Simplified virtual ground concept

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi! Been looking through the virtual ground concepts for creating a virtual pos/neg supply to opamps for headphones amplifiers that has been posted at HeadFi and at Tangentsoft's homepage. But, if the input voltage is known and stable, why not simply put a voltage regulator that regulates at half that voltage? 5V supply here because of designing for USB voltage, but the concept can be used at any voltage levels, with an adjustable regulator. Any reason this wouldn't work?

10185277.png
 
One reason it might not work well is a virtual GND needs to sink as well as source current. You could fix that by putting a current source on the 2.5V output to GND. Choose the current according to what the circuit is likely to sink into the 2.5V output, then add a bit more.
 
A very good point! The virtual ground need to SINK as well as source. This is of course the reason to the double regulators in the designs here. There will of course be an AC signal that both pulls and pushes on the virtual ground.

Please disregard my suggestion and consider it something for the trash can! :)
 
Basically you can do this - it's probably not the lowest-noise virtual ground in existence though. Even that LT1763 up there, with a 10 nF bypass capacitor, would still exhibit wideband noise levels of 20-30 nV/sqrt(Hz) under light load, which is not exactly stellar by audio standards. For comparison:
A classic TLE2426 rail splitter with 1 µF bypass cap - about 10 nV/sqrt(Hz), which is OK-ish.
A voltage divider with opamp - 1...10 nV/sqrt(Hz), depending on type.
A voltage divider with buffer - 1...3 nV/sqrt(Hz) or even lower, depending on transistors and current.
A plain voltage divider - almost arbitrary low depending on how big a filter capacitor you install.

Noise would not be a problem if all the inputs and outputs were referred to virtual ground. This only works, however, if either power ground or input signal ground is floating. Signal ground being shorted to power ground (e.g. USB power + signal input from same PC) or held at any other fixed potential could have disastrous consequences or would keep the circuit from working at the very least.

EDIT: Oh, too slow again. Being able to sink current as well as source is another very good point, of course.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.