Hello.
THis is a really in-depth technical question. Proabably one of the
original
Slim designers would have to asnwer it.
I've got my DAC going, and it uses the input Fs encoding in the input
data
stream to figure out what the output word clock F should be.
If the Transporter is running off
I've seen now that whenever I change the word clock frequency, the
Transporter is changing the Fs encoding to be all 0's (= 44.1 kHz code
point) for a short period of time!!
THis to me is an error and it's why there is a loop between the DAC
and the Transporter. When the DAC is in auto
My thinking is to have the Fs qualified by counting some number of
same encodings and then changing word clock when that number is seen.
Basically, I'm talking about an encoding debounce circuit.
--
wayne325
wayne325's
wayne325;558021 Wrote:
My thinking is to have the Fs qualified by counting some number of
same encodings and then changing word clock when that number is seen.
Basically, I'm talking about an encoding debounce circuit.
Cool. It worked.
What I did was stop changes to the oscillator output
wayne325;420007 Wrote:
I'm not doing parallel DACs in my current design, although I am using
the
same DAC chips and fully balanced (2 chips per channel). I want to
become
successful with my first foray into this before I get really tricky
and
start laying down FPGAs and have to learn
There's a white paper describing the circuit in a bit more detail as
well:
http://www.wadia.com/technology/technicalpapers/
As I guessed, part of the reason they're doing it this way, is to give
the DAC's more settling time. They're also claiming to get the full SNR
benefit of a correlated
DCtoDaylight;419925 Wrote:
There's a white paper describing the circuit in a bit more detail as
well:
http://www.wadia.com/technology/technicalpapers/
As I guessed, part of the reason they're doing it this way, is to give
the DAC's more settling time. They're also claiming to get the
seanadams;419570 Wrote:
In the normal source-DAC clock via s/pdif scheme yes, but here we are
talking exclusively about a word clock scenario.
Generally it's to not make you have to flip switches, but yes mixed
sample rates within a playlist would require this. In that case it's
Phil Leigh;419579 Wrote:
Sean - now I get it.
But, if the DAC is going to send its clock back to the transport at the
correct rate (and playback of mixed rate sources is required), it's
going to have to either read the status bit from the stream or recover
the clock from the stream and then
seanadams;419572 Wrote:
Huh? No you couldn't. The whole point of the DAC being in word clock
mode is that the source is going to give you the bits at whatever rate
the _DAC_ is running.
Well, actually I'd consider it optimal to not deal with s/pdif and word
clocking at all... this is
So the thing about the 4 DACs is a patent owned by Wadia. It's in their
9xx series devices. What they teach is to have 4 DAC chips per
differential
half of a balanced signal - so that's 8 DAC chips per audio channel; a
stereo
pair has 16 DAC chips. The DAC chips are all connected at their
DCtoDaylight;419811 Wrote:
Ok, I understand now
What they're doing is building a 4x oversampling unit, with a linear
transfer function, in hardware, rather than in software. Seems like a
slightly expensive way to implement it, although it would allow you to
use slower DAC chips than
wayne325;419815 Wrote:
Also it's not linear. Its some sort of a non-linear curve fitter. I
assume a cubic spline -ish math function.
Unless they feed each DAC with a totally different bit stream, then no,
it will be linear. If they do feed each DAC with a different data, then
you no longer
seanadams;419281 Wrote:
The other ones are
1000 = 88.2
1010 = 96
1110 = 192 (not supported by us)
BTW this is a good tool to have if you're doing anything with s/pdif:
http://www.nti-audio.com/Home/Products/Minstruments/DigilyzerDL1/tabid/82/Default.aspx
Excellent, thanks. I'll see
wayne325;419437 Wrote:
Excellent, thanks. I'll see what I can do to add this function to my
DAC.
As you said, it's really the way to go that the DAC will use the
correct
word clock out based on the input S/PDIF signal. Wow - I'll have one
up on the rest of the world (shouldn't last too
Phil Leigh;419438 Wrote:
??? - you want to use the clock out from the DAC to drive the transport
- you don't want to use the SPDIF clock
No, we're talking about using channel status bits (extra data embedded
in the s/pdif stream) so that Transporter can communicate to the DAC
what the
seanadams;419451 Wrote:
No, we're talking about using channel status bits (extra data embedded
in the s/pdif stream) so that Transporter can communicate to the DAC
what the correct clock frequency should be. These bits are already
defined in the s/pdif spec, although I don't think anyone has
Phil Leigh;419513 Wrote:
ah. hmmm...
But don't dacs auto-detect the clock frequency from the recovered spdif
clock? Is this about supporting a mixture of sample rate files?
I expose the DAC could detect the clock F, but why not do it as
intended
with the encoding in the bitstream - it's a
wayne325;419527 Wrote:
About the only thing I'm missing is paralleling, say, 4 DAC
chips and offsetting time samples to the DACs by 90 degrees.
Care to elaborate on this idea? I'm trying to understand the
advantage of doing this, and I'm drawing a complete blank on why you
would want
Phil Leigh;419513 Wrote:
ah. hmmm...
But don't dacs auto-detect the clock frequency from the recovered spdif
clock?
In the normal source-DAC clock via s/pdif scheme yes, but here we are
talking exclusively about a word clock scenario.
Is this about supporting a mixture of sample rate
wayne325;419527 Wrote:
I expose the DAC could detect the clock F, but why not do it as
intended
with the encoding in the bitstream - it's a LOT easier.
Huh? No you couldn't. The whole point of the DAC being in word clock
mode is that the source is going to give you the bits at whatever rate
seanadams;418536 Wrote:
If your DAC is smart enough to decipher the incoming S/PDIF channel
status, you could set the outgoing clock accordingly as TP will indicate
the correct playback frequency in those bits. I'm not aware of any DACs
that actually do this - AFAIK they all require you to
The other ones are
1000 = 88.2
1010 = 96
1110 = 192 (not supported by us)
BTW this is a good tool to have if you're doing anything with s/pdif:
http://www.nti-audio.com/Home/Products/Minstruments/DigilyzerDL1/tabid/82/Default.aspx
--
seanadams
Sean,
THanks for the reply. Wow I was not expecting the Founder of the
company
to answer a question. We have had similar experiences as I was one of
a handful of people who began a startup together in 2002 and that
company
was later purchased (alas, for a song). I am satisfied to see people
wayne325;418425 Wrote:
Fourth, for those who have a T3 alrady and are playing 96/24 files, is
there
a significant difference between those and CD ripped files? (I would
think
so). Along hese lines I downloaded the sampler 96/24 files from
HDtracks
and played thru my Duet. I
Hi all,
I'm building my own DAC and I want to perhaps use a Transporter 3 as
one of
the inputs for purchased music and I want also to rip my CD
collection
so I can get rid of that large cabinet beside my stereo
So... a few questions. I guess these are directed at Logitech
technical
wayne325;418425 Wrote:
Hi all,
I'm building my own DAC and I want to perhaps use a Transporter 3 as
one of
the inputs for purchased music and I want also to rip my CD
collection
so I can get rid of that large cabinet beside my stereo
So... a few questions. I guess these are
Funny, I googled, and found Sean's earlier forum reply as the second
match
The word clock input will accept a signal up to 100Khz of at least 1V
amplitude. It will accept practically any input but a 3V, 44.1KHz
square wave is typical.
The word clock is just that, a clock. The system relies
An ideal word clock source would be simply a 3.3 or 5V logic signal with
a 100 ohm resistor in series. The impedance doesn't matter, that's just
to protect you from a short. You could also put a ferrite in series to
slow the edge, since a sharp transition is not needed or desirable
here.
You
29 matches
Mail list logo