Is there any reason to add bulk storage capacitance to a Class A amplifier. If it uses constant power, then is there any point to a charge reservoir? Please assume the power supply produces no ripple.
Without bulk capacitance, there will be a rectified sine instead of a steady DC voltage.
The bulk capacitance supplies the charge for the current that is drawn by the load.
The capacitance holds up the voltage to a degree, depending upon the load. The remainder is the ripple.
There will always be ripple to some degree, since the rectifier cannot supply enough current at all times.
The rectifier supplies current pulses that recharge the bulk capacitance, while the load draws current from it.
Class A amplifiers actually require much more bulk capacitance than a class AB amplifier, since the
maximum current is required at all times, rather than on signal peaks as for a class AB amplifier..
The bulk capacitance supplies the charge for the current that is drawn by the load.
The capacitance holds up the voltage to a degree, depending upon the load. The remainder is the ripple.
There will always be ripple to some degree, since the rectifier cannot supply enough current at all times.
The rectifier supplies current pulses that recharge the bulk capacitance, while the load draws current from it.
Class A amplifiers actually require much more bulk capacitance than a class AB amplifier, since the
maximum current is required at all times, rather than on signal peaks as for a class AB amplifier..
Last edited:
For a Class A PSU that uses a capacitance multiplier to eat the ripple, only enough capacitance is needed to present the minimum level of ripple needed for the cap multiplier to flatten out the ripple to almost negligible levels. Since Class A amps have constant current needs, a cap multiplier is the ideal for smooth power with less demanding need for huge amounts of caps. For the SLB Class A PSU, for example, 5A of continuous current at 37v can be maintained with only 1mV rms ripple using 2x 15,000uF caps. Without a cap multiplier, you would need a LOT more bulk caps and probably never quite get 1mV ripple at 5A. At those kinds of currents it makes sense to use an active bridge to minimize thermal losses through the rectifier.
Last edited:
Consider a class A amplifier and a fixed voltage rail power supply as two good friends. The first is changing all the time while the second wants to keep their relation steady. The ac amplifier is not perfect and is depending on the stability of his friend, whereas the power supply, actually better understood as a dc amplifier, is not perfect too and depending on the evenness of his friend too. They're intertwined and no better then as good as the other. To ease the heated debate they've agreed to smoothen things out with large caps. The amp gets the trampoline on top and the supply the concrete block at the bottom. Both happy.
Strictly spoken, the supply should act as a perfect voltage source indifferent to whatever the load, so no large caps needed. Given reality limitations, applying class ab or b amps to inperfect supplies is asking for trouble. Save bridge amplifiers, who slosh the currents and need less solid supplies. But still needs this dampening caps however.
Strictly spoken, the supply should act as a perfect voltage source indifferent to whatever the load, so no large caps needed. Given reality limitations, applying class ab or b amps to inperfect supplies is asking for trouble. Save bridge amplifiers, who slosh the currents and need less solid supplies. But still needs this dampening caps however.
If we mean the push-pull output stage, this is a half-bridge circuit. Another half-bridge is formed by power sources.
Bulk storage is about storing energy during the mains zero-crossing, not about the amplifier load (other than it has to be sufficient for the maximum amplifier load). In class A the maximum is the same as the minimum load, unlike a class AB/B/D/G/H
If the supply has no ripple, its not a transformer/rectifier style analog PSU and has its own internal bulk storage.
Class A current draw
My understanding was that on average the current draw from class A is constant - but at any point in time it can be anywhere from zero to 2iq...
My understanding was that on average the current draw from class A is constant - but at any point in time it can be anywhere from zero to 2iq...
On true Class A circuits the current drawn from PSU is constant. Music is generated by changing the relative current that gets dissipated in the output transistors vs current going to the load. The current going to the load varies from 0 to 2x the quiescent current. It’s easy to test this by monitoring current through PSU on O-scope as music is played. It’s always quiescent current plus ripple. The ripple amplitude may change a bit with music but it’s small compared to main DC level.
Only if the class A design uses a constant current load and returns the AC speaker current to the switching device rail is the current constant. Class A isn't inherently constant current.
> Class A isn't inherently constant current.
Right; over-simplification.
In the old days, we only had needle meters. On a needle meter, the current in a class A stage is "constant". The needle is slower than most audio waves.
In the Tubes section, we would have a big choke transformer in the amp. Then the DC supply current is "constant". (Alternatively we have two tubes parallel on the power, pushing/pulling to cancellation.) (The half-bridge A power amps were very rare in tubes.)
But this is the Transistor section where two-device half-bridge is common. Any indicator faster than a needle-meter will show the varying current.
Right; over-simplification.
In the old days, we only had needle meters. On a needle meter, the current in a class A stage is "constant". The needle is slower than most audio waves.
In the Tubes section, we would have a big choke transformer in the amp. Then the DC supply current is "constant". (Alternatively we have two tubes parallel on the power, pushing/pulling to cancellation.) (The half-bridge A power amps were very rare in tubes.)
But this is the Transistor section where two-device half-bridge is common. Any indicator faster than a needle-meter will show the varying current.