Tube DIY Asylum: 6N1P versus 6SL7 (a bandwidth versus distortion question) by mikeyb

I am using a 'long tailed pair' phase splitter diectly coupled to a 6SN7 differential amplifier. This is a variant of Kevin Kennedy's circuit (see link below).

Life is full of trade offs - with a 6SL7 driving the 6SN7 and the gain reduced to about 25 due to an unbypassed cathode resistor (3000 Ohms between cathodes), the theorectical bandwidth is 90 Khz. Bandwidth is reduced by the rather high plate resistance of the 6SL7 (44K ohms), the fact that a high valued, unbypassed cathode resistor is being used, and the 'Miller effect' input capacitance of the 6SN7 stage. It measures around 65 Khz. However, the linearity is improved due to the 'local feedback'. The cathode resistors in the 6SN7 differential amplifier (8600 Ohms between cathodes)are also unpypassed, reducing the gain of this stage to around 6.

Here is the question. Is a bandwidth of 65 Khz adequate? I can improve bandwidth sustantialy to about 350 Khz by using a 6N1P in place of the 6SL7. The 6N1P has a lower mu (33) than does the 6SL7 (70), and a plate resistane of only 4400 ohms. To achieve about the same stage gain, I would nee to reduce the resistor between the cathodes to about 940 Ohms (versus 3000 Ohms with the 6SL7). This will increase distortion since there is less 'local feedback' (or degeneration, if you prefer that term).

How does the 6N1P 'sound' versus a 6SL7? Is it better to increase the bandwidth and accept higher distortion?

Thanks in advance for the advice.

Kevin Kennedy's circuit: http://www.svetlana.com/docs/TechBulletins/technoteNo57.html