32-bit DAC from TI

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Could somebody tell which parts of the converter are 32-bit? Kind of not elaborated in the datasheet. The performance looks exactly like in their 24-bit versions, which makes me think that in 32-bit mode the data could be simply truncated right away, or latest after digital filter. Any way, doesn't make much difference because the dynamic range is 22-bit just like in any well made chip and hard to improve. Yes I know that many people think that any digital processing should be made with 64-bits, but is the digital filter actually different from their normal versions? Asking TI would probably not help :)
 
It's got a digital gain function. It can shift your 21 good input bits down into the lower output bits, and reduce playback level without disgusting analog mechanical potentiometers.
 
reduce playback level without disgusting analog mechanical potentiometers.

:)

But so does PCM1796, which, I believe is the same chip, but just not capable of 32-bit input stream.
My next question is where is this 32-bit stream used? Is this marketing trick or is there serious need for this product?
 
Jonte Knif said:
Asking TI would probably not help :)

We could ask Rochey, but he might not be back from his honeymoon yet.

It isn't entirely impossible that users have been clamoring for a 32-bit DAC ever since those chips from ESS Sabre hit the market. Not that those offer anything like 144dB dynamic range, mind you.

JD 'mineshaft gap' B.
 
Jonte Knif said:
Asking TI would probably not help :)

...  :eek:

Hello Gents,

it's 50% Marketing Mojo, 50% Real Engineering.

The dynamic range of the device is the same as the PCM1796, and the analog section is very similar.
The digital system has been revamped to have a 32bit interface, and 32bit datapath through the device.

Dynamic range (and THD for that matter) is a number given versus RMS noise. (if memory serves...)

However, even in the PCM1796 datasheet (page 13 I think), we can see the output FFT of the device, where the noise floor is down at -155dB. Even with 24bit data, we couldn't get a signal down nearer to the noise. Offering a 32bit interface means we can get real output (at some frequencies) down near the 25th, and in some cases the 26th bit.

There are those who will argue that we can't even hear down to those silent levels... But some will claim that things like data truncation destroy sound, even at low levels.

Like I said, 50% marketing mojo, 50% real engineering.

Most importantly, it's better than the AKM effort  :D hehehhe

/R
 
playboss said:
user coefficients for DF would ve been soo nice  , maybe in next rev. ?  :)

Not bloody likely.

The digital filter already takes up a significant chunk of silicon, and moving from fixed multipliers (effectively an 1-input function) to configurable taps/weights would eat much more die area. There are not enough users that care to make it worthwhile.

If you insist on playing with the filter, put the PCM179x in external DF mode and supply your own DSP/FPGA.

JDB.
 
Rochey said:
However, even in the PCM1796 datasheet (page 13 I think), we can see the output FFT of the device, where the noise floor is down at -155dB. Even with 24bit data, we couldn't get a signal down nearer to the noise.

...so you're saying I couldn't generate a 24-bit stream/file with a single -150dB sine poking out of the noise floor (ie: below the theoretical 144dB limit of 24-bit sampled data)? Wanna bet?

JDB.
[if that were true, GPS wouldn't work]
 
A bit off topic but:

Say there is an industry expansion to 32 bits, most of the guys using it are going to 'squash to shite' the life out of everything anyway.

I frequently get to listen to first generation multitrack masters from the 60's.  Stuff from Capital, Trident, Sound Techniques... 
Usually, when you start opening the faders, not much has been squashed to within an inch of its life, maybe a vocal or bass will have a bit of compression on it.  We're talking 66dB available dynamic range on these tapes, no noise reduction.

All of it sounds terrible.      NOT!

I'm not anti-digi by any means but... just sayin'

 
Hello Winston,

There's a lot to be said about an "audible" noisefloor making music easier on the ears. I'm no expert, but I think it's one of the reasons why people like to dither a signal when they move from 24bit to 16bit, rather than just truncate. (i'm a little out of depth at 9.30 in the morning on this one though)

That, and the talent level with these guys was so much higher than modern musicians. While folks these days will quantise the living sh*t out any timing, then squeeze all the dynamic range they can out of the signal too.

It's sad really. The only real music left with any dynamic range is classical... and even there, it's becoming a challenge to leave the music untouched.
If I ever see/hear of a conductor using a click track, i may go nuts. (i'm sure there are some examples!!!)
 
Rochey said:
I'm no expert, but I think it's one of the reasons why people like to dither a signal when they move from 24bit to 16bit, rather than just truncate.

No.

Dithering (adding noise with a triangular PDF before truncating) actually prevents the distortion products of a quantizer by smoothing out its staircase-shaped transfer function. Adding equivalent noise after truncating's no use, BTW, as by then the damage has been done.

Rochey said:
(i'm a little out of depth at 9.30 in the morning on this one though)

Ah, posting before coffee  ;) And it's a Monday...

Rochey said:
If I ever see/hear of a conductor using a click track, i may go nuts. (i'm sure there are some examples!!!)

A few years back I went to see Bugs Bunny on Broadway in the Sydney Opera House. The conductor was wearing headphones (with a click track or other cues I don't know) to keep the orchestral music in sync with the cartoons on the big screen. Does that count ?

JDB.
 
Also, perhaps not many of us are fans of hybrid electroacoustic music, but there also, conductors often have to have clicks to synchronize.

Aaanyway,

Perhaps I was too skeptical about TI giving very specific info, but now I really crossed the line. I plain and simply wanted to know how much signal voltage could be tolerated at  PCM179x outputs if I wanted to try passive I/V conversion. I got a nice but short reply:

Anyway please note that the PCM1794A output stage has been designed (the internal equivalent circuit of Iout is a kind of floating source) to drive an active I/V converter. We provide in the datasheet design recommendations for the I/V stage in order to give proper guidance to users of this part.

We do not recommend passive I/V conversion and do not provide design recommendation for this.

I understand the policy. And surely I know an active I/V is supposed to be used, but the audiophool community does it passive anyway. I've seen distortion measurements for small I/V resistance, and everything looked good.  But I really would like to have some informed guesses about the behavior of the output stage when certain size of I/V resistor is used. Anyone?
 
Rochey said:
Hello Winston,

There's a lot to be said about an "audible" noisefloor making music easier on the ears. I'm no expert, but I think it's one of the reasons why people like to dither a signal when they move from 24bit to 16bit, rather than just truncate. (i'm a little out of depth at 9.30 in the morning on this one though)
As already pointed out, dither actually adds resolution. Think of it like operating the bottom bit as a PWM to duty cycle average lower than LSB amplitudes.

Regarding psycho-acoustics surrounding the noise floor, i recall hearing about studies where listeners thought they perceived more HF content when hiss was present with music, but that was decades ago so I don't recall details from that study.
That, and the talent level with these guys was so much higher than modern musicians. While folks these days will quantise the living sh*t out any timing, then squeeze all the dynamic range they can out of the signal too.

It's sad really. The only real music left with any dynamic range is classical... and even there, it's becoming a challenge to leave the music untouched.
If I ever see/hear of a conductor using a click track, i may go nuts. (i'm sure there are some examples!!!)

Electronic music potentially could be more dynamic than actual classical music. Distant micing (like M/S or simple stereo pair) of a classical orchestra is naturally compressed by multiple diffuse sources and room reverberation. Close micing of solo orchestra instruments actually delivers unnatural dynamics not experienced by the concert hall audience. 

JR

 
Back
Top