BJT overdrive gain problem

[TejfolvonDanone]

Hi everybody!

I've designed overdrive: https://drive.google.com/file/d/0BzGRfrmHSjb4dWhfdVNGNDU0Z2M/view?usp=sharing
I have a basic problem with my design:Bypassing more of the emitter resistors of the second stage decreases the overall gain.

I've figured out that the first problem is because if i decrease the emitter degeneration the input impedance of the second stage goes down and loads down the first stage resulting in a gain loss rather than a increasing it.

Are there any easy fixes for the first problem?

[thermionix]

Where is T3?

[TejfolvonDanone]

T3 was an input buffer  (emitter follower) i got rid of but will put it back.

[slacker]

Maybe buffer the ouptut of the first stage or you could try a series resistor between the stages, that might reduce the loading enough.

[TejfolvonDanone]

My ultimate goal was to be able to create a variable gain stage out of the second transistor. The series resistor would also reduce the gain. I'm thinking about a stereo pot with one side instead of the emitter resistors and the other side  in parallel with the series resistor to have a somewhat constant input impedance but increasing gain with the rotation of the pot. I'll crunch the numbers later.
The buffer stage looks fine but it would increase the partcount quite a bit and i'm not really sure i want that.

[TejfolvonDanone]

If anyone interested my solution was to bypass some of the resistor of the first stage's output resistor with a cap and putting a pot in the second stage's emitter.

[TejfolvonDanone]

I'd like to share the final design which works pretty well in simulations. Breadboarding is still nessessary.
https://drive.google.com/file/d/0BzGRfrmHSjb4WVlnaWZ1WGhMXzg/view?usp=sharing
The R1 R5 and R6 with the Gain knob sets the bais point and i didn't want to go lower than that. I also decreased the collector resistors of T2 (also with the bypass cap C8). The Gain2 knob is now in a range where the gain increasing effect of the rotation compansates for the input impedance lowering resulting in an overall gain rather than a damping.
I'd like to make this design with a germanium transistor at T3 so R1 should be a trimpot with 2K resistance to be able to set the bias point properly (around half supply).
As always feel free to share any comments.