Building a Class D amp acting like Class A. Is it possible?

jencelo

Member
2009-05-24 12:44 am
I even assume (with no technical knowledge) it is probably a stupid question :)

OK. There is Class A. It is like running all the lights in the room even though you don't need all. However, it sounds awesome.

Why isn't it possible to do the same thing with Class D amps?

Is it totally different philosophy which cannot be done technically or it would have no advantage sonically?
 
Class A never has the transistor switching off to avoid cross over distortion of non linear region near turn on. Class D fully switches on and off for highest efficiency and switched pulses are filtered and recombined to a recreation of the analog signal from discrete switched bits. To make class D behave like Class A makes little sense as it would not switch on off anymore and then not be class D.
 

jencelo

Member
2009-05-24 12:44 am
Class A never has the transistor switching off to avoid cross over distortion of non linear region near turn on. Class D fully switches on and off for highest efficiency and switched pulses are filtered and recombined to a recreation of the analog signal from discrete switched bits. To make class D behave like Class A makes little sense as it would not switch on off anymore and then not be class D.

Thanks for the explanation (not that I fully understand). It is not you of course. It is me:)
I wonder if anybody tried this and if so, would it make the sound better or worse.
 
It is a totally different system.
Have a look here. https://en.wikipedia.org/wiki/Amplifier#Amplifier_types

Class A100% of the input signal is used (conduction angle Θ = 360°). The active element remains conducting[11] all of the time.Class B50% of the input signal is used (Θ = 180°); the active element carries current half of each cycle, and is turned off for the other half.Class ABClass AB is intermediate between class A and B, the two active elements conduct more than half of the timeClass CLess than 50% of the input signal is used (conduction angle Θ < 180°). A "Class D" amplifier uses some form of pulse-width modulation to control the output devices; the conduction angle of each device is no longer related directly to the input signal but instead varies in pulse width. These are sometimes called "digital" amplifiers because the output device is switched fully on or off, and not carrying current proportional to the signal amplitude.
 

rayma

Member
2011-04-29 8:37 pm
Is it totally different philosophy which cannot be done technically?

Think of the difference as how often the output devices turn on.
Class A means they are all constantly on, but in an amount related to the signal.
Class D means they are constantly switching between fully on and fully off
at a high frequency in a pattern related to the signal, and the resulting
waveform is filtered. The two concepts could scarcely be more different.
The sound quality depends on the execution as much or more than the class.
 
Last edited:
Class B really has no place in audio, that is where one half conducts for exactly half of the cycle because loudspeakers are reactive loads. Reactance causes the current delivery to become offset to the voltage so that regardless of where an oscilloscope says the output voltage is, the current could be coming from either half or from a combination of both halves of the output stage.

You'd have to be driving a purely resistive load for this to actually be the case.
 
OK. There is Class A. It is like running all the lights in the room even though you don't need all. However, it sounds awesome.

Its worth exploring the reason it sounds awesome which in my experience isn't primarily the traditional one about lower crossover distortion (which of course is the case). I think this is a secondary reason. Rather it has to do with how much noise is generated on the power supply - since classA has such a large standing current there's less variation of current taken from the supply.

Why isn't it possible to do the same thing with Class D amps?

ClassD's efficiency gives it an advantage over classB in that because less power is wasted in the heatsink there's less current taken from the supply, with the result that it gets closer to classA in terms of sound quality in my experience, most especially in the bass. ClassD has the bass slam and tunefulness of classA.
 
At the rate of improvement of the better class D amp (with incorporated dsp, specific power supplies, etc) it won't be long before these amps rival the sound reproduction of the better class ABs amps, not sure about the class A amps tho...

And, as we gain better understanding of the connection between the amp, it's dsp and specific speaker driver alignment, we can look forward to a dramatic improvement in the sound reproduction of the 'paired package' - there still seems to be a question about the 'smoothness' of the treble and some residual 'fatigue' that isn't much of a problem with the more traditional amps these days, but it's slowly getting sorted out - the improvement in the bass reproduction with the class D amps is astonishing (regularly seen in 'pro-audio' use).

my 2 cents ...
 
Its worth exploring the reason it sounds awesome which in my experience isn't primarily the traditional one about lower crossover distortion (which of course is the case). I think this is a secondary reason. Rather it has to do with how much noise is generated on the power supply - since classA has such a large standing current there's less variation of current taken from the supply.

Any well designed amplifier will have extremely high PSRR at audio frequencies.

Power supply modulation is also a very well known about, measured and explored part of modern solid state amplifier design. Douglas Self has done a significant amount of research behind this and has shown how one can design amplifiers that reduce any impact that this may have (beyond simply the amplifiers ability to reject them) to below the levels of the amplifiers inherent non linearities.
 
It is possible to build hybrid amps consisting of a switching- and a linear- part.
One possibility would be a class A amp with tracking PSU .
The other possible topology is a parallel configuration with a small class A amp which is supported by a parallel current source (aka Devialet).
Both would help to improve the linear amp's efficiency.

Regards

Charles
 
Class d has its own crossover distortion due to mosfet switching dead times.

Yes, exactly, and this is the key to find the ClassD equivalent of ClassA. If you increase the current on inductor (and decrease output current), you can get a circuit where the current during dead time never reach 0, therefore dead time (and signal amplitude) doesn't affect transfer ratio. Inductor current can be increased in AC in a simple half bridge, or in DC in a BCA topology.
 

russel

Member
2005-10-07 9:38 pm
Devialet do a class D amplifier with class A error correction, a nice way to do an amplifier. Been wondering how to implement it with the new Ti chips, maybe Tanaka's fully balanced bridge amplifier plus a 3251, not a trivial project though, be truly awesome if it worked.