC. Quest for a superior "D.C. Amp" up through the 1990s, using predominately 6L6G push-pull output configurations.

Discovery of an inscrutable feedback effect within a SET amplifier led to years of research, group listening, and pondering as to what might be happening to cause the increase in the clarity of sound, especially around the upper midrange, including the building of a number of SET prototypes utilizing different output tube types, output transformers characteristics, etc.

6AH6 -> 812a Amplifiers = 10 Watts/Channel

Using dual 812a transmitting tubes at the requisite low 5-15 Ma plate current gave us the same audio quality at about 10 watts rms per channel, the same audio quality that we were getting out of the 1 watt per channel flea power type amps.

AWG 30 Solid Wire wrap Hand-Wound Secondary

The red winding is a hand-added secondary winding made from #30 wire wrap wire and threaded between the outer laminations and the primary winding.





Push-Pull Amplifier? I believe the two transformers were each a half of a complete push-pull output transformer.

Another view of the above amplifier.

I think the empty tube socket at center was actually just an adapter to convert a dual-tube socket into two single ones.

This is a 6G6 kit PCB we were trying to build, before addition of tubes, output transformers, and a voltage multiplier B+ power supply on the rear, along with a small filament transformer.

This is the bottom side of above.

In the prototypes, we simply bolted on the output transformers to the underside like support feet, and attached a simple AC/DC power supply to the back, with an external 12.6 vac filament transformer, feeding the 6G6Gs in series.









These photos depict a prototype amplifier that was essentially built on erector set parts for a chassis. I can get you the tube complement for this and the others, if you want.

This is a completed 6G6G stereo amp prototype on the kit PCB, though with a lot of tweaks for negative feedback. Also, the two potentiometers were added on the underside for input level control.

This is the very original of the "Golden Amps", originally a mono amp embedded in an old convertible LP console/radio system.

Dave, 2000

The couch on his right is the optimum listening position, or sweet spot. This was a glassed in porch, about 50/50 glass and brick, a pretty hard and reflective listening space.

Dave was educated at American University in Georgetown, Washington, D.C. and obtained a degree in Physics, although his love from childhood was electronics. At that point in history, electronics was not well enough developed that there was a degree available in it. Much of his chops he learned from self-education during a long childhood sickness and his job with Schrader Sound, where he repaired and modified, in some cases experimentally, the different hi-end audio rigs of the day, as well as radios. He tirelessly built amplifier prototypes in his basement, starting from 6L6 push-pull models, to building these mostly triode-type SET amps, until he was physically unable to do it anymore because of not being able to get down the stairs to his workbench due to problems with his knees.

From the piles of paperwork left around after his death, I was able to retrieve spiral-bound notebooks with his empirical observations and theoretical speculations about what was going on with these SET amplifiers that gave them that particular quality of clarity across the listening sound spectrum. His education and lack of thorough mathematical training left him with mostly a lot of guesses, but still he continued on tirelessly building and modifying different circuit patterns of this particular amp, noting all of his empirical observations in notebooks over the years, at my bequest.

We had numerous personal conversations as well as over the phone, but I knew I wouldn't be able to remember it all, so I requested that he take notes on his observations.

Back in the 1990's, Dave was still trying to perfect a "D.C. Amp", an amplifier that would be a stellar achievement in the field of hi-fidelity audio.

Something he always told me was that:

"Everybody understands audio amplifiers nowadays, Everybody understands speakers, and Everybody understands transformers. But No One understands how they interact to ultimately form the sound that you actually hear!"

One of his favorite quotes was the notation on the bust of Thomas Edison. This bust is located at the main entrance to the Washington, D.C., U.S. Naval Research Laboratory complex where he worked most of his career. It had the emblem:

"There's a way to do it better - find it!"

In the 1990s, he developed an appreciation for the SET type amplifier, or single-ended class A triode circuit, usually without the indirectly heated cathode in the output stage(the filament was the cathode). He also developed an appreciation for the KISS (Keep It Simple, Stupid) principle, keeping his amp designs down to two active devices per channel, to minimize the mutilation of the pure signal by passing through too many stages. One of his flea power amps used a 12AX7 to drive a 6G6G tube, for an output of about 1 watt or slightly less per channel. Before he died, he was working a PCB kit for a stereo version of this amp, but never quite sealed the design before his death. I think that his kit price target was to be under $100. The point was to sell a kit that people could buy cheaply and build for themselves. The big problem was getting the output transformers that will work with this.


We were using a batch of Berco make transformers that we found from the 1930s. The 15 M on the box meant 15,000 ohms in the 1930s.


In some cases we used interstage audio transformer primaries with a hand added secondary made of #30 gauge wire-wrap wire.

In any case the output transformers had to have a gap or two (E-I pattern laminations) in the iron lamination paths since we are talking Single-Ended-Triode and DC currents through the transformer primary winding. Without gaps in the laminations, the DC effects would override everything, as well as the laminations getting permanently magnetized. As it is, any DC through the windings will greatly knock down the value of primary inductance, another reason for keeping the quiescent plate current to low levels. Using an incremental inductance measuring Wheatstone bridge called a "Hay Bridge", it could be seen that a 15 or 20 Henry inductance primary would effectively be reduced to a few Henries of inductance when measured while a few mA of direct-current was passing through it. Its ironic that such a cheap transformer would work so in this feedback scheme, whereas the modern day Hammond designed-for-SET 125CSE model double-gapped, high iron content units were too low of a primary impedance and did not work at all with this. The primary impedance overall has to be high enough to retain the back- reflected signal from the speaker and pass that signal successfully into the vacuum tube to be mixed with the original driving signal in large enough proportions to be effective. It varies between triode-tetrode-pentode tubes and screen resistors connections and values, but the appropriate median value may be approximately a load impedance equal to the tubes effective rP.

My Early Years - My exposure to audio started with a 9 transistor radio that someone gave me. I used it with an ear bug or hid it under the covers so the parents wouldn't hear it. My first stereo was a 12BH7 amplifier formerly mounted in a record-playing console that was later converted to a stereo rig, playing through 10 watt rated Radio Shack Minimus 0.5 speakers. It later became the first prototype of this group of SET amplifiers. Later I moved to Fisher type integrated amps. I used a PAS-4 preamp as well as building the "Distortion-less Preamp" from the "Popular Electronics" magazine article of a few decades ago. There was another all-FET preamp built from a kit that I borrowed. During the 1970s and 1980s I lent my ears to every fad and fashion of those time periods. Many of my friends were competitive in building high-fidelity amp and speaker systems using vintage electronics, tubes, FETs, BJTs and novel speaker physical positioning phasing concepts and crossover designs, both passive and active, as well as motional feedback used with the woofer.

If it hadn't been for me pestering Dave, he wouldn't have kept any notebooks on this. Also, he wanted me to co-write an article about this in a magazine, but he was never sure what was really going on with it, and he died in 2004 before we could do anything on it, or figure out what was going on with it. So this is my contribution to it that all of the prototype building and all of the empirical research he did on this might not be lost. I've studied these prototype amps and his notes for four years now, and I believe that I understand the crux of the mechanism that is feeding back.

There were a dozen or half a dozen people that were making evaluations on each change that he would implement, or each new prototype that he would build using various triodes, pentodes, pentodes wired as triodes, etc.

For a long time, I was one of the half-dozen people that could hear what he was talking about, and was part of a team of listeners/evaluators to help provide an objective opinion to each tweak in design, or each new circuit topology. My college curriculum was heavily weighted toward calculi and math and I tended to want to analyze things in terms of comprehensive, exhaustive, end-to-end types of analyses. Since Dave did not have the exhaustive mathematical background that I did, he was often stuck with using simple electronic rules-of-thumb to follow his ideas. Most of his discoveries and innovations were through empirical means, he just kept building these things experimentally and through empirical observations.

When I finally got my hands on the control panel, so to speak, and we found all of his notebooks that we could, I collected all of the relevant prototypes that I could find from his basement. I then read all of his notebooks in a short enough of a time span that I could hold all of the information in my short-term memory long enough for some major points and correlations to filter out.

Dave built a number of prototypes, using typically a 12AX7 for the input voltage amplifier. For the output tube, a number of different ones were tried, in various operating modes of triode, pentode, and everything between. Tubes tried included the 12BH7A, 6G6G, 47, 2A3, 812A, 6L6 and 25L6, chiefly. We also had a single plate 2A3, a very rare tube indeed.

I came to the conclusion that through a type of feedback, that his amps were canceling out all back-emf of each entire speaker assembly and box, but without changing the slow damping or ringing of same, which occurred at the combined low, maybe 100 hz resonant frequency of the woofer and the closed speaker cabinet combination. Both mechanical and electrical inductive and capacitances of the speaker assemblies were being cancelled out at the higher frequencies or at least suppressed to a good degree excepting that this canceling effect was not so intense as to be harsh and detrimental to the sound quality.

This was done through the reflecting of the back-emf signal to the driving element (output tube) in a relatively strong enough proportion and inverted polarity to the original signal in order to de-emphasize the back-emf during its occurrence, regardless of polarity, capacitive or inductive. So while the plate circuit of the amp is the source for the speaker, likewise the distorted reactive signal that is created in the speaker becomes an inverted source that is reflected back into the plate circuit of the tube to mix with the original, in the correct magnitude to effectively mix with the original signal as a source of antiwise feedback.

To work, this required the correct proportions of idling plate current, and the correct ratios of rP, output transformer primary load, the transformer primary being a combination of high dc resistance and inductance as it turned out, and the correct turns ratio from primary to secondary (could be in the 30:1 to 56:1 ranges, typically, depending on vacuum tube impedances and output transformer characteristics) to make this work.

Looking at the vP versus iP plate circuit lissajous phase pattern, the primary V/I phase tends to become resistive (zero) around 2000 - 3000 hz, and mildly reactive close to these frequencies. This argues for a pretty flat phase shift difference between the signal going out to the speaker, and the correction coming back into the tube's plate circuit around these frequencies, which are the upper vocal ranges, for which the clarity effect is greatly pronounced, such as with young female solo vocalists.

The Ip of the output tube needed to be 5-15 ma for its ac operating range to be weak enough to be affected by this weak feedback signal from the speaker reflected up through the output transformer. This feedback serves to clean up the higher frequency "junk in the trunk" type of clutter that pollutes the middle frequency female vocal range, around 2000hz to 3000hz.

The fact that most tube amp designs are going for high plate current and low impedance is why you never see this fine-tuned feedback effect taking place in conventional tube amp plate circuits.

However, the limitation in plate quiescent current put a great limitation on the effective output power of the amplifier. That is why we had to use an 812a transmitting tubes with 750 volts on the plates and a 5-15 ma Ip to achieve a 10 watt per channel audio output, over the 1 watt per channel outputs obtained with other tubes. The 5-15 ma quiescent current had to be maintained for the plate current to be weak enough for the feedback from the speaker to have a negative contribution to it.

By the way, this speaker reactance canceling effect can also be made to work with a push-pull output topology, if the operating factors are scaled appropriately.

All of the waveforms that follow feature a pulse of 1 millisecond at about a 30 hz repetition rate fed into a Scott 299c (top trace) and a 12AX7 -> 6G6G "D.C. Amp" (bottom trace).

For several of these, a Fisher 500c was substituted for the Scott 299c, with very similar results to the Scott 299c integrated amplifier. The Scott and the Fisher are similar integrated amplifiers, with an FM/AM tuner built into the Fisher amp.

The first set of output scope traces are looking at the voltage across a 1 ohm resistor in series of each of two speakers, speakers which were used in the original setup in the original listening room in Washington, D.C. Therefore they are showing an approximate analog of the current going into the overall speakers. Remember, current is what creates the magnetic field, not strictly voltage, and current is what creates the force that moves the speaker coils, not strictly voltage, but ultimately current, so it is important to watch.

1 msec/division.

500 usec/division time base.

200 usec/division

200 usec/division.

Left channel miking of 6 inch woofer, close and halfway into the paper cone

Right channel miking of woofer cone, again with a Neumann TLM-103 condenser large diaphragm type microphone

2 msec/division - microphoned outputs of the Scott amplifier on the top trace, and the 6G6 amp on the lower trace

1 msec/division, miked

500 usec/division, miked

200 usec/division, miked

So What's So Great About It?

Well, let me write my own subjective opinion. He operated the 6G6 amps in a glassed in brick porch that was acoustically hard, about 50% brick and 50% glass. See my profile picture and imagine a listening spot centered 90 degrees to my right. The speakers were two 5 ½ inch paper cone speakers, some kind of Radio Shack model, with 1 ½ inch paper cone tweeters. The tweeters were coupled in with a simple series resistor-capacitor network, a 2.7 uF/50vdc non-polarized in series with a 3 ohm resistor, while the 5 ½ inch woofer was directly driven and had a frequency response that petered out on its own. Pretty cheesy setup, really, on the speaker end of things. But that was part of the goal, is that you wouldn't have to go out and buy expensive speakers to make this thing go. The speakers were usually positioned to radiate from behind us, from a parallel plane with our backs.


He was always tweaking the amount and path of what tiny amount of negative feedback he used. He experimented with feedback around the whole amp as well as around one stage only. Too much and the sound sounded squashed, and trapped within the speaker box, too little and distortion started creeping in.


On one day I was back visiting from Blacksburg, VA, and I found him in the listening room, (see my profile picture) rapt with attention. As he directed me to sit in the "sweet spot", I was taken with the crystalline quality of the singer's realism and clarity. If I closed my eyes, I could "see" Linda Rondstadt (Lush Life) up on the stage, with a crystalline positional image that did not smear, drift in position or fade in and out. Also, these aural images were not locked in the speaker cabinets. They appeared to float in midair. This is probably one of the most significant qualities of this sound.

These are the things audiophiles only have wet dreams of, having been one myself. That's one reason I'm writing this, that it may help someone else. I have only a few of the original Berco high impedance transformers left. If we could find some transformer winders to duplicate the characteristics, then we could produce more of these amplifiers, especially the 812a versions, as they have transformers with some different, beefed-up characteristics.



In so many systems the designers worship low THD numbers and end up applying so much negative feedback around the loop that they squash the sound and trap it and the image in the speakers.



In addition, I could "see" the Nelson Riddle Orchestra, below the stage in the orchestra pit. The midrange of the singer and the brass instruments were especially real clear, and had solid positional images. I've been a convert to these amplifiers that he made, since I've been listening to and evaluating mine and others stereo systems since the 1960s. I've been through all the fads and styles of the 1970s, 1980s, 1990s, including but not limited to, wooden monorail speakers, linear tone arms, motional speaker feedback, physically positioned phased speaker arrays, tubes, transistors, FETs, MOSFETs, all kinds of exotic crossover networks, bi-amping, tri-amping, extremely exotic stylus and cartridges, digital devices with extreme over sampling, on and on.




I've found that the enhancement of these amplifiers works best when used with recording sources that were recorded with simple left and right positioned microphones, and the signals then run straight through the left and right of a mixing board without much intermingling and mixing and mashing the left and right signals together with overmuch processing. Also a lack of left-to-right relative phase shift in the overall source is good.

From using a lissajous scope setup between right and left channel source inputs, I find for instance, that some of the early Pentangle (a Keltic folk group), has recordings that seem to meet this requirement. If a scope lissajous pattern between left and right source signals stays closed, a diagonal line, but changes in length with the music, that's good. If the pattern stays open as an oval or a circle a lot of the time, or continually opens and closes, you have phase shift or phase drift between the two channels that will mar your image positionally or make it drift in position with time.