This is what happens when layman specify what they want in a tube amp. They drive the designer crazy.
"An architect's dream is an engineer's nightmare."
LOL!
I'll say that some of the hurdles that popped up during this build were a learning experience, but will be a big contribution to another amp going into production very soon that will be available to purchase by anyone... unlike the Solar Flare that is unfortunately a one off...
One of the most original amplifiers I ever saw. Incredible. But it has manual fixed bias right? Otherwise you would be able to use 12A*7 and 6922 in the same sockets but with an adapter. Both are noval but bias differently. Right?
12AU7, 6922 and 6J5 are all quite close to each other electrically and could comfortable share the same operating point from my understanding even with resistor bias. I'm very interested in the output stage implementation myself. There are two different filament voltages, three different plate voltages and four different bias current requirements here. I thought what Mischa did with my amp, having two bias modes that changed both voltage and current (it's far more common for amps to only change the current and keep the voltage fixed when they have a feature like this), but this is another level of capability here altogether. Especially since we're dealing with DHT and the headaches they bring.
Do not worry, just tell your story. For output stage is clear with 300B but for input, seeing you have 6 sockets I was thinking "why?". Because you could have 3 and adapters and just adjust the bias for the tube type. This is how I do, I have one double triode socket and 2 single triode sockets wired for 6SN7 and 6J5. With adjustable bias I can put whatever there as long as I have an adapter towards 6SN7 or 6J5. Most probably you have an input selector and select 6J5 / 12AU7 / 6922 and do not care about the bias. But I was also wrong saying that "both are novals" because they are but have different pinout so you cannot put them in the same noval socket, you need adapter. Anyway another thing which bothers me is how you use 2 x dual triodes (6922 and 12AU7) or 2 x single triode (6J5) and you do not need a quad of these.
Everything in the amp is auto-biased/servo-biased. The driver tube stage uses a fixed voltage on the cathode (proprietary circuit), but allows the anode voltage to vary based on the tube that is in. For CCS, a constant ~3mA is run through the driver tube. For anode resistor driver tube bias, the current running through the driver tube, and the voltage on the anode, varies based on the setting that is chosen for the power tube, as the power tube setting changes the B+ voltage throughout the whole amp. As
@Xcalibur255 correctly mentioned here, the 12AU7, 6922, and 6J5 actually bias up quite closely together with either the CCS or the resistor setting. I believe I originally proposed doing just 6J5 sockets, and building adapters for the other tubes, but dedicated sockets for each driver tube type were chosen instead. The 12AU7 and 6922 sockets are run such that the dual sections in the 12AU7 or 6922 are run in parallel - this drops the tube white noise even more for these tubes. No need to select which driver tube is in use and no manual fixed bias for the driver tubes - that all gets set automatically.
The output power tube biasing is something new that I invented myself. Historically, many amps that offered multiple different tube choices would either only allow you to vary the B+ voltage (often by adding/subtracting dropping resistors in the B+ supplies - these would cause a spongy B+ supply and flabby sounding amps). Some amps that did multiple power tubes would instead do cathode bias using cathode resistor + capacitor, and the tube type chosen would vary the cathode resistor value to set the current through the tube. Unfortunately these amps usually chose a B+ voltage low enough to work with the lower voltage tube, and the performance/output power of the other power tube(s) would be compromised by the lower voltage.
The Solar Flare does things differently: In order to to not cripple the performance of power tubes when a bunch of different power tube choices are available like in an amp like this, you need output transformers that are optimized for the wide range of tubes, then the tube choice on the front panel has to dynamically set the B+ voltage, bias current, and heater voltage based on the chosen tube. The optimal amp would also avoid resistor+capacitor power tube cathode biasing (this is the historically defined "auto-biasing" for a power tube stage). The Solar Flare uses a different form of auto-biasing that eliminates the resistor/capacitor in the cathode - instead it uses a new form of servo-biasing (confusingly also referred to as a form of "auto-biasing") for setting the cathode current.
DHT's are also fickle and really require low noise DC voltage on their heater for there to be no hum (or ultra-high frequency AC heaters). One traditional way to do DC heater supplies is using simple voltage regulators - but those are believed/known to degrade the sound quality on DHT tubes in comparison to AC heating as voltage regulated DC heaters "fight" against the audio signal that can impose itself on the heater filament in DHT tubes. There have been a couple solutions to this over the years, including having DC voltage heater supplies that have big chokes in the line (chokes allow the DC voltage on the heaters to swing up and down somewhat in tune with the music), or using constant current fed heaters (e.g. Rod Coleman filament regulators). The Solar Flare implements a heater supply that I had to invent as well that allows you to set the heater voltage to 5V (300B and 71A) or 2.5V (2A3 and 45), but follows that with a "sloppy" supply that acts moreso like a constant current supply and gives the known sonic benefits of that.
Truly incredible feat of engineering
@A2029. I can’t even imagine how much work went into the planning and execution of this amp.
Congrats
@Zachik - you have years (decades?) of experimenting ahead of you! Top notch amp reveal as well. What a captivating prelude -- I look forward to the many chapters ahead in your impressions on trying out all these different modes and tube types. It will take some serious scientific rigor to design, conduct, and keep notes on comparing different modes across different tube types. I feel like this amp should come with a laboratory notebook In addition to the spiral binder operational manual!
I’m intrigued by the 1 stage / 2 stage output and SET/Parafeed switches. What are the pros/cons of each?
1-stage/2-stage:
When running 1 stage, pro is that it bypasses the driver tube and connects your DAC almost directly to the power tubes. This gives more signal purity, and less tube distortion/noise as there is no contribution of distortion/noise from driver tubes. Con is that a DAC can only supply so much voltage, so the power output from the 1-stage mode is limited by the max voltage that the DAC can supply. E.g. the 1-stage is good for purity of sound into headphones, but cuts the power of the amp down so much that it doesn't work well with speakers. For 2 stage, pro is more power as you get the driver tube stage to amplify the signal before driving the power tube, but with the con of the increased distortion/noise that the driver tube stage adds.
SET/Parafeed:
It gets a bit complex to explain the difference between parafeed and SET output topologies, but the following links do a pretty good job. See below the links for tl;dr.
http://tubelab.com/articles/ideas/single-ended-output-stages/
https://www.jacmusic.com/techcorner/ARTICLES/English/Parafeed/Index-Parafeed.html
https://www.tubecad.com/2014/09/blog0308.htm
The tl;dr is that both can use the same output transformer (though different output transformers can be specified for each to optimize the performance of each topology), but the parafeed couples the tube to the output transformer using a big capacitor and ideally feeds the power tube with a gyrator. Doing so linearizes the power tube more, dropping distortion. Also, because the transformer no longer has any DC current flowing through it, its distortion drops somewhat as well. A parafeed specific output transformer can also be made smaller, and with higher quality materials for cheaper due to the smaller size as there is no DC flowing through the transformer that could cause it to saturate like in a traditional SET topology. Negative is that output power drops for the parafeed mode in an amp that has both (and the difficulty of building both into an amp is high). There is also the distortion from the parafeed capacitor that is added (though this should be less than the extra transformer distortion in a SET amp topology). So with parafeed you have less transformer & tube distortion, more capacitor distortion, but overall less total distortion out and less power out.