A couple of years ago I've built this small DC-powered tube amplifier. It's based on the "Silverface" Fender Champion AA764.
Disclaimer
While this project is definitely accessible to most hobbyists and there's no mains voltage involved, the high DC voltages present must still be treated with the due fearful respect. I'm in no way responsible for damage to people or property caused by reproductions. Honestly, I don't suggest you to build a tube amplifier unless nothing else does it for you or you just find it fun. I've used both solid state and tube amplifiers of all sizes and while I hold this amplifier dear, I don't think tube amplifiers any better or worse than some other amplifiers. Also, the vast majority of tube amplifiers out there follow closely the topology and values of two or three classical types, so unless you're looking for something original, the amp you want is probably available used for cheap somewhere.
Powering the circuit
DC-powered? You heard that right. This thing is powered from any laptop brick and uses two switching converters to bring that input down to 6.3V for the heaters and 250V for the B+, not too great of a challenge for low power amps. DC-powered mini tube amps are nothing new, I'm a fan of the work on TENTEC and then there's of course Thomas Hafemann. I think they are a valid alternative to using a mains transformer, especially when I look at amp kits for beginners, both since you don't have to deal with mains voltage inside your project (although there's still the somewhat dangerous high voltage from the boost converter) and because you save the cost and space of the power supply and related rectification and filtering (more on this later).
Finding suitable DC converter boards for a low power project isn't too hard or expensive, and it just so happens that my wish was for a low power amplifier anyway.
My choice was a tiny MP1584 board for the buck converter, working comfortably with the heater currents involved, and the ubiquitous 8-32V to ±45V-390V, UC3843-based boost converter, colloquially called "the yellow boi". Again this one is working within safe limits, it's cheap and small, but isn't the cleanest of the lot, and this led me to some struggles down the line.
The laptop adapter can be of any voltage within the range of both converters. For these two something between 12 and 20V is probably ideal. The power draw, from some back of the paper calculations, is about 12W for the boost converter and 5.6W for the buck, so you want your brick to provide more than 20W to leave some room for efficiency and not run it at its maximum.
Output power
5W are suitable for home, recording and live use, where mic-ing the speaker is a requirement anyway if we want to be serious. This leaves out the specific case of band practice, when you have to deal with a drummer but don't want to deal with microphones, but I'm sure you can find an alternative.
Finally, low power amps allow you to get non-MV distortion while wearing only earplugs instead of firing range protection (I have sensitive ears).
About the claim of small tube amps being less clean or having less headroom, I think that's only if you want to. If we swap the unclear term "headroom" for "input sensitivity", which I prefer, which is the minimum input necessary for full drive in this case, we have clipping for any input larger than that, and a clean signal with progressively lower volume under it. Of course this is a tube circuit, with soft clipping, nonlinear single ended stages and tiny amounts of feedback, so there's quite the grey area in between, but for the sake of reasoning we can say we want to avoid most traces of the coveted tube clipping and stay well below that, as it would happen in a large amp driven well below its maximum.
So what's this input sensitivity? I went through the trouble of calculating an approximate value, considering the gain of all three stages and their headroom and output swing, but the good news is that we don't need to! We can make this value as big as we want with the "Volume" attenuator, after the first stage. And that first stage is the same in almost any tube amplifier, big or small, golden or silver, for good reasons pointed out in "Designing Tube Preamps for Guitar and Bass". Conclusion: small amplifiers don't have lower maximum input sensitivity than larger ones, they're just quieter at the same setting. Maybe most of the associated attributes are due to the speakers they came packaged with, which I have to admit, make quite the difference.
The circuit
Like I said, the circuit is based on the AA764 Champion. I chose this particular one because it's a two-bottle single-ended 5W amplifier, which means the circuit is trivially simple, with no phase inverter and a single output tube and that's the main reason for it. This is just like older Fender amplifiers under the name "Princeton", like the 5F2-A, while the Silverface Princeton was bumped up to a push-pull output 10W amplifier. Unlike the 5F2-A, this one has a two knob FMV tone stack with the "Mids" control fixed at full. I like the sound resulting from this type of filtering, just the same used in big Fender amps but with one knob less, even if it's somewhat lossy and limits the overdrive capabilities of the amplifier. If one wishes, a "Mid" control can easily be added in place of R11, or a switch to lift the ground and "defeat" the tone stack for the loud and raw kind of sound.
What I did instead was adding a bright switch (don't judge me, I'm a fan of bright guitar sounds). This is another feature found in many bigger amps but missing in small ones. I tweaked the resistor and capacitor values to something that was noticeable but not unpleasant.
Another difference from the original is that this amplifier uses the 6N2P and 6P1P tubes, which at the time of the build were easy to find for cheap. The 6N2P is a dual triode, basically equivalent to the 12AX7 in specs, but with no heater center tap, providing instead a shield pin between the two sections. The 6P1P is a tetrode equivalent to the 6AQ5, again with 6.3V heater. I've decided to run the whole circuit at 250V, instead of the higher voltages often found in commercial amplifiers. 250V is often quoted as a safe, standard voltage in many tube datasheets, as is in this case, and is still plenty high to not run into issues. The difference for the preamp is insubstantial, providing center bias is kept as in the original. For the 6P1P, I have drawn some load lines and came up with values that gave me a good bias with the given transformer ratio and supply voltage. You can find the measurements in the schematic, which agree with the calculated values. The feedback is shown taken from the 4 ohm tap, as in the original. If you only have a 8 ohm tap, you should double R16.
I've added the R17 screen stopper for peace of mind. Old amplifiers don't feature it, since they weren't meant to be overdriven (we've seen how that ended), but it doesn't affect normal operation and reduces the risk of damage to the screen grid.
The master volume is a recent addition, which replaced a 220k fixed resistor in the same position. In practice, I've actually used a B2M pot I had laying around tapered to that value. I can say that the distortion with the master volume down is tamer than the one with it up, in both cases with volume at full, which means grid current limiting on the output stage has a significant contribution. Other that not being painfully loud, I like the different character of the triode clipping. The amp by itself remains quite smooth this way, subtly distorting, but you just need a little bit of boost for the extra gain to tip it over the edge.
This is a single-ended amplifier, which means you need a suitable transformer. I was able to find one for cheap from aliexpress. The impedance ratio is 5k:8, which is suitable for this output stage, and features also a 4 ohm tap.
Power supply
The bottom left features a MOSFET "gyrator" and two RC filters for the B+ supply. In general, DC from switching converters is fairly clean and the high-frequency switching noise requires little effort to filter compared to mains ripple. The reason this much filtering was then necessary is because the "yellow boi", while featuring a supersonic switching frequency, sometimes shows low frequency bursts which appear as low frequency noise. I'm still not sure about the cause of this behavior or how to prevent it. Usually this happens after about half hour of use, before which the amp is super quiet, without the trace of hum often audible in most amps (also thanks to the DC heaters). After this, there's this faint annoying warbling noise. I can say a different DC converter with similar specs is probably a better choice. That said, the filtering suppressed most of it. The two RC filters feed the sensitive but low-current preamp and the screen. Since the noise doesn't go away with the preamp muted, I also added the gyrator to efficiently filter the whole amplifier. The mosfet in question is just an high voltage part I had desoldered from some power supply, you can use anything with at least 300V Vds. A capacitance multiplier with something like a MPSA42 is a valid alternative to the passive RC.
The build
Here are some pictures of the build. It's housed in a polished 1590DD enclosure, with a homemade front plate. The cage protecting the tubes is made from a steel sheet cut and bent with limited equipment, which I later painted black to prevent rusting. Later additions are the gyrator board, with a cute heatsink made from the same aluminum sheet as the front panel, and the master volume knob on top, switching to brass standoffs mounted to the bottom lid.
