GodDAC - 192/24 USB DAC

[BizonGod]

Hello!
I'm going to start new open project - GodDAC.
Some description of hardware:
- USB 2.0 High Speed - up to 480Mbps
- DAC - WM8740 - up to 24bit / 192kHz
- ADC - WM8785 - up to 24bit / 192kHz
- fully differential output and input stages
- differential and single ended inputs and outputs
- fully powered by USB
- CPU - ARM9 with 2x16kB cache, 192kB RAM, 2 I2S Tx and 2 I2s Rx,
internal clock generator (clock error < 0.1Hz), capable of boot over USB, etc.,
also can be connected: SDRAM, NAND memory, SPI flash, price < 7EUR

If You want some more specification, please ask

I'm looking for programmers with experience in ARM9 C, USB devices etc.

I have in plans to make few first pcs. of hardware for tests.
If You have experience in programming, send me You short summary.
I'm planning to give for free 2 or 3 pcs. to the best programmers to
start development of software in short time.

Regards
BizonGod

[elingeniero]

Ditch the ADC. We're audiophiles, not recording engineers.

[elingeniero]

Ah, and one other thing. Boost the amount of RAM, and add enough I/O (at least a serial port and an Ethernet card, and preferably a Mini-PCIe slot as well) to turn the thing into a computer in its own right

[BizonGod]

IMO ADC should stay. It enables measuring of DIY amps, etc.
I hope code will be < 96kB and buffer < 96kB, so all can fit ARM9 internal RAM -
as I see in examples, it should.
Addinationaly I'm thinking about SPDIF Tx and Rx. How about this?

[chaparK]

Originally Posted by BizonGod

Hello!
I'm going to start new open project - GodDAC.
Some description of hardware:
- USB 2.0 High Speed - up to 480Mbps
- DAC - WM8740 - up to 24bit / 192kHz
- ADC - WM8785 - up to 24bit / 192kHz
- fully differential output and input stages
- differential and single ended inputs and outputs
- fully powered by USB
- CPU - ARM9 with 2x16kB cache, 192kB RAM, 2 I2S Tx and 2 I2s Rx,
internal clock generator (clock error < 0.1Hz), capable of boot over USB, etc.,
also can be connected: SDRAM, NAND memory, SPI flash, price < 7EUR

If You want some more specification, please ask

I'm looking for programmers with experience in ARM9 C, USB devices etc.

I have in plans to make few first pcs. of hardware for tests.
If You have experience in programming, send me You short summary.
I'm planning to give for free 2 or 3 pcs. to the best programmers to
start development of software in short time.

Regards
BizonGod

Hi BizonGod,

Your project sounds cool

How about the host programming? What are the supported OS? Who will make the drivers?

Regards,

chaparK

[elingeniero]

Originally Posted by BizonGod

IMO ADC should stay. It enables measuring of DIY amps, etc.
I hope code will be < 96kB and buffer < 96kB, so all can fit ARM9 internal RAM - as I see in examples, it should.
Additionally I'm thinking about SPDIF Tx and Rx. How about this?

For programming the TAS1020, a $1.50 Arduino chip would do just fine; the ARM9 would be like fishing with dynamite ... what is it you are going to to with all that computing power, other than do software control on the DAC?

OTOH, something like the ARM9 with 512K RAM, using Ethernet as a transport (no USB, no S/PDIF, no TOSLink), and passing data back and forth using some kind of standard, already existing (therefore already coded) TCP protocol, would kill:

* all jitter issues can be handled at the DAC
* super CPU low utilization on the host
* At 500mHz, there's gobs of CPU left on the ARM9 for DSP: codecs, DRC, parametric EQ, compression, volume normalization, compander, resampling, what have you

I would want to avoid a unit that had everything all on one PCB, because hacking tightly integrated all in one boards is a pain in the ***. So keep
power supplies, ARM9, line stage and DAC/ADC circuits on separate boards (so builders can shield the DAC/ADC). At least Have the ARM9 on its own board, and set things up so that the other circuits are well separated and have well defined places where builders can plug in their own stuff.

2 other thoughts:

1. It might be interesting to set things up so that builders can use multiple DACs per channel, like some of the NOS 1541a boards are doing, where they stack the DIP packages. Maybe put the DAC chips on teeny-weeny edge mounted PCBs that can be put equidistant from the ouput connector(s). Enough of them, and op amps are unecessary.

2. The 8741 and 8742 DACs are pin compatible, and they do dual differential output just like the 8740. I like that the 8741 and 8742 have built in software volume control in .125 DB increments. I'd rather have an 8741 + Arduino than 8740 + ARM9, if you aren't going to do anything more than control the DAC with the ARM9.

3. Keep the firmware on a SD or CF card, so you can upgrade the firmware by putting a new image on it with a card reader, and plugging it back into the DAC.

[BizonGod]

This device will be using USB Audio Class, so it will use drivers delivered with any OS.

Yesterday I have added also new hardware feature - adaptative clock - using VCXO it
is possible to match perfectly sampling rate regardless ie. temperature drift etc.

Regards
BizonGod

[elingeniero]

How about using the Tenor 7022L chip for USB. I dunno if there's a USB audio class 2.0, but but at least you can have 24/96 USB audio class 1.1 without much programming.

[analog_sa]

Originally Posted by BizonGod

This device will be using USB Audio Class, so it will use drivers delivered with any OS.

What will the programming team do then? Fight USB jitter with prayers?

[chaparK]

Originally Posted by analog_sa

What will the programming team do then? Fight USB jitter with prayers?

Well, i think that the plan is to reconstruct a master clock from the USB stream.

The project looks interesting, i'm ok to take part to the programming burden.