"True" digital amp as 2nd amp project?

[Spasticteapot]

I'm working on my 100% DIY gainclone, and I think I've got it right.

However, after looking at the insanity related to DACs, and considering that I'm going to be playing music from a hacked Compaq IA (an "El Cheapo" linux computer) or a CD player anyway, I figure I may as well go digital. (I don't own any records, and NPR sounds just as good streamed to my PC.)

I'd like something small, inexpensive, and simple. I don't require much power, though; 15 watts would be more than I need, and even 10 watts should leave a nice margin of error. Quality is important, though.

 

[Pano]

Do you mean a power DAC?

 

[Spasticteapot]

Do you mean a power DAC?

I'm not quite sure of what the exact definition is, but I've been reading about amplifiers that take an I2S input, and produce a PWM output for a Class-D style H-bridge.

 

[Pano]

I havn't used one, but AFAIK they simply have an onboard DAC. The rest of the amp is analog, just like any other class-d. Don't know if there are any that actually modulate the power section directly from digital.

If the onboard DAC is decent, then why not? How do you control volume?

I'm sure someone here has used one of these chips. Maybe they'll speak up.

 

[peufeu]

Well, simple.

A full digital amplifier will take digital data in I2S and output a series of pulses. It is like a 1-bit DAC, except it has 3 output levels if you have a H-bridge.

Now a low power 1-bit DAC can use a high bit rate to get away with the inherent suckiness of the concept, something like 12 MHz usually.

However you can't possibly run a high power output stage at that frequency unless you want to prevent everyone from using their TVs and radios in a 300 m radius... and good luck designing the output stage.

You'd need something like TacT.

Also a full digital amp can't (by definition) have feedback around the output coil (unless using some sort of ADC feedback).

See the can of worms ?

UcD sidesteps the problems by :
- being balanced : much less EMI problems
- wrapping the feedback around the output filter
- being analog : the output waveform is amplitude quantized (-Vcc or +Vcc) but it is time-continuous (which means infinite resolution). A power DAC would have a time-quantized and amplitude-quantized output (which means finite, probably low resolution).

 

[Spasticteapot]

Well, simple.

A full digital amplifier will take digital data in I2S and output a series of pulses. It is like a 1-bit DAC, except it has 3 output levels if you have a H-bridge.

Now a low power 1-bit DAC can use a high bit rate to get away with the inherent suckiness of the concept, something like 12 MHz usually.

However you can't possibly run a high power output stage at that frequency unless you want to prevent everyone from using their TVs and radios in a 300 m radius... and good luck designing the output stage.

You'd need something like TacT.

Also a full digital amp can't (by definition) have feedback around the output coil (unless using some sort of ADC feedback).

See the can of worms ?

UcD sidesteps the problems by :
- being balanced : much less EMI problems
- wrapping the feedback around the output filter
- being analog : the output waveform is amplitude quantized (-Vcc or +Vcc) but it is time-continuous (which means infinite resolution). A power DAC would have a time-quantized and amplitude-quantized output (which means finite, probably low resolution).

What about that Panasonic reciever (the XR-55) and the so-called "D-amps?"

 

[peufeu]

Ah, apparently the Panasonic uses Equibit technology (seems TI bought TacT). While the famous TacT amp was a one-bit PWM design running at about 100 MHz this one runs at the usual frequency of 12 MHz... the datasheet specifies a SNR/dynamic range of 90 dB and doesn't tell anything about the innards of the chip ; but the original equibit converts the PCM data to something like 8 bits at 300 kHz and then PWM's it. This solves problems with the output stage.