Class AB with Regulated Power Supply...Class A-like?

[djoffe]

Class A amplifiers can be among the most linear of amplifiers, and the least efficient. They draw a constant current from the power supply, no matter what signal is being delivered to the speaker. That says that a Class A amplifier doesn't modulate the supply rails..constant current drain produces a constant load on the power supply, and hence a constant voltage.

Is a Class AB amplifier with an electronically regulated power supply almost as good in that respect? If the power supply's output impedance is low, then the output signal minimally modulates the rails. That should cause less distortion at the output. In that sense, the Class AB with regulated power supply should be about as blameless as a Class A amp.

The one way that the Class A amplifier still has it over the Class AB is that at low listening levels, the output device has nearly constant current. That is, its operating current is the large class A bias current with the small signal modulations.

Of course, at high output levels, the Class A advantage diminishes, as the output device then experiences wide swings in operating current.

Now, the inherent power supply rejection of a typical AB amp certainly helps reject the signals on the power supply, but might there still be some advantage to the regulated supply for AB amps?

[6L6]

Originally Posted by djoffe

Now, the inherent power supply rejection of a typical AB amp certainly helps reject the signals on the power supply, but might there still be some advantage to the regulated supply for AB amps?

As an amplifier is basically an AC oscillator for the DC power supply, yes, there is advantage in having a very quiet (regulated) supply.

One of the challenges is that as the amp gets bigger, the regulator for the PSU is almost as big (or sometimes bigger!) than the amp itself.

EDIT - It's probably easier and arguably better for the DIYer to just make a big honkin' supply. CRC with lots of capacitance. You can even try CLC. In production it is usually much cheaper to make a regulated supply rather than a huge one. Overkill in PSU is a time-honored tradition in DIY.

Wimpy PSU are the norm in production.

[5th element]

Self has shown that any induced distortion, from the modulated power rails in a class AB amplifier, can be reduced to inconsequential levels. Also any modulation of the supply lines shouldn't have any gross effect beyond that unless the amplifier is driven into clipping.

What sets the class A amplifier apart from the class AB are the switching components and the only way to get rid of these is by biasing into class A.

You said yourself that where the class A amp has an advantage is at low signal levels, but this is also wherr the unregulated power supply shine, with minimal ripple or sag present.

[MarcelvdG]

Originally Posted by 5th element

What sets the class A amplifier apart from the class AB are the switching components and the only way to get rid of these is by biasing into class A.

There are ways to make non-switching class AB amplifiers, meaning that when the current through one output device gets large, the current through the other output device approaches some non-zero value. One way to realise that is with a class AB control loop that controls the way the current is divided between the output devices. For example, you can use a harmonic mean control or an exp(-K*I1)+exp(-K*I2) control. In fact I have one playing right now.

[sreten]

Hi,

A good class aB design simply doesn't need a regulated supply,
and implementing one would be a false expense. The ripple
on the supply does vary with load for class aB, but its a lot
easier to design the amplifier for high inherent PSRR than
regulating the power supply.

Your logic is flawed. The high constant current of class A will
produce a constant high ripple on an unregulated supply, for
class aB at low levels the power supply ripple is also low.

Its class A that suits regulation, or CRC or capacitance
multipliers, to reduce the ripple on its supply rails.

rgds, sreten.

[Samuel Jayaraj]

Marcel, I remember you published a design several years ago in EW which had something like the approach you now mention. But can't remember more or place my hands on a copy of the article.

[djoffe]

Originally Posted by sreten

Hi,

Your logic is flawed. The high constant current of class A will
produce a constant high ripple on an unregulated supply, for
class aB at low levels the power supply ripple is also low.

Sreten...You're absolutely right...the high current drain of the class A amp will produce more AC mains based ripple all the time. To clarify my point, it is that the the class A amp produces no signal-based ripple on the supply rails.

[Boscoe]

I really don't know why everyone raves on about class A and it's linearity, I've never seen a class A amp good in distortion they are usually terrible. Class AB is superior in every way unless you want a heater.

[ilimzn]

Actually, sometimes it is beneficial to have regulated power rails even for a class AB amp, one notable example would be the situationwhere PSRR cannot be designed into an amp at an 'appropriate' level such as in no-NFB amps, some hybrid amps, and also sometims in topologies where power supply noise can get to the input of the power stage through Creverse of the output devices. Also, a regulated power supply offers a lot WRT protection, and in some cases it can be difficult to separate the roles of regulation and protection. One very rare but notable example would be using SITs for output devices, where their triode characteristics makes the output stage have poor PSRR and voltage should not be applied before bias.

It can also be shown that the total dissipation of the AB output stage and the regulator is equa to a raw output stage up to clipping conditions for a constant output. A regulator will provide clean clipping under these conditions whereas a regular AB stage will intermodulate it's output with 100Hz ripple when it clips. When operated at just under clipping the regulator transistors will dissipate the portion of power caused by the rippled part of the supply voltage. Regulated stages however cannot easily take advantage of added peak power due to PSU sag, because it has to be taken into account when determining the regulated output voltage. Regular class AB will produce higher output if the transient is short enough so that the power supply does not sag appreciably while it lasts.

[MarcelvdG]

Originally Posted by Boscoe

I really don't know why everyone raves on about class A and it's linearity, I've never seen a class A amp good in distortion they are usually terrible. Class AB is superior in every way unless you want a heater.

The distortion of the output stage is inherently small when the variations in momentary current are small compared to the bias current, like in a class A amp. That makes it easier to obtain low distortion with class A than with class AB. Still, there are many ways to mess up your amplifier design, even if it is class A.