This post brings the input stage and the buffer together into a single channel amp.
You'll remember from the previous input stage post that the grid of the second triode is grounded. But this grid is an extraordinarily handy place to apply some negative feedback. We don't want a lot of feedback, but a little bit will be a good thing.
Here's the schematic:
The output is brought back to the second grid through a voltage divider (an extremely common and broadly applied technique). With the component values shown the NFB is only about 6db. This is not much NFB, but it has two really nice effects:
1. It extends the effective bandwidth of the amp, causing the high freq rolloff to be higher.
2. It almost takes control of the gain of the amp. Not quite completely because there is not much NFB, but mostly.
The second feature is important because you can roll different kinds of triodes into the front end, including triodes that have vastly different mu values. For example, the front end will support 12au7, 12at7, and 6922 (among others). The mu for these tubes is 17, 60, and 33 respectively. This means that the open loop gain of the amp will change dramatically when you swap these tubes in and out.
However, with the NFB the gain of the amp for these three different tubes is:
12au7 - 9
6922 - 10
12at7 - 13
Notice that the gain hardly changes. This means that when you roll these tubes you won't have widely gyrating volume pot issues. If you want slightly less gain, increase the value of R8 by some small amount.
So folks, there you have it. Assuming that the amp will work as well as it has in dBel84's prototype, this is about as flexible, low voltage, tube hybrid amp that you can make.
The power supply also has some upgraded features compared to the SOHA. The B+ supply uses a much better voltage doubler, for example. In fact, the SOHA PS will not work for this amp.
But now I think we wait for the prototype team to get their boards. If you're anxious to get things moving you can rag on JeffR at GlassJarAudio.