Critique my USB DAC design

[MWP]

Hi all...

Ive come up with design for a DAC based on PCM2902, DIR1703, PCM1730 and OPA627s.

Anyone care to look it over and give me some opinions?

http://www.overclockers.com.au/~mwp/...WP-USB-DAC.pdf

[ukram]

You should use pull-up resistor in D+ USB data line. And 22 ohm resistors in series for D+ and D-.

[MWP]

Ok, thanks.

Also, is the power for the PCM2902 ok?
The way it is in that sch, its going be powered only by USB... not buy the DAC's supply.

[ghetto]

MWP, I would not recommend using the USB power method. You will be undoing much of your hard work by using the very nasty power available on the USB bus... I too am working on a USB DAC, well actually a USB Amp, and I will be going to great lengths to supply power that is extremely clean to all chips involved. You can look here for a PDF that outlines the "super regulator" which I will be using... the author claims it creates less noise than batteries! Good luck, keep us updated on your progress!

[tiroth]

Better bypassing everywhere is a good idea if you are going for high quality. I think this is a big advantage of DIY.

I am using 100nF 0805 || 1uF 0805 || 10uF 1210 || 220uF FC on my current project. The rails are definitely cleaner than I've seen in commercial products. I think resonance is generally not an issue due to the very small parasitics of SMD.

[ukram]

I think PCM2902 works fine with USB power if you are using it only as digital USB/SPDIF converter.

[MWP]

Originally posted by ukram
I think PCM2902 works fine with USB power if you are using it only as digital USB/SPDIF converter.

Yeh, i didnt see it as being a problem as the PCM2902 wont be used for its DAC outs, just the SPDIF out which shouldnt be effected by noise on the USB power.
The PCM2902 also has decent internal regulators anyway.

Thanks for the comments guys.

Ill beef up the bypassing also.

[ghetto]

I don't want to hijack your thread MWP, but I have just been told by a friend that his old USB speakers were very "laggy," ie, the sound would never synch with a DVD, etc. Seeing as I have not yet got my design fully working, I was wondering if anyone has experienced this "lag" before?

[jcx]

Separating the USB interface and the rest of the DAC at the S/PDIF with opto or Xmfr isolated coax interface would at least give you a OK DAC when not plugged in to USB and good ground isolation couldn’t hurt when using USB

A quick look at the data sheet for the 1703DIR leaves me wondering what purpose is served by the 12.288 MHz Xtal; it is only compatible with 32/48/96 K sample rates and the 2902 doesn’t handle 96 K, is there a lot of 32/48 K source material out there?

I/V converters benefit from low output Z to reduce the voltage at the DAC out/op amp input from the high frequency switching glitches that feed directly through the feedback C, the 50-60 Ohm OPA627 output Z can be reduced by ~10x with a buffer in the loop, 1 or 2 Qs and a few Rs per I/V stage would only add ~$2-3 to the cost (if you can afford ~ 25 mA per buffer class A bias current) current feedback op amps can also be used as buffers in the loop ( I think extra gain in the loop can be good too, but you need to know what you’re about to compensate the loop for stability ) C11/12 noise gain caps should reduce input glitches too, I would also put a RC filter in series w/Vcom2, you can always jump/not stuff these

[MWP]

Separating the USB interface and the rest of the DAC at the S/PDIF with opto or Xmfr isolated coax interface would at least give you a OK DAC when not plugged in to USB and good ground isolation couldn’t hurt when using USB

Very good idea, thanks.

Any sugestions on opto-iso's that will handle SPDIF?

A quick look at the data sheet for the 1703DIR leaves me wondering what purpose is served by the 12.288 MHz Xtal; it is only compatible with 32/48/96 K sample rates and the 2902 doesn’t handle 96 K, is there a lot of 32/48 K source material out there?

Well the 2902 will be getting 48Khz sound from the PC... ill make sure of that.

But really i had no choice on the freq of that xtal, its the only value ive got on hand that will work there, and getting other values here is expensive.

The 1703 wont be locked anyway, so itll still take 44100.

I/V converters benefit from low output Z to reduce the voltage at the DAC out/op amp input from the high frequency switching glitches that feed directly through the feedback C, the 50-60 Ohm OPA627 output Z can be reduced by ~10x with a buffer in the loop, 1 or 2 Qs and a few Rs per I/V stage would only add ~$2-3 to the cost (if you can afford ~ 25 mA per buffer class A bias current) current feedback op amps can also be used as buffers in the loop ( I think extra gain in the loop can be good too, but you need to know what you’re about to compensate the loop for stability ) C11/12 noise gain caps should reduce input glitches too, I would also put a RC filter in series w/Vcom2, you can always jump/not stuff these

Thanks... but this is still my first DAC, so i think ill leave the I/V simple for the moment.
I can always change it later.