drive an 833A SET with either an headphone amplifier or a 45 or 2a3 SET.

[DrVincelette]

I have successfully built and perfected an 833A pair of monoblocks and I am satisfied I have perfected it after more than two years experimentation. WAVAC made an 833A stereo they sold for $350,000 and it is nothing more than the 833A triodes driven by 300B SETs. My 833A final stage is driven by a 45 SET with a Lundahl LL2765 transformer taking the 45 through the 5K winding configuration stepping it down to 600 Ohms. The 600 Ohms is connected between the 833A grid and the same ground the directly heated cathode is connected to. At 1000 plate Volts and zero grid bias the 833A draws about 120 ma which does not strain this tube enough to turn a spot on the plate red-hot the way I saw in an AM radio station that was using this tube which is made for AM radio station transmitters. For the output of the 833A I use a Hammond 1642SE transformer which is rated for 300ma before core saturation. This is bigger than any other transformer you can buy for a big SET. The 1000 Volt power supply I use 2300 Volt rated polypropylene filter capacitors with a 10 Henry choke between them with the choke on the negative sides of the capacitors so they are connected to ground rather than the dangerous 1000 Volts. I put fast reacting solid state diodes together in parallel for surge current to the first capacitor when the plate Voltage is first turned on. While I had to build an elaborate pre-charging 10 Volt cathode heater for the 10 Amps the 833A cathode requires and 4.5 Farad final filter capacitors, I recommend using seated lawn mower batteries with 0.3 Ohm dropping resistors for simplicity. Tungsten cathodes with the least ripple in the DC they require will make the 120 Hz buzz in the speakers.
The cost of parts to build this final stage is well under $2000; the Hammond output transformers are the most expensive at just under $300 each, the Lundahl transformers cost under $200 each, you can get new 833A vacuum tubes on ebay for about$200 each and the 1000 Volt power supply capacitors, Hammond 1500 Volt center tap transformers and Hammond 10 Henry chokes from Mouser electronics. A Bottlehead headphone amplifier should drive it if you use a 600 Ohm output impedance direct to the 833A grid and you won't need the Lundahl transformers or you could use a Bottlehead amplifier with the Lundahl replacing the output transformer.
I have been playing with the possible idea of putting together kits for an 833A output stage of amplification like my prototype or building them for a modest price of my labor. My background for this is over 20 years building SET amplifiers using 300Bs 2A3s and 245s. I have done avionics work as an FAA licensed aircraft mechanic and I got my PhD in interdisciplinary applied mathematics and mathematical physics with emphasis on physics from Delaware State University class of 2009. I taught as an adjunct and visiting professor but the cost of commuting and apartment rent took my take-home pay so low that I quit. I start working as a janitor which pays better than university teaching. In short, I have the kind of experience to be able to make nice 833A booster amplifiers for a minuscule fraction of what they sell for anywhere else they have been built and sold. The 833A I have found perfect for Magnepan speakers, more than loud enough if I turn up the volume. One final note, I tried driving the 833A grid with parallel feed but it sounded watery and weak even if it was turned up loud. An air gaped transformer sounds much richer and full-bodied driving this tube and there is neither a parallel feed transformer and choke big enough for the output of an 833A.

[Paul Birkeland]

There are definitely bigger SE outputs than the 1642, but they are definitely custom order stuff.  I have some Monoliths on the shelf that await duty in either an 833 amp or a 250TH amp whenever there is time...

If you want simple for 833 filaments, a Meanwell LRS-200-12 turned all the way down works nicely.  You also get a nice soft start out of these modules.

Our headphone amps are parallel feed, so the grid current drawn by the 833 will saturate the output transformers and it won't work all that well unfortunately.  Our headphone amps also don't have 600 ohm taps on the output transformers since that's practically useless for headphone duty.  There are plenty of line transformers that have 600 ohm windings but they won't handle the voltage, then there are those weird headphone amp transformers that have 600 ohm windings for some reason, which I'm guessing is what you're using.

2W into 600 ohms from a 45 is 34V RMS, or about 47V peak.  It looks from the curves like being able to swing +/-55V peak would get max out an 833 with a 5K OT running at that operating point, but you're still making over 50W with what you've got so it seems hard to complain.

What you haven't addressed is that the plate impedance of the 845 will be quite high and the winding resistance of the 1642 is also going to add to that, which will lead to a very low damping factor (looks like maybe 1.25) and undesirable levels of distortion.  When you make a higher powered amp, people will pair it with less sensitive speakers that are designed for amps with higher damping, so you'd want to shoot for a DF of 4 or more if at all possible.  In order to do this, you need to use feedback of some sort or a much higher impedance transformer, and this is one of the more difficult aspects of designing amps like these.  Unless you can get what you need with cathode feedback, you'll find that your interstage transformer and output transformer together present far too much phase shift to make an amp with global feedback, so you'll have to get creative to make that work.  I have found that even a good series feed output transformer on its own can make an amp like this conditionally stable, and you don't want to explain to your 833 amp customers why their ESLs melted!

Lastly, you could certainly use parallel feed iron for an 833.  That was actually part of the point of parallel feed iron in the first place!

[DrVincelette]

I tried parallel feed using Lundahl but it sounded thin and had less body than the Lundahl LL2765.

[Tubegopher]

Very interesting post Dr. I marvel when people like yourself and Paul Birkeland can hold a conversation on such a high technical level. Much of it is over my head... but I do marvel! The kit idea sounds interesting! 

[Paul Birkeland]

I tried parallel feed using Lundahl but it sounded thin and had less body than the Lundahl LL2765.

For interstage driving the 833 or on the output side of the 833?

A parallel feed transformer won't work for driving the grid of a class A2 power triode.  That's not a deficiency of parallel feed...