Single Supply LM3886 - better virtual ground

[turantual]

Hi, my first post on this forum!

I'm designing an amp based on the LM3886 using a single 75V supply. The supply is a given. I don't have the option of using a dual supply.

I'm a little skeptical about the virtual ground and biasing scheme given in the application notes. I worry that it will introduce nonlinearities. In the attached image I show National's method on the left. On the right I've added an emitter resistor Re and moved the non-inverting input bias resistor Ra from the voltage divider over to Q1's output. Is this better? My reasoning is sort of fuzzy, but I think Re should be chosen such that the quiescent current flowing through Q1 is much greater than any current sourced or sunk by pin 7. Also I think it would be better if Ra is referenced to that low impedance virtual ground. Does any of this make sense?

Thanks!

[KSTR]

If Q1 can handle the increased current and dissipation, this in fact is a good way to do it. We always have to keep in mind that datasheet schematics are mostly only conceptual for sake of clarity and no explicit fine tuning is shown. I'd start out to try the emitter R and then check noise levels at the "GND" pin and the original node to where Ra was attached, and then decide from that which one is better (and confirm it by listening).

- Klaus

[AndrewT]

I would not couple the ground pin to the input pin.
There are reports on this Forum that this ground pin has dirty signals on it.