Point is better performance is achievable without paying a premium for the flavor of the month output device in an out of date circuit. I, and many others, would rather pay for engineering rather than marketing.
Yeah... nothing changed much... they still have their magic potion, so no mining until further notice...
On paper yes,
As those who know, global feedback is a "sometimes" necessary evil, just read papers of Nelson Pass, Dan Agostino etc etc
If you can get away without using it globally just locally and yet retain good specs still you get better "sound" advantages by not having it.
Cheers George
Yup. The input on the class D amp compares the input signal to a saw tooth oscillator signal. Everything I've been reading has been referring to that process as sampling the input signal. The saw tooth oscillator has to have a high enough fequency to sample quickly enough to accurately produce an encoded signal that will produce the desired upper frequency range on the amp's output. Seems I was wrong earlier about the timing being continuous. So there's a conversion step that involves sampling with a minimum sample rate for the desired bandwidth to be produced, and the result is a binary signal that can be decoded once produced by simply filtering the high frequency hash out of the signal. A line level DAC can also be made using this process, with PCM or other encoding being fed in to it, and then converted PWM using DSP, produce the PWM signal at the appropriate speed and then all you need is an appropriate filter on the output and out comes the sound. With the class D amp, we're not bothering to store the sampled information. We're just immediately producing it, filtering out the unwanted high frequency hash, and driving the speaker with it. There's no way around the fact that an analog sine wave signal got time and amplitude quantized through a conversion process, and then coverted back to a sine wave.
Last edited:
You are describing the interface to the output stage... An amps inout stage may be a differential input or perhaps just a impedance buffer. I hope we can agree that that is at leats analog 🙂 Then comes the output stage - and that is what the "Class" is all about.
I will now leave all poor souls that think Class-D is "digital" to their beliefs. I'm done here... good luck ;-D
//
Poor souls indeed.
How about that output stage, seems to me those transistors are either hard-on or hard-off, the devices enjoying the output current flow through RDSon values in milliohms, hence the low Pdiss of the design. The fully on/off nature of that particular circuit arrangement walks like a digit - to me. When that switching action gets integrated by the output filter, then the signal can be brought back to the amplifier input as, say, post filter negative feedback - an analog signal.
Pre output filter negative feedback is beyond my comprehensional continuity, though that's been done too. I think that latest voodoo is the post filter style feedback sounds better, trumping the digital all the way implementations - at least those of several years ago now.
In most descriptions of class D amplifier they start with the interface https://en.wikipedia.org/wiki/Class-D_amplifier
No input stage is described. Just an input
No input stage is described. Just an input
Because they only discuss the output part of an "amp". So they dont discuss what you call an "amp" - they discuss an output stage topology - dont you get that? The "Input" in that picture is not your usual RCA connector on the back of your Pioneer amp... capiche?
//
//
I say it again - the "Class" description has noting to do with the input part of a commercial consumer product - it is solely a definition of the last 1/3 of it, the output stage working.
//
//
The confusion here is "quantizised".
An analog system has an upper bandwith.
So the sawtooth can only be nearly perfect. Also the comparator has an upper bandwith. So they are not perfect. But we can still call them analog.
In a digital system the quantisiztion steps are clearly defined with a sampleperiod.
In an analog amp the steps are random if they are measured digitally. Its just the noise and bandwith that defines the smoothness of the sawtooth.
The same with the comperator. Only the bandwith and accuracy defines the presission of the conversion.
Also have in mind the fact that all signals can be caracterised by a sum of sinus waves. (Laplace?)
So the conversion adds sinus waves above the signal to make the square waves. Then they are removed with the low pass filter.
No quantizition involved or quantizition noise only analog prosesses and analog noise.
Then of course one can make the sawtooth digital or the comparator digital. Then there will be quantization noise and a digital interface before the output transistors
Because it's frustrating to read that some people don't get the basics right, which can be found in many books and literature?
(or like literally use google for 10 seconds)
At the same time spreading misinformation?
Last edited by a moderator:
An analog system doesn't have discrete samples, random or not is not very important.
But for a better explanation, go back a few pages.
Did I already mentioned the Groundhog Day movie.......?
There exist a lot of analog systems with digital control functions.
What du you call a Class D amp with a digital generated sawtooth? The output waveform will have quantization noise from the sawtooth even if the signal and comparator is analog.
What du you call a Class D amp with a digital generated sawtooth? The output waveform will have quantization noise from the sawtooth even if the signal and comparator is analog.
Read Bruno's paper and then get back to us with your objective reasons why he is incorrect rather than just waving your hands and giving a subjective opinion.
Was trying to say that. But failed. But think it is important if the system shows quantization noise or only the analog noise types.
Maybe the early telecom pioneers was right, it is only about the best signal/noise. Digital/analog who cares.
And feedback is just a tool to get the right distortion/bandwith/gain numbers + stability of cource
- Status
- Not open for further replies.