>The LM-317 & LM-337 become inductive at rather High frequencies tend to Be unstable and do inject lots of Noise.
But the plot above is of just such a regulated supply. Noise figures are very, very low, and flat across the entire audio band!
LM317's when suitably decoupled are not remotely unstable. Many low-dropout regulators can display instability if careful consideration of decoupling is not performed, but LM317 / 337 and all other standard dropout NPN regulators are very stable, owing to the dominant pole compensation scheme used.
Many people do not like using such regulators, since they intuitively feel they canot perform well, and assume them to be noisy, but this is very far from the truth. The system used to measure the above noise plot is capable of resolving -140dB and my experience is that a battery based power supply, when under load, is no quieter.
The idea of using a virtual ground is OK, but don't forget one now needs to consider the dynamic stability of this amplifier across the entire audio band, not forgetting (assuming it to be op-amp based ?) that it's performance in most areas will be degrading in a similar way to that you propose for the voltage regulators. Any noise generated by this virtual ground will be induced into your amplifier circuit.
With regard to sinking and sourcing current, the regulator will not be required to sink current, it is only sourcing it (current flow is always from regulator to circuit), and any noise injected back to the supply will be decoupled very effectively to ground, if suitable decoupling is used. I cannot think of a single circumstance where a low impedance actual ground will under-perform when compared to a virtual ground. The impedance of a passive ground will always be lower than one using active circuitry.
Andy.
But the plot above is of just such a regulated supply. Noise figures are very, very low, and flat across the entire audio band!
LM317's when suitably decoupled are not remotely unstable. Many low-dropout regulators can display instability if careful consideration of decoupling is not performed, but LM317 / 337 and all other standard dropout NPN regulators are very stable, owing to the dominant pole compensation scheme used.
Many people do not like using such regulators, since they intuitively feel they canot perform well, and assume them to be noisy, but this is very far from the truth. The system used to measure the above noise plot is capable of resolving -140dB and my experience is that a battery based power supply, when under load, is no quieter.
The idea of using a virtual ground is OK, but don't forget one now needs to consider the dynamic stability of this amplifier across the entire audio band, not forgetting (assuming it to be op-amp based ?) that it's performance in most areas will be degrading in a similar way to that you propose for the voltage regulators. Any noise generated by this virtual ground will be induced into your amplifier circuit.
With regard to sinking and sourcing current, the regulator will not be required to sink current, it is only sourcing it (current flow is always from regulator to circuit), and any noise injected back to the supply will be decoupled very effectively to ground, if suitable decoupling is used. I cannot think of a single circumstance where a low impedance actual ground will under-perform when compared to a virtual ground. The impedance of a passive ground will always be lower than one using active circuitry.
Andy.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.