Low voltage virtual ground

[garyr]

I need a half-rail virtual ground in a 3.3 volt circuit. I couldn't find
anything like a TLE2426 that would work at that voltage so I thought I would
use a low-voltage opamp configured as a follower with VCC/2 connected
to the positive input. The TI LMV721 has a supply voltage range of 2.2 to
5.5 volts and a GBW of 10 MHz. That would be probably be acceptable
but I would like to extend the range of low impedance output to as high a
frequency as possible and so perhaps a cap on the output would be helpful.
The data sheet implies that the '721 is stable with 2.1 nF on it's output
with a 2.2 V supply and 21 nF at 5 volts, but doesn't state a maximum
value of capacitive load. Linear extrapolation would imply that 9.5 nF
would be OK at 3.3 volts. Any idea wha the actual maximum value
might be?

Also, my circuit will not impose a DC load on the virtual ground; should
there some DC load on the opamp?

 

[Robert Baer]

I need a half-rail virtual ground in a 3.3 volt circuit. I couldn't find
anything like a TLE2426 that would work at that voltage so I thought I would
use a low-voltage opamp configured as a follower with VCC/2 connected
to the positive input. The TI LMV721 has a supply voltage range of 2.2 to
5.5 volts and a GBW of 10 MHz. That would be probably be acceptable
but I would like to extend the range of low impedance output to as high a
frequency as possible and so perhaps a cap on the output would be helpful.
The data sheet implies that the '721 is stable with 2.1 nF on it's output
with a 2.2 V supply and 21 nF at 5 volts, but doesn't state a maximum
value of capacitive load. Linear extrapolation would imply that 9.5 nF
would be OK at 3.3 volts. Any idea wha the actual maximum value
might be?

Also, my circuit will not impose a DC load on the virtual ground; should
there some DC load on the opamp?

Well, i would put a resistor in series from opamp output to "load"
/virtual ground and a capacitor from there to ground for the HF bypass
needed.

 

[whit3rd]

If there's no DC load, why not two resistors and a big cap?

or two resistors and two capacitors (if the power-on/off transient behavior matters).