While I was searching for more info on "all digital" or "Digital class D" amplifiers, I didn't find very many diy projects.
I am looking for projects only on PCm->PWM->Analog audio amplification, and not Analog->PWM->Analog.
Why is that so ? Are there any limitations like availibility, cost, etc ? Or there are just not any favourable opinions from people who have built them ?
I was specifically trying find out more about projects based on TI Equibit. And TAcT and also PowerDAc.
I am looking for projects only on PCm->PWM->Analog audio amplification, and not Analog->PWM->Analog.
Why is that so ? Are there any limitations like availibility, cost, etc ? Or there are just not any favourable opinions from people who have built them ?
I was specifically trying find out more about projects based on TI Equibit. And TAcT and also PowerDAc.
but what do they know about dacs 😉 - ok just kidding 
.
I mean I posted it in the digital forum on purpose because I was specifically interested in comparison with 'conventional' (for the lack of a better term) means of digital to analog conversion.
For example, if anyone built and/or listened to both a diy nos dac and a pcm->pwm->audio amplifier then I'd like to hear about it.
I will try searching in the class d forum again but its mostly analog input class d amps that fly over there.
.I mean I posted it in the digital forum on purpose because I was specifically interested in comparison with 'conventional' (for the lack of a better term) means of digital to analog conversion.
For example, if anyone built and/or listened to both a diy nos dac and a pcm->pwm->audio amplifier then I'd like to hear about it.
I will try searching in the class d forum again but its mostly analog input class d amps that fly over there.
Ummm.
All DAC manufacturers embraced the new religion of 1-bit sigma delta when it came out, not because it was better, but because it was cheaper. Next thing you know, they all wake up with a major hangover because the sound really sucked so bad (remember CD63 from Marantz ?), and all switch to multibit sigma delta running at higher frequencies.
IMHO full-digital amplifiers would just repeat the same errors from the past...
All DAC manufacturers embraced the new religion of 1-bit sigma delta when it came out, not because it was better, but because it was cheaper. Next thing you know, they all wake up with a major hangover because the sound really sucked so bad (remember CD63 from Marantz ?), and all switch to multibit sigma delta running at higher frequencies.
IMHO full-digital amplifiers would just repeat the same errors from the past...
Comparing TI Equibit in Panasonic receivers to UCD180AD kit... I prefer the UCD by a fairly large margin.
The TI Equibit receivers seemed to be at their best when fed 24/96 source material digitally. Otherwise, they average sounding, but unspectacular. How well I liked them seemed very dependent on speaker load they were hooked to.
The TI Equibit receivers seemed to be at their best when fed 24/96 source material digitally. Otherwise, they average sounding, but unspectacular. How well I liked them seemed very dependent on speaker load they were hooked to.
> How well I liked them seemed very dependent on speaker load they were hooked to.
Do they have feedback across the output inductor like UcD ? If not, speaker impedance matters.
Do they have feedback across the output inductor like UcD ? If not, speaker impedance matters.
Why do you want to do the PCM to PWM conversion in the digital domain? As Bruno Putzeys and others have pointed out many times, this is a foolish way of making a PCM-input power amplifier. D2A audio have a solution that costs many times more than a UCD amplifer + low-level D/A, but has significantly poorer performance.
well, had I looked at how really the pcm->pwm conversion is done I would have never posted this question.
For some reason I thought the chopping of the input signal with triangular/sawtooth carrier wasn't required for pcm data. I can see the problem now...
thanks all for your inputs on the subject.
(now you see why I posted it in the digital forum 😉 ? )
{edit: this paper helped so I thought I'd link it here - http://www.mathworks.com/applications/dsp_comm/pdfs/hbreschpadgett.pdf }
For some reason I thought the chopping of the input signal with triangular/sawtooth carrier wasn't required for pcm data. I can see the problem now...
thanks all for your inputs on the subject.
(now you see why I posted it in the digital forum 😉 ? )
{edit: this paper helped so I thought I'd link it here - http://www.mathworks.com/applications/dsp_comm/pdfs/hbreschpadgett.pdf }
The main problems of the "digital amplifier" are IMHO :
1- feedback from the analog output !!!!! (output coil impedance vs speaker impedance) this is the major gotcha, since the amp essentially works open loop, so the sound will depend a lot on the speaker's (mis)behaviour
2- time-discrete output waveform, versus analog-time for analog class D like UcD (UcD has a on/off output like all class D, but since the switching times are analog, it has an infinite number of pulse widths to choose from, versus a pulse width that must be a multiple of the clock period for a digital amp)
3- you only get 1 bit, well maybe 1.5 bits if both halves of the H-bridge are independent
4- switching frequency can never be as high as a low-power MOS DAC, therefore some mix of sigma-delta and PWM is required, and sigma-delta needs to compensate the low-bit samples with a really high frequency to work well
1- feedback from the analog output !!!!! (output coil impedance vs speaker impedance) this is the major gotcha, since the amp essentially works open loop, so the sound will depend a lot on the speaker's (mis)behaviour
2- time-discrete output waveform, versus analog-time for analog class D like UcD (UcD has a on/off output like all class D, but since the switching times are analog, it has an infinite number of pulse widths to choose from, versus a pulse width that must be a multiple of the clock period for a digital amp)
3- you only get 1 bit, well maybe 1.5 bits if both halves of the H-bridge are independent
4- switching frequency can never be as high as a low-power MOS DAC, therefore some mix of sigma-delta and PWM is required, and sigma-delta needs to compensate the low-bit samples with a really high frequency to work well
There's now one project described in this thread; http://www.diyaudio.com/forums/showthread.php?threadid=91148
- Status
- Not open for further replies.