Is there a sonic advantage to using a regulated power supply for a solid-state power amplifier? Let's say, a "typical" power amplifier of 100 watts (≤ 0.1% THD) driving an 8 Ω load, with a supply of ~+/-45 volts filtered by 10,000µF capacitors.
Less audio power for your dollar? Absolute clipping instead of allowing a few short peaks far above the steady-power rating?
Seriously: if you find a marvelous amplifier with crap PSRR, regulation is an obvious path.
Seriously: if you find a marvelous amplifier with crap PSRR, regulation is an obvious path.
And even then, boosting the rails for the diff pair and VAS (and regulation) can bring 99.99% of the benefit without needing much of anything power-wise versus regulating the output stage..
The advantage of a regulated power supply for an audio amplifier is 1) it reduces the ripple on the storage capacitors (the huge electrolytics just after the transformer and rectifier) and it allows you to use the power capability of the amplifier to its maximum.
The maximum output power of an amplifier is determined by the maximum supply voltage (assuming the current capability is sufficient).
Let's for a moment assume the maximum supply voltage (in operation) to be 40V.
A normal transformer-rectifier-storage capacitors power supply (unregulated) can be designed for 40V at low loading. But when it is loaded fully, the voltage may sag (lower) some 10-15% due to winding resistance, increased voltage ripple etc. Therefore, when you actually needed the full supply voltage of 40V for full power, it has reduced to around 35V. The output power goes with the square of the voltage so you loose more than 20% of the power capability.
If you have a well-designed regulated power supply it has an output of 40V at little loading and 40V at full loading. Thus, more power and less hum that else could affect the amplifier. The downside is increased complexity and some power loss in the series element.
An SMPS does not have a series element and the ripple frequency is much higher (typically above 100KHz). Their problem may relate to the speed of the regulation loop such that if the output voltage starts lowering, it is not so quick in getting more power to the output and the voltage becomes less stable. Most SMPS are designed for general purposes and not for the very dynamic loading from an amplifier.
The maximum output power of an amplifier is determined by the maximum supply voltage (assuming the current capability is sufficient).
Let's for a moment assume the maximum supply voltage (in operation) to be 40V.
A normal transformer-rectifier-storage capacitors power supply (unregulated) can be designed for 40V at low loading. But when it is loaded fully, the voltage may sag (lower) some 10-15% due to winding resistance, increased voltage ripple etc. Therefore, when you actually needed the full supply voltage of 40V for full power, it has reduced to around 35V. The output power goes with the square of the voltage so you loose more than 20% of the power capability.
If you have a well-designed regulated power supply it has an output of 40V at little loading and 40V at full loading. Thus, more power and less hum that else could affect the amplifier. The downside is increased complexity and some power loss in the series element.
An SMPS does not have a series element and the ripple frequency is much higher (typically above 100KHz). Their problem may relate to the speed of the regulation loop such that if the output voltage starts lowering, it is not so quick in getting more power to the output and the voltage becomes less stable. Most SMPS are designed for general purposes and not for the very dynamic loading from an amplifier.
Last edited:
A regulated SMPS != a regulated linear supply, where the excess ripple (and sag?) need to be regulated away. Perhaps a misunderstanding, but I figured the OP's question was for the latter. With the ease that we can acquire good-sized SMPS modules, it definitely has its value.
My power amps don't need a regulated supply as I always use front end decoupling to get rid of hum.
The maximum power is determined only by how much money you have. You can always raise the voltage, caps, current, etc, to reach whatever power goal you can *afford*.
An unregulated supply will usually be more watts per buck.
There are exceptions. The Dynaco 120 used transistors right-TO their Vce ratings (like 72V those days). In that economic it was cost-effective to insert a regulator. Some of Nelson Pass' amps (not intended to be high W/$) have low PSRR and a regulated supply is part of his design. And in this Brave New World, SMPS is the key to portable 9,000W amps and they can incidentally regulate.
Last edited:
Actually this is not true. There is no change in the ripple on the storage caps because the regulated supply comes AFTER it.
Not correct. You always loose some volts across the regulator so the supply voltage (with a given transformer/rectifier/storage cap) with a regulator is lower, meaning lower output power.
Further, with an unregulated supply the output power may be higher for a short moment before the supply voltage starts to sag to what it is at constant full power. That gives the amp something that is called dynamic headroom and can be several dB. With a regulated supply the dynamic headroom is 0dB.
So when your goal is max dynamic power output (with a given transformer/rectifier/storage) you should not use a regulated supply.
Jan
Thanks for your well-respected reply. Why is it then NOT true for lower-voltage applications (preamps)?
Two main reasons: preamp power supplies are relatively easy to set to whatever voltage you need without much additional cost. And supply voltage doesn't sag with output level so dynamic headroom is always 0dB.
With a power amp, an unregulated full-power 40V supply may on short peaks actually be 45V so delivering short peaks higher than the 40V would suggest. This is not the case with preamps.
Jan
You could hugely over rate your regulated power supply, but it would probably end up costing more than the amp for no noticeable improvement. Then, why not consider batteries instead?
That would seem to come at a cost, however. By allowing a higher peak, you may be running outside the SOA of the output device, which would NOT be the case with a regulated supply.
Of course not. A competent design has sufficient SOA space for anticipated operating conditions.
Besides, the crux of SOA is that it allows high level short duration excursions.
Jan
- Status
- Not open for further replies.