This would be the basis of my ultimate dac, too, I have to say. After lots of experimentation over a lot of years, I don't think there is a genuinely better dac chip than the 63. PCM1702/1704 only come close, AD1860/1865/1864 & TDA1541-types distant third, and no bitstream chip I've heard yet can come close.
The Accuphase DC91 used 16x PCM63PK per channel, in single-ended mode(balanced out was done with phase-splitter/bal-lin-driver hybrid module), and although I never acquired a schematic for it, in extensively rebuilding one two years ago with my own no-loop-feedback complete analog stages, I can't recall seeing any extra logic driver chips for distributing the DF chip's output lines to the many dacs. I think just a small series resistor, say 47-100ohms in front of each dac's input pins, should be sufficient to prevent excessive loading of the signals, although I could be wrong. Worst case is you may need to add a line driver, such as a 74HCT08, to each of the data/clock lines. As far as getting negative phase signal for balanced, as far as I've seen on dac units that do this, all one needs to do is add an inverter, ala 74HCT04, to the audio data line to the negative bank of dacs.
Personally, I would use a DF1704 digital filter, set for 20bit output obviously. I think the whole NOS craze is just a "side effect" of feedback i/v & lpf circuits. The lower frequency glitch artifacts are a lot less damaging to the sound in such circuits, IMO, so people assume that NOS sounds better than OS, when, in fact, IMO, the opposite is actually true once loop feedback is eliminated after the dac.
As for passive output direct to amp, it's not going to work without a VERY high gain amp. If you don't have an i/v resistance that's low enough, you run into clipping caused by the internal protection diodes. With the 16 dacs in the DC91, I ended up with an i/v value of 7ohms. You could very, very effectively use a really high quality transformer(to me, this means Cinemag in currently produced ones), though, as the 16 dacs could drive a very low impedance primary to a much higher impedance secondary with superb preservation of dynamics. The only problem then is preserving line drive current after the transformer & through whatever attenuation device is used for volume control before the amp.
Not sure why you'd want to use a WM8804. Were I to use a Wolfson(now Cirrus, sadly) part, it would be the way easier to use WM8805, but I personally prefer the B-B DIR9001.
The Accuphase DC91 used 16x PCM63PK per channel, in single-ended mode(balanced out was done with phase-splitter/bal-lin-driver hybrid module), and although I never acquired a schematic for it, in extensively rebuilding one two years ago with my own no-loop-feedback complete analog stages, I can't recall seeing any extra logic driver chips for distributing the DF chip's output lines to the many dacs. I think just a small series resistor, say 47-100ohms in front of each dac's input pins, should be sufficient to prevent excessive loading of the signals, although I could be wrong. Worst case is you may need to add a line driver, such as a 74HCT08, to each of the data/clock lines. As far as getting negative phase signal for balanced, as far as I've seen on dac units that do this, all one needs to do is add an inverter, ala 74HCT04, to the audio data line to the negative bank of dacs.
Personally, I would use a DF1704 digital filter, set for 20bit output obviously. I think the whole NOS craze is just a "side effect" of feedback i/v & lpf circuits. The lower frequency glitch artifacts are a lot less damaging to the sound in such circuits, IMO, so people assume that NOS sounds better than OS, when, in fact, IMO, the opposite is actually true once loop feedback is eliminated after the dac.
As for passive output direct to amp, it's not going to work without a VERY high gain amp. If you don't have an i/v resistance that's low enough, you run into clipping caused by the internal protection diodes. With the 16 dacs in the DC91, I ended up with an i/v value of 7ohms. You could very, very effectively use a really high quality transformer(to me, this means Cinemag in currently produced ones), though, as the 16 dacs could drive a very low impedance primary to a much higher impedance secondary with superb preservation of dynamics. The only problem then is preserving line drive current after the transformer & through whatever attenuation device is used for volume control before the amp.
Not sure why you'd want to use a WM8804. Were I to use a Wolfson(now Cirrus, sadly) part, it would be the way easier to use WM8805, but I personally prefer the B-B DIR9001.
- Status
- Not open for further replies.