Preamp section - schematic review, revisions, etc.

[jordane]

Good afternoon folks,
I have an Epiphone Valve Jr. chassis I am converting --- using a different preamp topology.  I designed this by (heavily) borrowing ideas from different amps.  Being new to this, I am hoping some experienced builders can give me a once-over and suggestions before I order parts and try this.

Attached is a hand-drawn schematic.  I've made it as neat as I can; hopefully, legible.

Input is pulled from Marshall 18W type.  Attached is the input from the 18W "Modern Classic" for reference.  Some of the values are changed, but I tried to stay true to the basic idea.

I lifted the "FAT" section from the "Tweed OD Special," but implemented in a different way.  I'm not 100% sure it will work.  The 12AU7 cascade is pulled from one of the tube data sheets (attached as 12AU7 cascade example).

The main question here is, will the FAT section work as drawn?  I think I need mixing resistors as marked at the FAT part of the circuit.  If so, what values?

Secondary question: can NFB be incorporated?  Or can I worry about that after I have the amp up and running?

[pdf64]

The last stage is missing a grid leak resistor.
I think the shared cathode resistor is supposed to be 18k, rather than 1k8.
What makes the low input lower than the high input?

[sluckey]

What makes the low input lower than the high input?

The low input only feeds one grid. The hi input feeds both grids.

[PRR]

The "cascade" with common cathode WILL motorboat unless the cap is very large relative to the interstage coupling bass cut-off. 5uFd is far less than the 50uFd in the factory hint-sheet. Even 50uFd may decay over the years and give you an unexpected thu-thu-thump in 2032. In this case: don't be cheap, use two cathode R/C networks.

[jordane]

The last stage is missing a grid leak resistor... What makes the low input lower than the high input?

It is missing a grid leak resistor, yes.  I will add a 680k as suggested in the datasheet.  Thanks. 
Sluckey is correct that the low input feeds one grid, while hi feeds both in parallel.

The "cascade" with common cathode WILL motorboat unless the cap is very large relative to the interstage coupling bass cut-off. 5uFd is far less than the 50uFd in the factory hint-sheet. Even 50uFd may decay over the years and give you an unexpected thu-thu-thump in 2032. In this case: don't be cheap, use two cathode R/C networks.

Excellent tip.  I was wondering why the data sheet wrote in 50uF, that seemed high to me.  I've got more reading to do.  I will use two cathode R/C sets. 
Question: am I at risk of the same motorboat on the "parallel" (first gain) section with the small cathode bypass cap?  I have those cathodes tied together as well but it would be no problem to run two separate cathode R/C networks.

[PRR]

> parallel" (first gain) section

Not a cascade.

[jordane]

> parallel" (first gain) section

Not a cascade.

Got it, so parallel acts differently in this regard.  Thank you.