I haven't done much with a single supply, the job I'm working on has some restrictions.

The circuit is actually powered from an AC supply, not batteries, so the voltage regulation is not perfect. The main positive supply is an LM317. The output is 10V, the voltage regulation is probably about 4mV or so. This is fine to power all op amps and other IC's, transistors.

I'm not sure if it's feasible to split this 10V supply to derive a 5V bias voltage to use as the ac ground. I'm not sure how clean this virtual ground needs to be. The circuit is all pre-amp stuff - with signals in the range of 20mV to 1V.

If my virtual ground has 4mV of ripple in it isn't that going to effect my circuit by causing a lot of distortion?

or is it not giong to be audible?

Would I get a better reference by using a 5V reference diode like an LM336Z-5.0 (since it's dereived from ground and not the shaky positive supply)?

 

You don't really need to be too picky about it. For example, I'm designing a circuit that uses Op Amps and a 9V battery, which I'll swap with a wall wart when/if I get it working.

http://forum.allaboutcircuits.com/showpost.php?p=63653&postcount=3

With your typical transistor circuit the voltage divider is already included. Here is a link to the AAC eBook...

http://www.allaboutcircuits.com/vol_3/chpt_4/8.html

If you get a chance read this book, while its not finished the parts that are are pretty good.

 

if you really want a virtual ground, you could use an lm675 power op amp. output is 3a. look at the datasheet on national semiconductor's website.

 

The important thing is to avoid a common return path for the grounding on the input side and the output (power) side of your amp.
Charging/discharging the blocking capacitor will generate significant voltage drops along its ground return.

 

If you use two equal-value resistors to divide the supply and produce a new "ground" with 2mV of ripple then connect a capacitor to ground on it to reduce the ripple down to nearly zero.
The opamps have plenty of power supply noise rejection anyway.

You will need an active "ground" if you want the opamp to have a response including DC.