XMOS-based Asynchronous USB to I2S interface

[Wolfsin]

Finally, a glimmer of understanding. Thanks. So capturing the output of my Oppo i2s pcm of an SACD really provides all the information available?

Mow I would like to cut through what for me has been confusion. I have the simple choice of providing a small amount of power to the GMR, achieving isolation, but at a cost of added jitter or powering the entire WaveIO and using the non-isolated but lower jitter output. The second alternative seems superior.

Without redesign is it correct that those are my choices?

[Lorien]

What is the deal with driver updates, will thesycon provide these or how is the
"next" operating system supported?

Drivers for WaveIO card are customized according to its USB VID/PID and other specific informations. If a new release is available from Thesycon then I'll customize the drivers and make it available for use.

Is there anything in your hardware that limits the ability to apply the firmware changes

unfortunately it will not work if you have only the executables due to few pinout switches in my application. Should I reconsider this and make this board fully compatible with XMOS design?
Regards,
L

[Wolfsin]

Forget my WHQL notes.

The point of WHQL was to allow developers to keep their source code secret while avoiding support issues for legacy components. The choices you make should have much to do with your plans for WaveIO as a product and how much support you foresee providing in the future.

[rsdio]

Originally Posted by Wolfsin

So capturing the output of my Oppo i2s pcm of an SACD really provides all the information available?

Theoretically, yes. However, keep in mind that the Oppo may be applying digital filtering in the process of conversion from SACD/DSD to PCM, so it is not necessarily a lossless translation in all implementations.

Personally, I see the whole SACD process - from the start of recording to final playback - as more severely limiting than PCM. Thus, I would not worry too terribly much about lossless translation. Granted, many great A/D chips are sigma-delta at heart, and thus they are translating to PCM on the fly. You do get whatever 'coloring' is imparted by the particular A/D, but we are already accustomed to thinking of the A/D as an imperfect piece of the system. Using the Oppo PCM output just means that you've combined a little coloration from the original sigma-delta DSD A/D chip as well as any digital filtering from the Oppo PCM translation. It's not necessarily worse than any other PCM A/D, it's just a hybrid; and the key is that another SACD player with digital outputs might give a slightly different stream of data.

[Wolfsin]

@rsdio Thanks. Now I understand why going from DSD to PCM workz but 'tutherWay doesn't.

WaveIO is perfect for diy applications where the same power supply can be used for it and the DAC. If this is not the case please tell me not to proceed as planned. It seems to me that the key to the common power supply may come down to a single set of rectifiers and multiple regulators for the case where 3.3, 5 digital, 5 analog, . . . are required. An eBay seller (item #270567301199) provides a modular solution for providing multiple different regulators with a common ground. Connection to WM8740(s) almost doesn't require more than a tiny adapter pcb with bypass caps and possibly ferrite beads if you feed the analog outputs into, say, a BAL-BAL 'wire'. I will test this conjecture soon after wiring an Opus, then Opii

I am assuming that having the frequency pins broken out allows me to make adjustments to the DAC registers (if required) using i2c. Setup is easy once you have thrice (re)read the email, cleared away pages of jitter posts from your brain, and thought about how it fits into your system.

[Wolfsin]

Wolfsin's test plan was broken. The DDE3 expects 16 bit words and it is his guess (given that it will not synch) that the word length is wrong or a signal is inverted. Anybody tried the AudioAlchemy yet?

[rsdio]

Originally Posted by Wolfsin

@rsdio Thanks. Now I understand why going from DSD to PCM workz but 'tutherWay doesn't.

I really hope my comments are somewhat on topic, and in any case I'll try to be brief.

Going from PCM to DSD works just fine. In fact, some PCM DAC chips use this technique internally without really making it apparent. A DSP or FPGA can simply run the DSD output at a much higher rate than the PCM (but DSD is always at a ridiculously high rate), and basically increment or decrement a register until it matches the incoming PCM sample value. The output of the DSD converter is a single bit at each super clock saying whether an increment or decrement was done.

Granted, PCM to DSD requires a 6 dB/octave lowpass across the entire signal bandwidth (so, at a maximum, 20 Hz and higher, if not then starting even lower than 20 Hz), but this can be done as a digital filter, and in any case it's exactly the same requirement as in the analog stage of a DSD ADC.

So, I guess it depends upon how you define 'workz' - if the lowpass is acceptable then PCM to DSD is rather straightforward. Keep in mind that SACD and DSD grew out of a period in the industry when nearly all A/D and D/A was handled internally as sigma-delta. Sony simply asked "why convert to PCM and store that when you can just store the raw bits from the converter?" Trouble is, converter technology has moved on since that window of time when everything was pure sigma-delta, and now DSD is stuck with one size hammer.

[hochopeper]

Originally Posted by rsdio

I really hope my comments are somewhat on topic, and in any case I'll try to be brief.

Depends how strictly you apply 'on topic'; regardless I have found them a great read, thank you!

Originally Posted by Lorien

unfortunately it will not work if you have only the executables due to few pinout switches in my application. Should I reconsider this and make this board fully compatible with XMOS design?
Regards,
L

The design choices are all yours, I was just wondering what future upgrades to the firmware might be possible. DSD is way down my personal list of priorities in any case. Reading between the lines you've made these changes based on a need to achieve a better layout. I would certainly prefer not to compromise the layout to add a feature that may well not be used very often, if at all.

I am now in the queue for a WaveIO of my own too!

[Wolfsin]

I took a look at the software architecture doc on the XMOS site and have yet to recover. The options list is not short.

[regal]

Originally Posted by Wolfsin

Wolfsin's test plan was broken. The DDE3 expects 16 bit words and it is his guess (given that it will not synch) that the word length is wrong or a signal is inverted. Anybody tried the AudioAlchemy yet?

I am pretty sure that bitclock is inverted I know one of them is inverted, I can send you a schematic for my DDE DAC which shows all the glue logic required to make its "i2s" input compatible with its PMD100, its rather amusing. I am not sure how it would handle the 64bit words as the thing runs at 256fs. Honestly though not sure why you are messing with AA I2s, their implementation is very poor and archaic to todays understanding of how to deal with jitter.