Even some of the latest, TI, digital input class D, single chip amplifier solutions actually have a DAC on board.
The TAS5756M, for example, has a BurrBrown DAC and TI class D amp inside the same chip as individual operational parts.
Maybe this was done for more flexibility vs the previous generation, the TAS5706B.
The TAS5756M, for example, has a BurrBrown DAC and TI class D amp inside the same chip as individual operational parts.
Maybe this was done for more flexibility vs the previous generation, the TAS5706B.
I agree as currently implemented PWM amplifiers can arguably be called analogue but that's again an accident of historical implementation. Pulse widths are allowed to vary arbitrarily with an analogue input. If the allowable pulses widths were instead held to 2**16 or 2**24 discrete values would it still be analogue?
Admittedly spit balling here, a detector capable of pulling 'numbers' from such a PWM stream might exceed the limits of physics.
Yes. The signal is still being represented by an analogue, even though it has been forced to be discretized. It would still be an analogue amplifier even if the pulse width was produced digitally - this point would be the DAC in the system. The DAC is the point where a number is turned into an analogue.rdf said:
Exactly, the amplifier output/speaker interface instead of the amplifier input. That's what makes it interesting.
No. The DAC would occur at the point the pulse width is formed. This would have to be somewhere before the amplifier output, probably before the output stage and likely straight at the amplifier input.rdf said:
No it converts PCM to PWM. It converts digital to digital although that is debatable. As we have read what is analog and what is digital leaves some question marks. My teacher always taught me digital is analog too 😉 Anyhow, it does not convert digital to analog in the classical DAC way so one step (digital to analog) is omitted. It is analog at the outputs of course.
Patent US6594309 - PWM power amplifier with digital input - Google Patenten
In techno babble:
A digital input PWM power amplifier comprising:
an input circuit receiving pulse code modulated (PCM) digital input data organized in words of a first number of M bits at a bit rate, and outputting PCM digital data organized in words of a smaller number of N bits at a multiple of the bit rate; a first bus transmitting a first number of most significant bits (MSBs) of the N bit words output from said input circuit, and a second bus transmitting a second number of least significant bits (LSBs) of the N bit words output from said input circuit; first and second converters respectively fed from said first and second buses, each converter comprising a counter driven by a clock signal having a frequency equal to the product of the bit rate of the MSBs or LSBs transmitted on the respective buses times the base two raised to the respective first or second number of MSBs and LSBs, and a digital comparator receiving through a first input the reference digital words and through a second input the respective first or second number of MSBs and LSBs, and outputting a respective PWM signal; and an output power stage having an input receiving a driving signal defined by a PWM signal output by said first converter summed with an attenuated PWM signal output by said second converter.
Last edited:
The position appears to be PWM can't be digital by definition, even though it can carry numeric data, uses a clock, is subject to anti-aliasing, etc..
The PWM can be digital, the definition does not prevent that. But for it to be digital it needs to represent some symbols that are recovered without loss at the receiver side. It is definitely not what is happening in the context of the discussion and the signal discussed is not only not treated as symbols by the receiver, it does not even recoverable to the original symbols. So, all even imaginable properties of 'digital' signal are lost once PCM is converted to PWM in any class D amp. The process that does the conversion is clearly a DAC in such case, but if done at he high-current stage it is clearly a poor DAC as you can't have stable reference voltages and stable/predictable switching at high and variable currents.
Wrong question. They don't. While amps that appreciate analog nature of PWM measure and sound amazing.
Well the technologies are inherently different. A digital input class D amplifier uses digital processing to turn the PCM input into a PWM stream that directly drives a class D amplifier or class D power stage. From my limited knowledge of digital audio and processing there is no reason why this should be low quality it's just a different way to skin the cat. Traditionally class D output and power stages have been fairly lacklustre when it comes to absolute performance but things on that front are changing.
Certainly if you look at TIs digital input stuff that uses a PWM processor, followed by a PWM input, output stage, you'll see that the performance is nowhere near close to what flagship DACs can achieve but I can't see why this has to remain the same. TIs own new range of analogue input class D amplifier (TPA325x range) have much greater linearity in their output stages compared to the previous generation.
To be perfectly honest I am not sure how or what goes on in a typical D/S DAC I mean clearly it doesn't have a mini class D output stage inside being driven by a PWM train, so the comments on stable references are a little moot as they probably aren't required. Case in point if the conversion from PCM to PWM has already carried out the actual D/A conversion process, then all the PWM output stage is doing is working as a buffer to convert that PWM stream into a format that can actually drive a loudspeaker.
Certainly if you look at TIs digital input stuff that uses a PWM processor, followed by a PWM input, output stage, you'll see that the performance is nowhere near close to what flagship DACs can achieve but I can't see why this has to remain the same. TIs own new range of analogue input class D amplifier (TPA325x range) have much greater linearity in their output stages compared to the previous generation.
To be perfectly honest I am not sure how or what goes on in a typical D/S DAC I mean clearly it doesn't have a mini class D output stage inside being driven by a PWM train, so the comments on stable references are a little moot as they probably aren't required. Case in point if the conversion from PCM to PWM has already carried out the actual D/A conversion process, then all the PWM output stage is doing is working as a buffer to convert that PWM stream into a format that can actually drive a loudspeaker.
In my experience they do sound pretty good and that's from my only experience with mediocrely specified TI stuff.
I'll admit that's what I'm still not following. Ignoring amplifiers for the moment and using the simple general example of a 4-bit PCM signal, in general principle converting it to a 4-bit PWM signal with fully recoverable symbols appears to doable. The absolute level of that PWM signal would also seem irrelevant to retrieving the symbols. That should be true across devices from optical transmitters to voltage amplifiers as long as the waveform remains consistent with one modification in hindsight. In the case of a PWM audio amplifier those symbols would only be retrievable ahead of the final LC filter, not at the speaker terminals.
Again, it's understood that's not the way Class D amplifiers typically work.
Plain rubbish. I have several in my possession that sound from very good to excellent.
Yes. You can have even have it pretty noisy and unstable levels of "ones" and "zeroes", but as long as the receiver actually recovers the encoded symbols without errors or guesses - it is a digital signal. On the contrary for analog PWM all that matters a lot while "recoverability" doesn't matter at all.
LC filter just attenuates excessive high-frequency content, so does not change much if anything at all. And in case of audio amplifier they are not retrievable simply because of the error in switching times alone is already above the resolution of input PCM signal. Anyway it does not matter at all as this signal is never ever treated as such and no receiver ever attempt to reconstruct the symbols.
Exactly none of them work that way and never will. Simply because the fact that our ears and voice coil does not and will never understand symbols. They want an analog signal. As long as we are still humans and not robots. But then it is not an audio & sound anymore as we understand them today.
So, lets return back to our physical reality.
Last edited:
Then you should be able to name them... The ones that take feedback from the output doesn't count of course as this is an analog error correction. There is no much point arguing that the "DAC is removed from signal chain" or that there is nothing in the chain after a DAC and before the output if you replaced it with an ADC, that is in the feedback loop and so does the function of a low-power quality DAC. Still best designs like that are worse than good DAC + best pure-analog class D amp. No ADC can be as fast as a pure analog feedback. And you actually do more conversions between Digital and Analog on such designs... Just to fool yourself.
PCM is digital. PWM is analogue.jean-paul said:
PWM is not digital because it represents the original signal (a voltage) using an analogue (pulse width). No numbers, so not digital. The fact that you can, if you wish, represent a number using PWM does not make it digital; you can represent a number using a voltage level or current but that does not make all analogue circuitry actually digital.rdf said:
The essence of digital is using numbers - digits! There are various ways of representing numbers using electronics, binary being the most common. How you do it is not the issue.
PWM shares some issues with digital systems. Both use sampling, for example. Sampling alone does not make a system digital.
Last edited:
- Status
- Not open for further replies.