The DAC development platform

[peufeu]

While I wait for reviews on the FPGA module that will be included in my next DAC, I am thinking about the next step in the project.

The idea is to get a DAC development platform, which which will allow anyone willing to participate, to build modules, exchange them (by mail) via other participants, and share listening results. The purpose of this project is to make some sort of standard pinout for little DAC or I/V daughterboards, so that all of you people can exchange little modules.

This will allow comparison of various DAC chips or I/V schemes, all other things being equal. It should be interesting...

This "platform" will be a board, with several connectors on it, to plug modules. It will have an on-board Tent clock, supply decoupling, audio outputs, and some power supply regulators. It should not have too much stuff, because ideally everything should be a plugin, to allow for more experimentation.

It will have its own clock to eliminate jitter issues and the issue of everyone having a different digital source. Therefore, it will have a clock output and an I2S input for slaving the source. I am doing this for my Ethernet audio project, so the board will connect to the Ethernet FPGA module. It will also have a slot for an ASRC and a SPDIF input, so anyone will be able to experiment with it provided they have a SPDIF source at hand.

The modules I am thinking about are :

- SPDIF Input and ASRC
- Digital Filter
- DAC
- I/V
- Balanced to Unbalanced (for balanced-out DACs)
- Lowpass filter (if needed)
- Volume control
- Output buffer

So basically this is just a board with 8 connectors, a power supply input, a clock, decoupling, and some RCAs for audio output. Modules will be daughterboards plugged in at a 90° angle. The board itself is really nothing special, the important thing is that everyone works with coherent pinouts to be able to share modules.

So :

- Who is in ?
- Who has other ideas for other modules ?
- Who can suggest a cheap, and practical connector solution ?

I was looking at board-edge connectors. This would make module boards very simple, having simply an their edge pushed in a female connector on the main board. Leveraging the economies of scale of the PC platform is a good idea : PCI and PCI-Express connectors are quite cheap. The drawbacks of this approach is that the width of a module board is fixed, some connectors need a lot of holes in the PCB (PCI has too many pins) and PCI-Express needs the module boards to be milled to shape.

I an thinking about using 90° Sub-D connectors. These are dirt-cheap and work well. The main board would have straight female connectors and the modules would have 90° male connectors. There is no restriction on module board size.

WHat do you think ?

[Carl_Huff]

Count me in. I have been interested in doing a project like this for quite some time but have never got around to it.

Best Regards,
Carl Huff

[analog_sa]

Great concept. And at the right time - i've just been scratching my head over where to test an AD1955. Such a unified test bed will greatly help in comparisons and development.

An unwanted ASRC module, per example, can just get replaced with a shorting plug.

DB9 connectors are probably all that is required. They are low-tech but cheap, easily available and most importantly allow for improvised prototyping plug-in modules.

I would like to see the option for multiple independent power supplies, rather than a shared raw dc. At least one single PS per module with local regulation on the daughter boards. Higher performance regulation (Jung) may also be available on plug-in modules for the analogue stages. Not sure if this is practical.

As many won't have use for spdif inputs a USB input module should be easy to fit in place of spdif. This means the USB/SPDIF slot needs to be positioned appropriately regarding access to the port.

[pentode1]

I have just built a Audio Note Dac1 kit and plan to a few experiments

a) using I2S rather than Spdif. (See my I2S thread)
I have located the I2S signals in an Arcam Cd72's Sony drive mechanism and will use an Atmel At Tiny44 instead of glue logic to
'convert' the 24bit drive output to 18bits for the dac.

b) I also plan to parallel 4 AD1865s* to increase the output current by piggy backing them. Though it will require mods to the onboard 5v regulators.

*These are still available from Digikey for about £10

I feel the Dac1.1 kit is a good basis for experiementing - penty of room in the enclosure, valve output stage with scope for upgrading, dac pcb with socketed CD8416 and AD1865, Audio Note interstage transformers are available for £70 for a further upgrade (couple the dac out to valve input)

This is my first foray into digital audio but from the detail and clarity improvement already provided by the Dac1, I'm 'hooked'!!
P

[dddac]

Nice idea, I wanted to do this years ago (connectors are still in the drawer )

I would vote for:

removing the volume control and output buffer. This has little to do with the idea itself

adding USB input module
adding power supply module

You should "design" first a BUS "Standard" where all possible signals and power supplies are present, so every one can design any module according to this standard.....

If everyone would except this standard, every one can build a masterboard and modules ....

AND, you need to set up a seperate WEB Site for this, otherwise it will run dead for sure...

[Luke]

Love your work guys

[AR2]

This is the greatest idea I have seen around here. I am in!

[koolscooby]

I'm totally into this one, I've been thinking about doing something like this for the past couple months... Here are a couple ideas I have for something like this...

I think there should be at least 2 board-board connectors; 1 for power, 1 for signal.

I think each board should have a (3?)-bit ID which could be set via DIP switches or read serially via an on board eeprom. Any board could send or receive I2S audio from the base board for distribution to other boards... For my personal application, i'd like to have enough room to carry around 3 sets of I2S so that 5.1 could be achieved... daughter boards that use the I2S as input should have DIP swtiches to select which channels to use for whatever output method is going on (IE: headphone amplifier.)

I don't think the PSU should be a module, but part of the base main board... or have alternative inputs for PSU... so that the base is more of a platform [which includes the PSU]

I thought of making this a vertically stackable system -- each board has a top and bottom connector for expansion. The bottom board would be PSU (PSU caps would be on the bottom of the first board, with standoffs to a case.) The second board would be the FPGA/logic-level stuff that includes an LCD display, rotary encoder and a couple buttons. The third board and beyond would be configurable to taste. I like the vertical method; however, the disadvantage lies in the fact that each board would be required to be within the size of the base.

If more than one board were an input IE: a board that could do USB in, Optical in, and SPDIF in, the board would have 3 ID assignments and would enable or disable tristate buffers to be in high impedance mode to allow one set of inputs flow to the other devices.

The PSU would push around standard levels; 12, 5, 3.3, 1.2-1.8 (for FPGAs) and ground. If a module needed a significant amount of current, or a specific voltage, either cables could provide it via external connector input or through extra pins left on the PSU board-board connectors and specialized powersupply below the logic/low-current powersupply.

These are all just ideas, kicking them around... I might throw together some sketches

[AR2]

I agree with having PSU as part of the main board, but than additional regulators could be part of other boards, so they are close to the circuits.

[peufeu]

I'm excited to see the idea raises enthusiasm !

Keep the good ideas pouring, guys !

Right now I'm swamped... back in a few days