Modular multi channel DIY AD/DA Box

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Hey, I don't wanted to supress ideas. Ideas are always welcome. I just wanted to give it a more serious direction by having a clear 'must-fulfil'-list. :)
Thanks JDB!
The DDS approache looks interesting and I'll definitly put it on my 'projects-in-the-future'-list. But for now I think an AD9912 is not the way to go. Not only because of the package. On the internet you can find tips how to solder it at home. But I think it is simply to expensive! We are talking about 88Eur only for the chip. In my university time we always made projects with the best performing stuff but then later in my job as an engineer I learned, that you also have to keep in mind: Is there a market for it or rather are there people who want to spent that amount of money for it. And I think that there are not many DIYer who want to spent ~150Eur only for the parts for the clocking (Maybe I'm wrong). Therefore, I think now it is better to do it in the following way: First make a board with a VCXO driven clock on it providing the clocks either in master mode or locked by a PLL which is only catching fixed rates like 44.1,48,88.2,96,... and not to support varispeed (as JDB mentioned) and later to make a board with DDS on it and supporting then varispeed. But this board will be designed a little bit more complex and will be a project for advanced DIYer. But I can imagine that this board is then also able to work as a standalone master clock like iClock or Big Ben (not thinking about video sync at this moment) because then it is worth the effort and the money because you will get more features.

Raphael
 
That's it right there Raphael. A stable, low-jitter master clock which works at the common fixed rates. I believe that is a good fit for this project. We can get all fancy and DDS later (not trying to be disrespectful). Simple, reliable and stable. As we all know, some guys will be able to whack this out and some will have trouble. I can do surface-mount surgery in my sleep. I can solder TSSOP's with a fine-point iron tip. LFCSP's are a pain-in-the-ass. I would skip those. I could solder those too but what's the point. And then there's the cost you mentioned. 88Eur IC's? I'll pass on that. DW.
 
Here is a drawing of the PLL. Please note that it is only showing the concept. Many parts are not in at this moment, like input/output and control circuits. I've tried to keep the gate count low for the clock signals but also to use widely available parts. Especially for the output select matrix we could use another part instead of the 125's. I'll have to insert a multiplexer between the VCXO and the divider to support other clock rates like 44.1 or an externally applied superclock. Depending on that the output matrix logic will be a little bit different. As soon as I know all the needed states I can probably simplify the control logic via some boolean algebra.

Raphael
 

Attachments

  • vcxopll.pdf
    52.1 KB · Views: 102
rkn80 said:
Please note that it is only showing the concept.

Even so a few random notes:

- 12 bits is borderline. The usual approach is to implement a software sigma/delta-modulator and feed its output to a single/multibit DAC with a LPF after it. Keeping this differential helps in rejecting noise; you could consider a discrete DAC with a bunch of 4053-type analog switches

- turn off the power from unused VCXOs. Similarly, digital MUXes have non-infinite off-channel rejection; it helps a lot if deselected inputs are grounded (and yes I realize that that effectively implements a double MUX).

- 74AC(T) logic is notorious for producing lots of noise on supply(/ground) lines. Consider using a different family like AHC (if fast enough) or LVC.

- I'm not sure your PFD is race-free. Why not use the standard design with internal (full-speed) clock on D and external WC on CLK, with a second DFF as synchronizer? This gives effectively a phase-detector; frequency detection (and rapid capture) can be done with one of the ATMEGA's internal timers.

JDB.
 
jdbakker said:
- 12 bits is borderline. The usual approach is to implement a software sigma/delta-modulator and feed its output to a single/multibit DAC with a LPF after it. Keeping this differential helps in rejecting noise; you could consider a discrete DAC with a bunch of 4053-type analog switches

Well, I wanted to keep is simple. Therefore, I decided to use this parallel DAC because it is simple and cheap.

Turning off unused VCXOs is on my TODO list. You see that the enable pin is unconnected...
Yes the ACT is not the correct choice. Therefore, you see the note in the schematic regarding the HCT family. I simply used what I had in my component lib.

jdbakker said:
- I'm not sure your PFD is race-free. Why not use the standard design with internal (full-speed) clock on D and external WC on CLK, with a second DFF as synchronizer? This gives effectively a phase-detector; frequency detection (and rapid capture) can be done with one of the ATMEGA's internal timers.
I'm not sure if I understood what you mean. Perhaps you have a drawing of it? The design I used is the standard design for a PFD for both situations where f1 > f2 and f1<f2 (sign memory). The capture range is only limited by the adjust range of the VCXO and only phase offset between 0..2pi can be adjusted.

Raphael
 
rkn80 said:
jdbakker said:
- 12 bits is borderline. The usual approach is to implement a software sigma/delta-modulator and feed its output to a single/multibit DAC with a LPF after it. Keeping this differential helps in rejecting noise; you could consider a discrete DAC with a bunch of 4053-type analog switches

Well, I wanted to keep is simple. Therefore, I decided to use this parallel DAC because it is simple and cheap.

Pity it eats so many I/O pins, that makes it hard to connect multiple DACs for, say, two VCXOs and a LC VFO. I'd consider something like the MCP4922. (For extra fun, hang a low-grade audio DAC on the AVR's SPI port)

If you must use this DAC, at least include a filter at its output (referred to AGND).

rkn80 said:
Turning off unused VCXOs is on my TODO list. You see that the enable pin is unconnected...

For best noise performance the power needs to be removed from the oscillator, as the enable pin only switches the internal output buffer.

JDB.
 
Because I got some emails regarding the next steps of the project here is a general roadmap after the ADC group buy is done:

- GroupBuy of the DAC
- Release of the AES in/out board with/without ASRC on board
- Release of the the clocking module (currently designing)

Ideas for further modules:
- stand-alone ASRC module
- ADAT interface with ASRC on board
- helpful stuff like bar graphs etc.
- other interfaces like firewire

Any additional ideas are welcome!

Raphael
 
This page is 17 pages now!!
Interesting discussion but I will have to start at page 1 and read everything I guess!!  ;D

But a short search on digikey site for the AD9912 dds shows not available and no price for it, which generaly is bad news.... then there is also an ''end of life alert''  which means there is a replacement comming soon, or a plain abandon of the product by AD  :-[

There should be a replacement solution.
 
kibadachi said:
Is there a frontpanel or case design available?

Not yet. I think it would be helpful to have something like bargraphs etc. on it. The ADC/DAC board don't have a bargraph driver or something like that. Only the ADC has a clip signaling pin that can be used.
Feel free to make a frontpanel or case design. :)
 
I want to inform you that I've just ordered a prototype PCB for the AES output board that can be used standalone with the ADC board (no need of programming a microcontroller or something like that...).

Raphael
 
blue_luke said:
[...] then there is also an ''end of life alert''  which means there is a replacement comming soon, or a plain abandon of the product by AD  :-[

Check again. It's the AD9912 (first revision silicon) that's on Lifetime Buy; the AD9912A is still In Production.

JDB.
[looks like ADI is transitioning from the old to the new rev right now, leading to a parts shortage. Not that it's very likely that a DIY DDS clock board will appear very soon]
 
sorry for my ignorance but if im going to put the money and effort into this project i'll have to ask some newbie questions. i understand i will need some type of interface to connect this to my pc sound card. but what do in need on the input end to connect an xlr out from my preamps to the a/d converter? im sure its been discussed but can some one tell me the bitrate this a/d will give me? sample rate?  if its variable what is the range of this parameters?
 
Good questions seavote. Try to go back and read the thread. Raphael is designing modular circuit boards. One for ADC, one for DAC, another for spdif/AES input and output and so on. The first board he is making is the ADC board. It is based on the Texas Instruments / Burr-Brown PCM4204 IC. Go to the TI website and download the datasheet. That device will do all the common audio sampling rates. 44.1, 48, 88.2, 96 and so on. The ADC board does not have a digital output section on it. That will come on one of the next boards. So you would have to home-brew the digital output circuit right now. That's what I plan to do. The PCM4204 works up to 24-bit but can most likely be put into 16-bit mode too. I would have to re-check the datasheet. And on the analog input side, Raphael's board has circuitry to take the balanced analog signals and send them directly into the ADC's. He has two opamps per channel to do this. So all you would really need to add is XLR or 1/4" connectors for that portion. I've written all of this from memory so if I have made any mistakes here, other guys will correct me. But this is pretty close to accurate. This is not a really simple project. I commend you for being interested and asking questions at the beginning. But this is a fairly complex project. One last thing.... the PCM4204 is four ADC channels in one package. So 4 analog signals can be converted. And each digital output stream consists of 2 encoded analog channels. So the board will ultimately have two digital outputs going spdif or AES or TOSLINK. But again, this ADC board only has I2S digital outputs which need to be converted to spdif or AES or TOSLINK.
 
On way to connect this A/D to PC for an example would be to use something like TI C5515 eZdsp USB dongle kit ($79 or something, ordered mine from Digikey for 65€ inc. postage which was zero):

http://focus.ti.com/docs/toolsw/folders/print/tmdx5515ezdsp.html

This is manufactured and supported by Spectrum Digital:

http://support.spectrumdigital.com/boards/usbstk5515/reva/
http://support.spectrumdigital.com/boards/usbstk5515/reva/files/usbstk5515_TechRef_RevA.pdf

Quite a bit features for so low a price. However, Windows does not support USB Class Audio 2.0 but Apple does, guess Linux as well.  Let's hope that Microsoft soon adds support for USB Audio 2.0 to it's Windows products.

There is also the XMOS USB Audio 2 Reference Design:

https://www.xmos.com/products/development-kits/usbaudio2

You interface these boards using I2S (slave mode only with the TI kit, don't know about the XMOS). So ADAT is not the only way though I guess that's what most are waiting for.
 
mhelein  not sure if your reply was to my post. im in a bit over my head with this project but its my understanding that the devices you referenced are used to connect the digital signal from the A/D board to a PC. cool they are great options. though i would need one for each channel of audio,no?
my post was asking how i get my audio out of a mic pre with an xlr connector into the A/D. i see no xlr audio input on the A/D converter.
 
I am super keen to learn from this project and thank Raphael a lot for it.

Like Seavote however, I am a bit of a featherweight when it comes to this stuff and admit that I don't understand too much either.

Personally if someone was prepared to create an "RKN AD/DA for Dummies" little post or page explanation in its most basic form I would be chuffed. Even if it was something like:

1) "xlr inputs will go here",
2) "you need another card to talk the computer and the AD and DA boards and that card needs to go between here and here" etc.

Perhaps even a little diagram for extra stupid people like me.

It does beg the question that if I am this stupid perhaps I should not be embarking on such a project. In reality - and I suspect there are a few people on this board in the same boat, -  I am a little bored with my own knowledge of preamps and comps which I feel I have mastered to the level that I am happy with. Now I would love to learn something new and learning is why I am a member of this board.  This project seems like a really interesting new journey for me to embark upon and into new territory - digital electronics.  I am super keen to learn.

So with this being said, if anyone wanted to create a little post explaining this I would be so appreciative.

Kindest
Deuce
 
Perhaps I can do a basic building instruction the next days. But some remarks already here:

The XLRs can go into X102: Hot on IN+, Cold on IN-, Ground to Ground. Note: The inputs are DC coupled. Depending on your studio stuff and on your preferences it might be advisable to put either two 10u electrolytic caps or a transformer between XLR and X102.

Connectivity to a PC: That will become much easier with the AESOUT boards I've designed and I'm currently waiting for. This board has exactly the same pin layout for the I2S interface like the ADC (X103) so that you can connect them with a simple flatband ribbon cable. The AESOUT gives you 2 AES signals (=> 2 Stereo/4 Mono channels) on either  BNC connectors or on a pin header that you can connect it to any connector type you want.

Raphael
 
On page 2 of this thread is the PCM4104_module.jpg image. That is the schematic for the DAC board.

On page 10 of this thread is the DIX4192 board image and schematic. The DIX is the digital interface transceiver. A transceiver transmits and receives. That board will transmit and receive AES or spdif signals to and from a sound card. This is one way to get the digital audio into your PC and out of your PC.

Also on page 10 of this thread is the ADCmodule.pdf. That is the schematic for the ADC board. Look way over on the left hand side of that schematic. See the nodes labelled IN+:1, IN-:1, IN+:2, IN-:2......? Those are the balanced inputs. You would connect your mic pre output signal to those nodes. Balanced out from one mic pre goes to IN+:1 and IN-:1. That's one channel right there. Same for 2, 3 and 4. The PCM4204 is a four-channel device. It has four completely separate converters inside that one chip.

Go grab those three documents and save them to your PC and look at them closely. RAPHAEL reserves the right to change his designs at any time. Those drawings and schematics are preliminary and have probably changed already. But once we have etch from Raphael, I'm sure he'll provide us with updated schematics. DW.
 

Latest posts

Back
Top