(Moderator's note: This thread was created from a post in the Finished Projects Forum, http://forum.allaboutcircuits.com/threads/project-cmv2a-virtual-ground-circuit.86442/)

Hello CumeSoftware,

I want to use your rail splitter to power a OPA548 as Audio Power Amp from a 12V battery in a bridge mode configuration as per the following file, what is your recommendation ?
Optionnaly I would like to use the Amp as Howland current source, what do you think ?



Best regards,
Jean

 

The two 10K resistors on the right hand side already form a rail splitter. However, the 0.1 uF capacitor is woefully inadequate. More capacitors are needed in parallel, the values depend on the max and min frequencies being amplified.

ak

 

However, the 0.1 uF capacitor is woefully inadequate.

?? I don't understand why. The time constant of 0.1µF into the op-amp's high impedance should be quite long, and there's almost no noise on that rail splitter.

 

Actually, the circuit will perform better if you leave out the subject capacitor...
That way the PSRR (power supply rejection ratio) intrinsic in the opamps can work its magic...

 

When creating a centerpoint ground out of two resistors, usually that point should have a zero AC impedance to both power rails for all frequencies of interest. In this case the corner frequency is 318 Hz. I think a non-decoupled medium-impedance point that is driving two inputs that are 180 deg out of phase is asking for trouble.

Also, there are two common ways to configure two independent amplifiers as a single bridge-tied-load (BTL) circuit, and the schematic in post #1 is neither of them. Usually, either the input to the right amp is taken from the left amp output (rather than its inverting, virtual-ground input), or both amps are driven directly from the circuit input, with one amp configured as non-inverting and the other configured as inverting. Has this circuit been tried, and where did the schematic come from?

ak

 

The virtual ground was created from battery, so if using two 100 uf~220 uF/10V to stabilize the voltage for +V, Gnd, -V is better then just used 0.1 uf for noise, if the 9V is come from adaptor then 0.1 uf maybe necessary and need two 0.1 uf.

 

Actually, the circuit will perform better if you leave out the subject capacitor...
That way the PSRR (power supply rejection ratio) intrinsic in the opamps can work its magic...

The circuit is DC coupled. Leave out the capacitor(s); it will just unbalance the circuit and possibly inject common-mode noise...
Note that the gain to V(x) is 2 and to V(y) is -2, so a total gain of 4.



The freq response is (of course) a function of the GBW of the amps.

 

You don't want to remove the decoupling cap, you want to add some more.
0.1μF ceramic caps should be placed from both the positive terminals and the negative terminals of the op amps to the virtual ground.
That's standard decoupling practice.