MDAC First Listen (part 00100011)
- Thread starter highfi
- Start date
- Status
I've been thinking a bit on how to do things like these in as "clean" was as possible. By "things like these" I mean DSD ADC interfacing, for example, but changing channels on the fly is virtually the same.
The problem is that each USB (slave) device has to announce its capabilities during initialization, there's no "on the fly" change allowed.
Now, I assume the ADC will normally use the PCM recording UAC2 interface and "report" all supported bit modes and sampling rates to the OS, so I can record at 44.1KHz if I wanted. However what if I switch the MDAC2 to "DSD mode", to record DSD as DoP-packed PCM? The lowest supported rate by the DoP "standard" is 64FS, which is 176.4KHz/16bit. What should MDAC2 do in case I request recording in 44.1KHz?
There's no clean way of saying "I don't support that", because it already (earlier) said that it can. There's also no way for the OS to distinguish between "real" PCM and DoP. Triggering errors / invalid replies from the MDAC2 side could easily kick the device out of the system as faulty. Causing an USB reset, forcing device re-initialization, would work, but could also break the recording application as it's the fact of opening the device for recording 44.1KHz that triggers the problem.
The only "clean" solution known to me is to make DSD a preference possibility and indicate (on the display) whether the recorded PCM data are DSD or real PCM. If some application opens the MDAC2 for 44.1KHz where it has been "set" to DSD, it would fall back to normal PCM.
It's these seemingly nonexistent problems that can creep up anywhere.
Multiple channels may be a problem if the device has already announced only 2ch, but I guess you (or Dominik) could make it so that MDAC2 chaining will be detected on device initialization (power on or USB re-plug), ... multichannel "hot plug" might not be very useful.
JohnW
pfm member
The MDAC2 will always do what its "Commanded" by the OS - and indicate its mode of operation on the OLED display. The MDAC2 will never try to force itself on the OS.
The MDAC2 will announce its capabilities to the OS upon enumeration - if the user changes the configuration then the MDAC2 will simply re-enumerate.
adamdea
You are not a sound quality evaluation device
Ok John, level with us-
a)have ever had the urge deep down to design an entirely proprietary digital crossover active speaker system?
b) do there exist enough bulk foil resisitors *in the world* for a 7.2 m-dac crossover system employing 7 3-way speakers (23 channels)
a)have ever had the urge deep down to design an entirely proprietary digital crossover active speaker system?
b) do there exist enough bulk foil resisitors *in the world* for a 7.2 m-dac crossover system employing 7 3-way speakers (23 channels)
adamdea
You are not a sound quality evaluation device
JohnW
pfm member
Hell no!!! - I'm an ESL fan, nothing else come close... Oh shame about the all important bottom end...
I true HQ Music surround sound system is appealing
Well sadly we can only daisy chain 8 audio channels - lets see the crazy Mega Fusion (60 resistors) x4, thats 240 bulk foils per setup!
Its times like this that I start to wish we where not selling the PCB's "at cost"
fusion5
on the dark side of hifi...
2nd, 3rd & 4th MDAC2's L2 / L3 for multichannel systems can be offered at GBP300 each + PCB "at cost" (so long as the offer is not abused)
Is there any reason for the slaves to be L3 instead of L2, besides standalone use?
How many inputs are left with 3 slave dacs?
JohnW
pfm member
Only for those who require 8 ADC inputs (multi-track studio recording), otherwise the L2 and L3 are identical. So must systems would be L3 for the Master unit (for a pair of ADC inputs) and L2's for the expanded audio channels.
With the current rear panel arrangement with all 3 BNC's dedicated to Digital outputs, 3 Optical inputs remain (and USB), but note SPDIF can only support 2 audio channels (this is not relevant to Active speaker set-ups where the expand channels are just EQ'ed for each driver as you are still processing a "Stereo" input).
MartinC246
Member
John
It's very interesting to read all this technical stuff although most of it is way over my head. Like everyone here I'm really looking forward to the improvements you'll make to an already fantastic DAC.
Can you please confirm that L2 will include ClockLock? The last table of MDAC2 options showed "Wordclock & Video Clock Synchronisation" on L3 only, or is this something else? I don't need ADC but would like the opportunity to clock lock if I had a suitable transport or player. Is optical the best way to do this, as it does limit the transports we could use?
Thanks
Martin
ultraviolet
Member
On the MDAC it defaults to 0dB when I plug my headphones into my Little Dot Headphone Amp. Is there any benefit or point in lowering the dB on the MDAC and raising the volume on the amp, or vice versa, does it affect the sound quality or add to your listening pleasure.
I just had a quick try and it sounded the same. Just wondering.
I just had a quick try and it sounded the same. Just wondering.
I would like to see the clock output via bnc too if possible
JohnW
pfm member
Martin,
I'm not sure why I suggested Wordclock would only be only available on L3 - as it can be implemented on any unit with the advance clock circuit, so thats basically MDAC2 L2 / L3.
Wordclock / Video Synchronisation is important in a studio environment - where the ADC versions would be at home - but it can be included in the L2.
The basic Clock-locking via optical would be included on all versions - at this time I'm not sure if the L1 version makes any sense, I'll see how many MDAC2 owners decide to go for the L2/L3 version once I open the next instalment option - hopefully everyone will atleast go for L2 as this offers the superior Audio quality (and the important DSP option) - to this end L1 makes little sense.
As far as I'm aware, basically all clock-lockable transport (Arcam, DPA, Cambridge audio) employ an optical clock link - the technical reasons is to galvanically / RF isolate the "Analogue" DAC from the Digital source.
There is no reason why the bidirectional "multi purpose" BNC connectors could not be used for Clock I/O, defined "on the fly" by the FPGA firmware but be aware that there are only 3 BNC connectors available on the rear panel.
I like to stress that just as with the Analogue PCB, the final Spec. of the digital PCB is in a state of "Flux" until I actually complete the design - the design just flows with my ideas as I work on the PCB. Making the BNC connectors "Definable" to allow multi channel expansion is just one example of this "natural" progression
Richard Kimber
pfm Member
Yes, that was me. I don't regard that as a stability issue, though. Just a bug - for which there was a work-around. A.10 is perfectly useable.
- Richard.
JohnW
pfm member
Hi Richard,
Thank you jogging my memory
- that agrees with the point I was trying to make, that once you get to grips with the "misbehaviour" there are normally ways to work around it If we encounter a "fatal" problem then we will work on a patch (if there's a solution)
"+1" for multi-purpose BNC connectors, a neat idea if it can be done (ie. via multiplexers).Making the BNC connectors "Definable" to allow multi channel expansion is just one example of this "natural" progression
Clock in, clock out, regular SPDIF, ..., more future uses.
finesensations
pfm Member
I prefer -9dB but if you can't hear any difference no need to worry.
John, my technical knowledge of digital circuitry is very thin, and I wouldn't normally waste your time on design aspects. However, over on Computer Audiophile there has been a lot of experimenting on how to get the best signal from the source to the DAC, and John Swenson has just added something which might be significant here.
He thinks, and he has data to back it up, that "packet noise' is a major problem for most ways of getting data to the DAC receiver chips, including USB, and that design of the power distribution networks in the DAC is going to be crucial in minimising the effects of this noise. I hope I've summarised him correctly; the full spiel is #211 at
http://www.computeraudiophile.com/f...server-step-step-17666/index9.html#post366377
Is this likely to be relevant to MDAC2, or am I really wasting your time?
- Status