[joneeboi]

Presenting the babyDAC
 
*loud applause and other self-congratulations*
 
Since the world hasn't ended, I thought I'd post this DAC up for comments. I had been thinking of doing this for a long time, but school loads are always heavy for me. Four months later, I had a chance to come up with this little number. It's a TI PCM2706(/7?) USB-to-I2S feeding a Wolfson WM8740. I initially tried to squeeze a dual differential configuration on the board, but I thought I'd take a smaller step before tackling that behemoth of a project. It's all pretty straightforward after that.
 
I separated analog and digital grounds, and I went with a simple crystal oscillator for the clock. I left footprints for the box-type caps used in the BantamDAC, and I still have to work out clearance for the two non-corner notches. I also want to rework the mini-USB plug since I just grabbed it from the EAGLE library, but I had bigger issues with getting the two chips to play nicely on the board. I tried to reduce the length of the I2S lines as much as possible (pins 5, 17, 18 and 19 on the PCM), but I think overall the delay is negligible at this size. Also, components are tiny at 0805, cuddling nice and tight since I was a bit paranoid about not having room since I started with a second Wolfson chip on board. Thus, if anything is too unrealistically close, let me know.
 
If you look kinda closely, you'll notice a lot of vias overlapping with SMD pads. I operated with the principle that this is fine and acceptable since I couldn't find any standard stating the opposite, but I wanted to double check that this is alright. Some vias are completely engulfed by pads.
 
Schematic
http://dl.dropbox.com/u/886264/babyDAC/babyDAC_v012_schem.png
 

 

 

 

 
Though the board is cut for a cableDAC-ing, I still want to make use of the Hammond 1455C80x enclosures. That's why there is so much clearance on either side for tall parts. The idea is to eventually pair this with a balanced headphone amplifier with the Carrie's footprint. That's why it's hooked up for bus-powered operation so that the computer will give the USB as much power as possible. I just thought I'd try my hand at DAC design first since I didn't want to let cobaltmute have all of the fun.
 
There's lots of room and time for wiggling and rearranging, so fire those comments my way. Don't hesitate to be specific.

 

[nullstring]

I am very interested in this. I can't comment on the design, but I've been looking for a cheap balanced output DAC with usb interface, and this looks like just the ticket.
 
 
Just a few thoughts-

• Would it be difficult to add I2S input? Just in case someone wants to feed it 24bits from another source?
• I assume elna silmic II's fit in C7-C10?
• Is there a reason you're using WM8740 rather than WM8741 or WM8742? How hard would it be to support all 3?
• Should you add analog ground points near the outputs? It seems like most balanced dacs have this, I'm exactly sure why.

 
 
Concerning differential..
Is dual mono the same thing as differential?
I am wondering if you could possibly integrate your differential design into this.
You could use two dac boards. One would piggyback the I2S off the other.
All you would need to do is add I2S input and output to the boards..... And maybe a switch? (Does dual mono require a setting change?)  Err, I actually don't know enough to say that. But it seems doable to me.
 
I think this might be a better idea than trying to stick two dacs on one board.

 

[joneeboi]

1. I can add through holes for I2S. Should be fairly simple.
 
2. I'm going to revise this and try to maximize the amount of space possible for caps. Since this design is ambitious, I want to make it possible to use good quality caps. What we have now is a worst-case scenario for parts layout. Very rudimentary and early.
 
3. The WM8740/1/2 are pin-compatible. Should be able to drop in, though I'll double check that. (When did the WM8742 come out?!)
 
4. I can add the analog ground hole.
 
For more information on the differential outputs, see the Application Information in the WM8740 datasheet. In other words, I don't know.

 

[cobaltmute]

Just some random comments:
 

1. Take a very good look at your trace routing and your ground planes - the planes are cut up very, very badly.  You may want to consider 4 layer
2. Under the PCM, the vias look like they are either touching or won't pass a DRC.  You could use smaller pads for your vias
3. Is the via near pin one of the WM supposed to be touching that trace going to pin 1?
4. Does the top side notch have enough screw head clearance to not hit the PCM?
5. Will RR- clear the stand-off in the case?
6. I may not be seeing it but where are you tying top and bottom planes together?
7. Some of your parts labels seem weird - X1 for the mini-USB and Q1 for the crystal?  I've usually seen X being a crystal and Q being a transistor.
8. This is a nice SMD crystal that will suit this project.  I have used it and it works fine.
9. I'm not seeing vias on some of the caps around the regulators, but they may be the pics.  I know you've said you're doing via-in-pad, but I still should see the drill hole.
10. Thermal relief is a good thing.
11. Are C3W and C4W actually de-coupling the pins?  They should be right on each pin, not just on the power line coming to the chip.

 
The first board in your pics is not the same as the last three - it makes it hard to judge what to actually comment on.
 
Seems like a decent start.
 
 
 

 

[Gotez]

Hi,

I'm very interested in building this.

I would like to see the ground points near the output. If possible a SPDIF feature would also be very nice.

However, since there a very many small (DIY) usb dacs around you might want to add something 'extra' to the dac, as a unique selling feature. The balanced outputs help alot, but this might not be enough.

Please keep us informed. I'd love to build one. When are you expecting to make the pcb's?

 

[nullstring]

Quote:

Hi,

I'm very interested in building this.

I would like to see the ground points near the output. If possible a SPDIF feature would also be very nice.

However, since there a very many small (DIY) usb dacs around you might want to add something 'extra' to the dac, as a unique selling feature. The balanced outputs help alot, but this might not be enough.

Please keep us informed. I'd love to build one. When are you expecting to make the pcb's?
 

 
I am sure joneeboi can speak for himself.. but...
 

• SPDIF input would never happen. It requires a SPDIF receiver chip and isn't really part of the project goals. Since he will be adding I2S input, you can add your own SPDIF receiver board if necessary.
• I don't think any "something extra"s need to be added. This is a uses a higher class chip than any of the competitors. As a result, it also can't be used with single ended amplifiers without a balanced to SE converter. It's already very special. Besides, what did you have in mind?

 

[Gotez]

 
I am sure joneeboi can speak for himself.. but...
 

• SPDIF input would never happen. It requires a SPDIF receiver chip and isn't really part of the project goals. Since he will be adding I2S input, you can add your own SPDIF receiver board if necessary.
• I don't think any "something extra"s need to be added. This is a uses a higher class chip than any of the competitors. As a result, it also can't be used with single ended amplifiers without a balanced to SE converter. It's already very special. Besides, what did you have in mind?

As I said, I would really like to build this, for these specific reasons you mentioned (high class chip and balanced output). And you are very right, I could add an SPDIF reciever board and use the I2S input. I meant no disrespect and would like to apologize if i caused any. I have great respect for designers, and my own talents in this area are non existent.

But my line of thought was this: people who are going to build this already have a balanced amp. These amps are often larger and more expensive than SE amps. They would also most probably use balanced headphones. The users would therefore probably not want nor need a small USB dac, unless of course they have a small portable balanced amp. If paired with a larger and more expensive balanced desktop amp it would not be a problem to add some features to this dac, like a spdif reciever, of even extra SE output. Size would not be an issue.

Now, I might be wrong, (I probably am) and if so, please disregard my comments. But I only wish to help joneeboi.

 

[cobaltmute]

 
Quote:

 The idea is to eventually pair this with a balanced headphone amplifier with the Carrie's footprint.

Maybe this explains the size?
 
And two problems with doing a SPDIF in this size of board:

• Size of parts in the receiver chips as well as the isolation transformer
• Where are you going to draw power from if not the USB bus?

 
Now, given what joneeboi is trying to do - put a board this size into a Hammond case, I know exactly how I'd try to do it to have both USB and SPDIF in the same case.  Ruins the cable dac version, but that isn't his goal.
 
 
 

 

[nullstring]

Quote:

But my line of thought was this: people who are going to build this already have a balanced amp. These amps are often larger and more expensive than SE amps. They would also most probably use balanced headphones. The users would therefore probably not want nor need a small USB dac, unless of course they have a small portable balanced amp. If paired with a larger and more expensive balanced desktop amp it would not be a problem to add some features to this dac, like a spdif reciever, of even extra SE output. Size would not be an issue.

 
I think this is the point. Trying to break ground in this area.
Having a small balanced amp doesn't really make sense right now because it requires a very large and expensive DAC to source it.
Therefore, he is creating a small balanced DAC to be paired with a small balanced amp.
 
 
"people who are going to build this already have a balanced amp."
Maybe or maybe not. There is a chicken before the egg issue here.
You need both a balanced amp and a balanced DAC. Either one is less useful without the other.
From what I can tell, it's actually much more common to have a balanced DAC running a SE amp than the other way around.
 
I plan on pairing the DAC with this:
http://www.diyaudio.com/forums/headphones/179298-wire-ultra-high-performance-headphone-amplifier-pcbs.html
 
This amp could easily be semi-portable and seems like a great office amp. Pairing it with a small balanced DAC would be the optimal solution.
 
 
"If paired with a larger and more expensive balanced desktop amp it would not be a problem to add some features to this dac, like a spdif reciever, of even extra SE output. Size would not be an issue."
There are already solutions like this. There is little reason to compete with them.
 
There is also little reason to add these features to a single PCB.
Not even other larger DIY DACs do this. Separating them into different boards helps keep the cost of a single component.
Put all the features in one PCB would reduce flexibility and increase the minimum cost of the component.
I don't see the upside. There are already places to source SPDIF receivers and Balanced to SE converters if so inclined.
 
 

 

[joneeboi]

Thanks for the comments, guys. I really appreciate the interest. It's going to take some time before I can even think about producing PCBs.
 
I also appreciate the detailed response, cobalt. I'm going to re-do with the layout with 1206 and a more realistic clearance. I'll try to throw as many components as I can underneath the board, which should make room for decoupling caps and other distance-sensitive components. I didn't use any vias around the regulators. I was wondering what the outer diameter is for the washers at the notches? Also, where did you use that SMD crystal? On the grub, you went with the Crystek, so how does the performance differ? (Will that just be in the grub thread?)
 
I'm not interested in SPDIF. I don't have anything that uses it, and it's not what I want for this DAC. Sorry.

 

[cobaltmute]

Quote:

Originally Posted by joneeboi /img/forum/go_quote.gif
 
I also appreciate the detailed response, cobalt. I'm going to re-do with the layout with 1206 and a more realistic clearance. I'll try to throw as many components as I can underneath the board, which should make room for decoupling caps and other distance-sensitive components. I didn't use any vias around the regulators. I was wondering what the outer diameter is for the washers at the notches? Also, where did you use that SMD crystal? On the grub, you went with the Crystek, so how does the performance differ? (Will that just be in the grub thread?)

 
0805 is very reasonable for a project such as this.  My comment about the vias around the regulators was that it looked like some of the caps were on the bottom, but the regulator was on the top.
 
As for the clearance around the notches, I looked at the sizing of a #2 screw head and used that for my basis as the #2 is the Hammond size for the mounting screws.
 
As for where I used the SMD crystal, I don't post everything I build

.  Using a crystal requires equal traces to the PCM.  For the grubDAC production board, that wasn't practical so the Oscillator was a better choice.  It does affect the sound as well,  For my ears, and it has been a while since I compared a grubDAC proto to a production board, it seemed to provide a cleaner more detailed high end.  If you go searching and I recommend that you do, you can find some other peoples opinions on the matter.

 

[joneeboi]

I have read that clock quality affects the sound, such as on CD players, but so far I've only done a bit of reading on crystals and the Kwak clock. I'll dig deeper.
 
I'm still gonna give it a go with 1206. Despite the higher performance I'm hoping to achieve, I still want to make it accessible and easier to build. That's something I ran into with what you see above. It was basically impossible to build, so the pendulum is swinging the other way. We'll see how far I get with this revision.

 

[nullstring]

Just FYI, I looked into what it would take to go differential.

1. I2S output and I2S input
2. jumper/switch to switch pins 6 and 24 from 0 to 1
3. Pin 6 is the Diff HW Mode.
4. Pin 24 is to select the mono channel
5. For a pin to register at 0, does it need to be shorted to ground? or can it be left disconnected.

 
To use Diff mode the user would:
Board 1:

1. Jumper I2S output to I2S input
2. Set Pin 6 to 1
3. Leave Pin 24 at 0 (Left)

 
Board 2:

1. Wired I2S output from Board 1
2. Set Pin 6 to 1
3. Set Pin 6 to 1 (Right)

 
To use Stereo the user would:

1. Jumper I2S output to I2S input.
2. Leave Pin 6 and Pin 24 at 0.

 
 
If you decide to do this, you may want to also allow separate external power for each digital and analog.
 
 
 
This would make the DAC alot more flexible, but I can understand if this isn't in the project goals.
Just something to think about.

 

[nullstring]

Another idea for consideration would be to allow volume control of the DAC via I2C.
It might work well with this because:
 
USB is only 16bits, so we have 8 bits left over for attenuation.
You would be able to get rid of the cheap and bad sounding pot in the portable amp..

 

[joneeboi]

I'm working on making this DAC work with 1206 and more realistic for building. If a person wanted dual differential, they'd need to use the Hammond 1455C120x. I could add a hardware jumper for pin 24, but it's already really hard to make things work. I'll try.
 
Also, I'd go 4 layers if I had more than EAGLE Light. I got the license, but only the 49 dollar one. I'd prefer to have a chopped up ground plane than to manually route every power and ground signal unless there was a compelling reason against it.