Usb dac

[5th element]

I mentioned this briefly in another thread and said I'd post about it when it was completed and here that thread is.

The design is based around the PCM2706 USB interface chip from TI and is able to output as I2S. The output of the 2706 feeds into a SRC4192 where the signal is reclocked, before being fed into a PCM1794.

The 4 power supplies are based on the ALW super reg.

First there's a picture of the main PCB, then two of the finished unit.

[5th element]

The box is a model from Hammond that I got through Farnell. What surprised me was how nice the box actually looks. Enclosures are usually expensive, but this one only cost about £23.

The next picture is of the insides and then a picture of some of the tests I ran. It performs well enough considering its beginnings, although the jitter is higher then I'd like (that's if these measurements can be trusted).

Sound-wise it doesn't match up to the last thing I posted about, but then I didn't expect it to. It does sound fine in practise however and you only really notice its shortcomings when you do a direct A/B comparison. As you'd expect the sound has a similar flavour, considering the DAC/IV and headphone amp are exactly the same.

[ratza]

Nice project. Mine is with PCM2704, bus powered. The computer makes enough noise so I don't hear the 0.06% THD. I wouldn't hear them anyway

[mravlca]

Any chance to post the PCB and the scheme of the main board?

Regards Aleš

[5th element]

Originally Posted by mravlca

Any chance to post the PCB and the scheme of the main board?

Regards Aleš

A schematic I don't have. I basically designed the PCB whilst looking at various data sheets and schematics from other projects.

The PCM2706 is configured pretty much how the TI data sheet suggests. I use a combination of figs 33 and 34, + logic gate.

The PCB around the 1794 was a simple cut and paste job from another DAC PCB I had, but was designed following the 1794s data sheet + some small resistors in series with the feedback capacitor of the I/V opamps - the THS4031s don't like a capacitative load.

The SRC chip was implemented using a combination of how it's used in the EZDAC, theAnonymous1's microdac and the TI data sheet - so thanks to those designers.

I have found the data sheets from TI to generally be very good, but I think the SRC4192s could be made clearer wrt the sampling frequencies, perhaps with an example.

Unfortunately the program I used to design the PCBs is quite unfriendly when it comes to exporting the design in different formats. It won't allow me to export it as a PDF or print to file etc.

Having said this the PCB is not difficult to design, although I have had quite a lot of practise over the years. If you spend a decent amount of time using the free version of Eagle, you should have no problems at all designing your own.

Designing and making PCBs isn't difficult at all, it just requires a bit of practise and not caring if the first ones you make are not so good or complete failures. My first ones were just awful compared to what I can make now.

[5th element]

Originally Posted by ratza

Nice project. Mine is with PCM2704, bus powered. The computer makes enough noise so I don't hear the 0.06% THD. I wouldn't hear them anyway

It'd be interesting to compare what differences there may be between bus and self powered. I did read somewhere that when fed from a laptop, the jitter and distortion performance was significantly improved when the laptop was powered by its internal battery, rather then plugged in.

I would imagine if using a seperate power supply that it would negate most of the benefits perhaps. With my design I deliberately chose to leave the ground connection from the USB cable at the PCM2706s end lifted. That way the noisy ground of the computer isn't transmitted directly to the USB DAC. What with the USB signal being balanced already, the fact the ground is broken shouldn't really be an issue. Obviously the USB DAC has a connection to ground itself.

[soundcheck]

PCM270X are a total mess. These chips are extremely sensitive to incoming jitter
generated by the PC and causing quite some jitter on the I2S.

I am running an Opticis optical USB cable for galvanic isolation and a battery supply.
I reclock the I2S and my PC is heavily tuned to avoid jitter on the PC side. And still
you're not 100% satisfied.