New Posts  All Forums:Forum Nav:

usb dac - Page 3

post #31 of 519
I really think this is cool.

But what about the software drivers required to use the TI part? Something is going to have to tell Windows that this is a sound device and all that jazz. Any thoughts?
post #32 of 519
Any device that complies with the USB Audio spec works without any special drivers. It's like USB keyboards and mice -- you don't need special drivers for them, either.
post #33 of 519
Well, OK.

I'm in to build one, assuming I don't have to open the Windows DDK.
post #34 of 519
Thread Starter 
[IMG]ez[/IMG]

Second layout, slightly bigger (I could still fit 2 of these on an Eurocard).

Differences :

- Room for a pot
- Less jumpers and shorter ones
- Amp is now multiloop and not simply buffered
- Use of 2 batteries instead of 3.
- Voltage of the amp down to +/-4.5V.
- Use of two 2200uF/10V instead of four 470uF/25V for the analog supply.


Estimation of the cost : roughly 75$.


EDIT : layout withdrawn, very unsatisfactory
post #35 of 519
Thread Starter 
stupid question... do I need that pot or is the digital attenuation of the pcm2702 satisfactory ??
post #36 of 519
Would the digital attenuation need a controller of some sort or is that also part of the specification (USB Audio)?

This project seems seriously SMD. Is there a plan to demonstrate construction using a toaster oven and solder paste? I can see my wife already wondering why I'm taking the toaster oven into the garage. And then I can see myself at the store the next day buying her a new one.
post #37 of 519
Thread Starter 
From the datasheet :

Quote:
Digital attenuation and softmute features are included, and are controlled via USB audio class request.

Digital Attenuator: 0dB to –64dB, 1dB/step
There are not that many smd parts. Of course, there's the pcm2702, that's a ssop-28. All the other ICs are so-8, like the ones used in a Mint. The other parts are full size.
post #38 of 519
The only other thing I could see that might make this a little sweeter is the ability to bypass the headphone output and just get the signal out of the unit to connect to another amplifier. I see this as an extra input to my desktop PPA amplifier that I would mount inside the case with a rear-mounted USB terminal.

If the op-amps could be configured to work as the gain adjustment/digital filter as a preamp to the main amplifier, that would be sweet. I haven't read the datasheet, but on other DAC's I've used an active low-pass filter is done using an op-amp and some caps/resistors. There is an applet on the Analog site that helps you tune your filter values for a 96kHz rolloff, etc. to get the high-level noise out of the signal.
post #39 of 519
Thread Starter 
Here's the "definitive" layout.

I scratched the pot, moved things around to get more room, added 2 more caps in the analog supply, made all the power distribution for the headamp by wires to avoid big power traces all over the board, added 2 pads to get the signal before the headamp.



Edit : power consumption problem :

This thing should eat at least 40mA for the DAC and Sijosae claims a 20mA consumption for the amp section. So, it gives us a 60mA total power consumption. On two rechargeable 9V batteries, with 200mA, it gives us about 6 hours and half of life. On 8 AA rechargeable batteries, at 2000mA, it gives us a more satisfactory 33 hours life

Another option would be to use the power the USB could give us. It'd take a 5 to 9V dc-dc converter and quite a bit of regulation to clean the power coming from the computer.
post #40 of 519
Just allow for a DC power connector and we'll worry about getting enough juice to it.

I'll probably power it off linear supply (LM317-style)...
Or a simple wal-wart would suffice for many like the Audiophile USB.

The evaluation board uses a 6.3V power supply itself.

There is always Tangent's battery board with 10 AA batteries as well..
post #41 of 519
thats looking nice 00940! however, I was wondering if shoehorning the diamond buffer into makes sense both sonically or practically:

using a space for a monolithic buffer (say the HA3-5002, or BUF634) would take a lot less space, while at the same time leave the option of the discrete buffer open, as can be seen with sijosaes implementation of the Jung buffer on a piece of board not much bigger than a DIP chip, or by ppls superlative discrete buffer. The monolithic buffer will likely sound better than any cheap 2N3904/06 implementation as well, but that's by no means definite

also, it means people would have to buy excess output transistors so they could be matched to minimise DC offset, which is going to boost up the cost, complexity, and patience level of those wanting to make it.

this is your design, I know, but if you wanted others to construct it (and I'd build it either way), the DIP option is definately worth considering seeing as there are components on that side of the board anyway

edit: just another thought on this; how many people would use the DAC with the headphone amp on board, and how many would connect to a dedicated headphone amp?? If no one wants it to have an amp on board, you could gain a lot of space, reduce the cost, and crucially, reduce the power consumption!

I'm also going to question the use of the discrete rail splitter (sorry!). I can see the big advantage in this design being all discrete in that all the parts are readily available (especially seeing as we're in Europe!), but the performance of the discrete rail splitter does not even come close to matching a TLE2426 unless, again, you use some serious hand matching and excess component purchasing and even then, an imbalance is still more likely to occur than with a laser trimmed resistor pair.

g

edit2: bit of housework
post #42 of 519
Thread Starter 
Quote:
Originally posted by guzzler this is your design
Actually, I could hardly consider it as "my" design. The dac is straight of the datasheet and the analog section copied from different designs Sijosae posted on headwize. And it was more like an homework to get my feet wet with eagle.

Quote:
using a space for a monolithic buffer (say the HA3-5002, or BUF634) would take a lot less space, while at the same time leave the option of the discrete buffer open
Well, if we're going the road of monolithic buffer, we could as well shrink the space. The builder would have to solder a difficult piece (the pcm2072 ain't big). He'll probably be able to solder so8 parts.

Quote:
just another thought on this; how many people would use the DAC with the headphone amp on board, and how many would connect to a dedicated headphone amp?? If no one wants it to have an amp on board, you could gain a lot of space, reduce the cost, and crucially, reduce the power consumption!
Honnestly, I didn't intend this design to go commercial in any way or be widely built. I planned to build one for me because I could have a need for such a design. An integrated solution would be cheaper than an sonica+pimeta combo. I'm not sure the DAC alone would offer a better quality ratio than a sonica. However, swapping the headamp for an output stage could be interesting considering most diyers able to solder such a thing already have an amp or two or even three.

So... I'm gonna play a bit more with this. I'm gonna suppress the headamp, implement a low pass filter, an output stage (perhaps just a pair of buf634, the pcm2702 outputs up to 3V !) and try to get the power from the USB port.
post #43 of 519
I would look at the schematics of the ART DI/O. It's an excellent DAC with a nice op-amp based low-pass filter and highly regarded sonics. Use that as an example to create a line-driver stage.

If you are truly looking to make a "Sonica with nuts" then use some monolithic buffers with some rollable DIP8 op-amp sockets and room for a gazillion batteries.
post #44 of 519
Thread Starter 
actually, i'm looking at the evaluation schematic that TI released. There's a low pass filter just waiting to be picked up.

I'll probably swap the opa2353 they used for an AD823 : they're cheap, they're good and above all, they work from a single 5V !

We could thus power the whole thing from 5 AA with a comfortable lifetime. A charger could probably be easily implemented.
post #45 of 519
keep us posted 00940, i would certainly be very interested in building one of these, and i've got a UV exposure unit, so if you leave it single layer, i could run you off a couple from that. I might run off a bit of the digital section with the line out, and hook it up, see what comes out...

all the best,

g
New Posts  All Forums:Forum Nav:
  Return Home