Hi builders!
If one builds a class D amp. Can the input be digital all the way in the pre section? So for to say you can connect Spdif direct to it.
Cheers!
If one builds a class D amp. Can the input be digital all the way in the pre section? So for to say you can connect Spdif direct to it.
Cheers!
It would be called a 'FDA' (full digital amp) rather than a class D. Class D amps usually have analog inputs.
Unfortunately the digital signal must be decoded first using a D-A for class D or an algorithm to decode the signals and turn them into something analogue to feed your loudspeakers as the bits are jumbled up and then have to be decoded before amplification.
If you read the Encoding for Audio Digital, it explains how the digital signal is encoded and of course what the flags are to decode it. Very interestingly cunning.
http://www.google.ch/patents/US6161088
If you read the Encoding for Audio Digital, it explains how the digital signal is encoded and of course what the flags are to decode it. Very interestingly cunning.
http://www.google.ch/patents/US6161088
Last edited:
Class D amplifiers are NOT digital amps. Yes, the 'D' might be a bit misleading but it does not stand for digital, the amplifier types were just alphabetically named. That said, you indeed need a decoding of some kind.
Like already stated, Class D isn't digital, therefore they can't be 'full digital' at all. That's a marketing slang.
If you go back to my era, digital meant either on = 1 or off = 0.
The output stage of a class D amplifier is ideally either on or off so that is where the name came from. It is actually a PWM or an FM amplifier.
The output stage of a class D amplifier is ideally either on or off so that is where the name came from. It is actually a PWM or an FM amplifier.
Yes, that's absolutely correct.
Well, no. Neither. FM radio is also not digital. Just because something is switched on or off does not make it digital ("My desk lamp is digital!", "My fan is digiital, it only got on and off!" 😛). The signal is still analog, the signal is in the width of the pulses and that varys analogue, there are no steps or numbers defined in it. And yes, I know you know it, you just say something else.
The names come from the naming convention of amplifiers, class A, class B, class C and so on. So in 1958 when they started the development, they've reached the 'D'.
As a matter of interest (or maybe not!) how about an amplifier with 16 switching transistors in a ladder formation of some kind directly switching the 16 bit (in this example) code. Don't tell me.....it's been done....!
What do you hope to hear? Without an algorithm and memory, how are you going to put the bits in the correct order?
Decode the bit stream into 16 bit parallel code, as in 2 bytes? then each transistor is switching a corresponding voltage, using a parallel voltage summer for example
Last edited:
Direct digital class-D
You can get a DSP or dedicated custom IC to convert a PCM input to PWM purely in the digital domain. TI have full-digital versions of some of their integrated class-D amplifier chipsets.
It is very difficult to implement the NFB loop if doing it this way though. I've not heard the latest solutions, but the early ones were pretty dire.
You can get a DSP or dedicated custom IC to convert a PCM input to PWM purely in the digital domain. TI have full-digital versions of some of their integrated class-D amplifier chipsets.
It is very difficult to implement the NFB loop if doing it this way though. I've not heard the latest solutions, but the early ones were pretty dire.
What I mean is a parallel voltage summer dac using power transistors to do the voltage switching right at the output. Presumably there is a reason this can't be done?
Well... you could try!
But think of what is involved. Imagine the MSB is switching a 64V signal, so in a 16-bit word (65536 steps) the LSB would be switching a 1mV signal. That's a lot to ask, both in terms of the switching speed, accuracy and noise rejection.
But think of what is involved. Imagine the MSB is switching a 64V signal, so in a 16-bit word (65536 steps) the LSB would be switching a 1mV signal. That's a lot to ask, both in terms of the switching speed, accuracy and noise rejection.
Ofcourse that can be done. The problem is, you can only 'create' a certain number of voltages. And for every voltage you need at least one power transistor because adding up like a normal digital number does not work. That means you get either a low number of different voltages and and a very low resolution, especally at low volume/voltages or you get an insanely high number of parts and an equally high count of possible points of failure. Not cheap, too.
Or did you mean a DAC which directly controls a PWM? That's not digital then. It doesn't really give you an advantage.
Hi, this is the kind of thing I meant, expanded of course https://en.wikibooks.org/wiki/Circu...ge_Summer#/media/File:Par_summer_dac_1000.jpg
What advantage do you expect from that? I mean, you still need electronics to translate the digital stream into the bits. Do you want to implement it on the power or on the input?
Right at the output, it obviously won't work or someone would have done it! 😀 Is was just a thought, why not do the parallel 16 bit (or whatever) decoding right at the output?
Why are you convinced that nobody has done it?
There are enough crazy guys in this world who cannot calculate,
who cannot simulate, but some of them can build.
In any case it is completely unattractive.
- For 16Bit resolution you need 16 halfbridges + gate drivers.
- Massive losses in the resistive summing network.
- Output impedance is massively depending on the signal voltage.
There are enough crazy guys in this world who cannot calculate,
who cannot simulate, but some of them can build.
In any case it is completely unattractive.
- For 16Bit resolution you need 16 halfbridges + gate drivers.
- Massive losses in the resistive summing network.
- Output impedance is massively depending on the signal voltage.
Haha. When I first heard of what were once called "digital amplifiers" many years ago, I had this vision of 16 parallel transistors doing what I've tried to describe and thought, what a great idea. I since found out this was not what was happening at all and was immensely disappointed and felt I'd somehow been lied to.
This thread appeared and I just couldn't help myself.......call it a wish/fantasy fulfillment, I don't know, almost like a kind of therapy, I want to know why I can't have my cake and eat it.......🙄😀
This thread appeared and I just couldn't help myself.......call it a wish/fantasy fulfillment, I don't know, almost like a kind of therapy, I want to know why I can't have my cake and eat it.......🙄😀
- Status
- Not open for further replies.