I changed the output stage from the "Devil begotten" ultra-linear configuration to pure pentode operation. (Stu Hegeman's expression).
I increased the power supply energy storage five-fold.
I changed the output tubes to the Tung-Sol KT120s.
I modified the screen - grid circuit with a current limiter.
I substantially changed the open - loop feedback compensation.
I changed the time constant of the output tube grid dive such that the volt-time product impressed on the control grid of the output stage is ALWAYS lower than the saturation limit associated with the output transformer.
I modified the feedback in order to increase the performance when driving a loudspeaker, especially when used with a high resolution loudspeaker. It was here I voiced it to yield a large soundstage, and with pinpoint imaging inside that soundstage.
Pure Pentode: The major problem with the "ultra - linear" screen connection is that it does not sound as good as pure pentode operation because when the output tube plate pulls the top end of the transformer to ground, it necessarily must simultaneously push the screen grid low. This extremely low momentary screen voltage causes the tube to become current starved in that moment for driving the speaker. Not good. It was Stu Hegeman (designer of the Citation series of amps and pre-amps) who explained this to me long ago. Ultra-linear became popular for a variety of technical reason; it was considered negative feedback (It's not) and reduced distortion for a resistive load. If it were feedback it would reduce the gain when applied, but it does not. It is simply a different output stage configuration. It avoided the complexity of including a second power supply for the screen voltage, and to do it right required the screen power supply to be regulated, an additional unwanted complexity. Ultra-linear works modestly and quite well when applied to small (6BQ5) amplifiers. But that's another story.
KT120 Output tubes: By using the new Tung-sol KT120 tubes instead of the 8417 tubes that the unit comes with, the power was increased to almost 80 watts rms per channel continuous. On speech and music, we find that the power becomes quite enormous - easily 200 watts rms. This update required a change in the bias circuitry to make it work. It was necessary to increase the charge time of the bias circuit, and simultaneously increase the discharge time such that when the amplifier is shut down, it does so without causing excess current to flow in the output tubes during the power supply discharge. Far better reliability in the long run.
High-frequency oscillation: It was necessary to add a "pole and zero" into the frequency compensation to make the amplifier super stable for the next 100 years. This consisted of a series RC network at the plate of the input pentode, the 7199.
Grid drive volt-time product: This was easy to do, all that was needed was to quantify the grid drive as a function of time in order to determine where the transformer core saturated. In order to prevent saturation, all I needed was a correctly sized grid drive capacitor with a time-constant well inside the output transformer core time constant for saturation. A walk in the park!
Feedback: I changed the feedback in order to allow the amp to easily sound good when driving difficult loads, as well as normal speaker loads. This was done by adding several resistors and capacitors into the forward path as well as the path from output to input. This allows the amp to listen to the room, as I have done in several of my other amps. By doing so, I have the room engaged in the presentation in a way no other amp can, allowing all the beauty to be captured.
Power supply energy storage: The power supply capacitors were changed from 100 uf to about 490 uf. This change eliminated "gulp" distortion, an insidious distortion in which the power supply is modulated by the output signal, resulting in a signal related bounce in the output of the amp. It is completely absent now!
High voltage drive to the output grids: In order to drive the grids of the KT120s, it was necessary to increase the B+ voltage to the 7199 driver tube. This change allows the voltage drive for the grids to be just the right value to allow full power operation with room to spare.
Majestic Output Transformer: Now for the best part: If you look at the pictures of my prototype Silver Seven amplifier in the assembled photos, you will see that it sports two huge Dyna A451 transformers for outputs. It took two of them because the power level was 350 watts. This very transformer is the unit used in this Dyna amp. The only difference is that it is not potted with a nice case. It only has a simple end bell, but inside it is one of the best transformers there is!
The final design: The amplifier has become a world class vacuum tube amplifier, capable of standing with the very best units there are. It is far advanced compared to the way it was when I started working on it, and I hope you enjoy its sound. I believe it will probably last another 50 years. If it ever stops working, all you have to do is get it to me and I will fix it free. Even the output tubes are guaranteed for life. It's my life, and I intend to live a long time!
Sold for $5655.55 ...