Hi all,
I joined diyAudio this morning, as I am interested in conversating to member Michael vdG about class D amplifiers.
I am "currently under moderation" and "can start new conversations with other members" after aproval, e.g. by review of posts of mine.
This post, if there's interest, I could conversate to you all, about my DIY MP3-player, I build back in 1990, before MP3 / MP3-standardisation existed.
First I encoded a digital audio-source, compressing the audiodata 6 times. After, I decoded this data, regenerating the original audio without hearable quality-loss. This, according to human ear psychoaccoustical principles (I recall mr. Zwicker, a scientific researger in this area.
My algorithms were based on a German university-script of mr. Detlef Krahe, that I had access to. One or two years later, Krahe was involved in the Fraunhofer organisation (I believe), establishing / defining the MP3 standard (very similar to his university-work).
Back to my encoder in 1990: I encoded 6 audio-sources and joined them. This gave 1.4 Mb/s (standard CD: Fs 44100 Hz, stereo, 16 bits PCM). I bought a Marantz CD recorder. This was one of the first devices that could RECORD A BLANK CD!!!
I was concerned if this would work, because I wanted to increase the capacity of a CD (playable in a CD-player) 6 times, and the Marantz could only record AUDIO data, whereas I wanted to store/record PURE data. My data sounded as noise, but luckily the Marantz could handle it (input by S/PDIF).
I put a DSP inside of a normal CDplayer and only needed 5 hack-points of its internal circuit: power (5V and GND), bitclock (going to the DAC of the player, sampleclock and data, going from CD to DAC (in my case carrying the complessed audiodata) and from my DSP (outputting standard PCMaudio to DAC
CDplayer could be operated as usual (play, stop, pause, next, prev), and one extra button of mine, to select which one of the 6 channels.
Hope you like the story...
I joined diyAudio this morning, as I am interested in conversating to member Michael vdG about class D amplifiers.
I am "currently under moderation" and "can start new conversations with other members" after aproval, e.g. by review of posts of mine.
This post, if there's interest, I could conversate to you all, about my DIY MP3-player, I build back in 1990, before MP3 / MP3-standardisation existed.
First I encoded a digital audio-source, compressing the audiodata 6 times. After, I decoded this data, regenerating the original audio without hearable quality-loss. This, according to human ear psychoaccoustical principles (I recall mr. Zwicker, a scientific researger in this area.
My algorithms were based on a German university-script of mr. Detlef Krahe, that I had access to. One or two years later, Krahe was involved in the Fraunhofer organisation (I believe), establishing / defining the MP3 standard (very similar to his university-work).
Back to my encoder in 1990: I encoded 6 audio-sources and joined them. This gave 1.4 Mb/s (standard CD: Fs 44100 Hz, stereo, 16 bits PCM). I bought a Marantz CD recorder. This was one of the first devices that could RECORD A BLANK CD!!!
I was concerned if this would work, because I wanted to increase the capacity of a CD (playable in a CD-player) 6 times, and the Marantz could only record AUDIO data, whereas I wanted to store/record PURE data. My data sounded as noise, but luckily the Marantz could handle it (input by S/PDIF).
I put a DSP inside of a normal CDplayer and only needed 5 hack-points of its internal circuit: power (5V and GND), bitclock (going to the DAC of the player, sampleclock and data, going from CD to DAC (in my case carrying the complessed audiodata) and from my DSP (outputting standard PCMaudio to DAC
CDplayer could be operated as usual (play, stop, pause, next, prev), and one extra button of mine, to select which one of the 6 channels.
Hope you like the story...
Thank you Netlist.
Could you please inform https://www.diyaudio.com/community/members/marcelvdg.4072/ about my wish to ask him a question? And give him my email address?
(my <Start conversation> button still is disabled)
Could you please inform https://www.diyaudio.com/community/members/marcelvdg.4072/ about my wish to ask him a question? And give him my email address?
(my <Start conversation> button still is disabled)
good old days...
DSP I used was Motorola DSP56001: 10 mega-instructions / second, 24 bits fixed point. Audio needs at least 16 bits, however 8 bits extra, important for more precision e.g. to do recursive filtering of audio. Fixed point costs less than floating point precision, so Motorola decided to make their product (recomended for audio) 24 bits fixed point. Motorola was sold to Freescale, that resold it to NXP, a Dutch company I admire so much: related to royal Dutch Philips semiconductors)
DSP I used was Motorola DSP56001: 10 mega-instructions / second, 24 bits fixed point. Audio needs at least 16 bits, however 8 bits extra, important for more precision e.g. to do recursive filtering of audio. Fixed point costs less than floating point precision, so Motorola decided to make their product (recomended for audio) 24 bits fixed point. Motorola was sold to Freescale, that resold it to NXP, a Dutch company I admire so much: related to royal Dutch Philips semiconductors)
Your sigma/delta ADC (drawing above) will feed my very simple sigma/delta 1bit amplifying DAC: four transistors in fullbridge configuration, driving a loudspeaker through an analogue reconstruction filter, like STA518 (ST semiconductor; digital in, analogue out). I am new to sigma/delta, and a (to me) an intriguing phenomenon seems to appear, when looking at TPA3138D2 (Texas instruments; analogue in, analogue out). This class D digital amplifier seems to inherently do datacompression factor two. Its clockfrequency is 320 kHz, so 320 kbits/s. CD PCM is 700 kbits/s. Maybe its speakerdriver is more than 1 bit, because out+ and out- are driver independantly, resulting in 2 log 3 = 1.58 bits? Another reason could be TI is less than CDquality (THD+N, dynamic range)?
Looking at the functional block diagram (figure 9.2) in https://www.ti.com/document-viewer/tpa3138d2/datasheet I get the impression it uses normal analogue pulse width modulation. No time quantization like in a sigma-delta, hence infinite resolution.
I guess in analogue domain, it is possible to have infinite resolution even though "clocked" at a fixed rate. THIS IS THE CONFUSING PART (to me, at least at the moment). Digital amplication (meaning increasing analogue current in a wire) seems to be "somewhere inbetween" analogue and digital, as far as 1 bit PWM or sigma-delta is concerned.