A lot of commercially available and pro design voltage regs for audio have an error amplifier and then deal with the noise and ripple, whereas DIYers tend to want to have successive layers of noise reduction. Is one approach just wrong or too expensive for commercial application?
Are the diyers getting away with adding impedance to the PSU by buying big transformers in the first place?
Are the diyers getting away with adding impedance to the PSU by buying big transformers in the first place?
A string of filters is not a voltage regulator, the supply droops quite considerably with load, it is not difficult to make an error amplifier have supply impedance less than one ohm with line regulation in the mV range, a filter will not have these characteristics.
Filters and voltage regulators are complimentary, a filter gets the ripple down to manageable level and removes the 10kHz + noise that error amplifiers have trouble with then a error amplifier regulator remove the low frequency which includes line transients and sags
Filters and voltage regulators are complimentary, a filter gets the ripple down to manageable level and removes the 10kHz + noise that error amplifiers have trouble with then a error amplifier regulator remove the low frequency which includes line transients and sags
Best combo IMO is a hybrid of both Vreg + filter. Works great for low-noise SMPS or regulator.
First have your Vreg with error amp with capacitor filter while sensing voltage at that point, then feed that regulated DC through a series inductor and another bank of capacitors.
To simplify I mean: Voltage Regulation in 1st stage, then filter output of that with inductor + capacitors. - - Vreg+C+I+C
The Inductor has to have just milliohms of resistance, so it should have near ZERO DC voltage drop to not affect voltage regulation, but has enough inductance to properly filter out high frequency ripple that makes it past the first Vreg stage.
First have your Vreg with error amp with capacitor filter while sensing voltage at that point, then feed that regulated DC through a series inductor and another bank of capacitors.
To simplify I mean: Voltage Regulation in 1st stage, then filter output of that with inductor + capacitors. - - Vreg+C+I+C
The Inductor has to have just milliohms of resistance, so it should have near ZERO DC voltage drop to not affect voltage regulation, but has enough inductance to properly filter out high frequency ripple that makes it past the first Vreg stage.
Last edited:
Huge cap banks makes the life for the reg much harder. You get lots of harmonic sprays with a large cap bank, due to the very narrow and very large charging current spikes. This is a case where you can have too much of a good thing.
If you use a reg, you get cleaner power with moderate cap bank, just enough to get the ripple below the reg drop out. Then it has the least high freq harmonics and the ripple is no match for the reg anyway.
It pays to know what you're doing (literally and figuratively
).
jan
If you use a reg, you get cleaner power with moderate cap bank, just enough to get the ripple below the reg drop out. Then it has the least high freq harmonics and the ripple is no match for the reg anyway.
It pays to know what you're doing (literally and figuratively
).jan
The charging current spikes from large capacitors is not as extreme as some might imagine. If the cap bank is fed from a low impedance source then some series inductance is called for, this provides 3dBV octave rolloff in the harmonics, the capacitor bank also provides 3dBV rolloff on the capacitor side.
Doubling the capacitor bank size halves the corner frequency at which this filter starts to act.
This is a trivial circuit to simulate, the simulation will give good results if the full equivalent circuit for a transformer is used and the supply impedance is included.
Use ideal circuits and the current spikes will be far more severe than reality
Doubling the capacitor bank size halves the corner frequency at which this filter starts to act.
This is a trivial circuit to simulate, the simulation will give good results if the full equivalent circuit for a transformer is used and the supply impedance is included.
Use ideal circuits and the current spikes will be far more severe than reality
Just to be clear here... [addressed to all, but also jan.didden]
The purpose of the "R" in an "RC" filter stage (just taking one section) ... is to lower the frequency response of the section. In the case of using ever larger primary reservoir capacitors ... having a resistor R in series from rectifier-assembly to reservoir serves substantially to limit charging current, thus stretching out in time the charging "spikes", which in turn spectrally acts as a low-pass filter.
"R" - resistors - are the poor man's inductor in this case. Much better (though more tricky to deal with at design-time) are inductors in series. These ... in the A/C domain, act as variable-impedance devices (like frequency-dependent-resistors, sort of) ... but don't have anywhere near the voltage-drop of resistors (and power waste, heating) for the DC that flows through them. The only drawback is that they cost like 20× what the substitute resistor might cost. Bummer.
Finally ... when it comes to the actual regulation of the power output of the power supply, it is absolutely true that either LC:LC:LC or RC:RC:RC does no regulation, but just filtering. The regulator circuit is in charge of keeping the voltage at power-supply output as near constant as is practical. And the it is the designer's job in turn to ensure that the audio-path has a high PSRR (power supply ripple rejection) ratio, as well as symmetric or 'constant power draw' amp sections.
Can't talk about one without imputing that the other is also near-magical in its fight to "cure" the "problem".
GoatGuy
The purpose of the "R" in an "RC" filter stage (just taking one section) ... is to lower the frequency response of the section. In the case of using ever larger primary reservoir capacitors ... having a resistor R in series from rectifier-assembly to reservoir serves substantially to limit charging current, thus stretching out in time the charging "spikes", which in turn spectrally acts as a low-pass filter.
"R" - resistors - are the poor man's inductor in this case. Much better (though more tricky to deal with at design-time) are inductors in series. These ... in the A/C domain, act as variable-impedance devices (like frequency-dependent-resistors, sort of) ... but don't have anywhere near the voltage-drop of resistors (and power waste, heating) for the DC that flows through them. The only drawback is that they cost like 20× what the substitute resistor might cost. Bummer.
Finally ... when it comes to the actual regulation of the power output of the power supply, it is absolutely true that either LC:LC:LC or RC:RC:RC does no regulation, but just filtering. The regulator circuit is in charge of keeping the voltage at power-supply output as near constant as is practical. And the it is the designer's job in turn to ensure that the audio-path has a high PSRR (power supply ripple rejection) ratio, as well as symmetric or 'constant power draw' amp sections.
Can't talk about one without imputing that the other is also near-magical in its fight to "cure" the "problem".
GoatGuy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.