USB DAC Selection

[hagtech]

He measured double the jitter, if I recall it right

The jitter coming out of the USB chip is the same whether it is I2S or S/PDIF. The difference occurs in the S/PDIF receiver that you have to add. That's where the extra jitter is added.

As the I2S isn't perfect either (nothing is jitter-free), it still helps to re-clock. That's what I do.

So I figured, I'm re-clocking anyway, it's not going to hurt too much to use the S/PDIF output. The benefit of choosing this approach is that it is much easier to isolate grounds. Most of the I2S implementations you see tie the ground of the computer directly to your audio system. To me, that was actually the worse of two evils. So I use the S/PDIF output from USB chip, and run it through a tranny for complete isolation of supplies and grounds. This method has the side benefit of making an input selector switch easy. Only one 75-ohm signal per input. Much simpler than trying to mux several I2S inputs (which ties together even more grounds).

jh

[fluckscapacitor]

hagtech,

So if you were going to use your USB->SPDIF converter in a single box with a wall powered DAC, you would still use the 3.3V supply from the USB line to power that chip?

[soundcheck]

Originally posted by hagtech



As the I2S isn't perfect either (nothing is jitter-free), it still helps to re-clock. That's what I do.


jh

Why not taking an I2S output of an Envy-chip based soundcard and reclocking this I2S signal with a precision clock!

Is anybody aware of a reclocking module for I2S?

The Envy chip on such as soundcard can even be driven as clock slave, if the I2S reclocker runs as master! That's most probably
even better.
(Do I mix up things here?)

I am wondering in general if the ground issue is such a big issue.
Lynx managed to come up with an internal PCI solution, where
a noisy PC environment doesn't seem to be an issue at all.
I think the ground issue is becoming a real issue the more (independend powered) stages you connect togehter.
My idea: Limit the involved stages!

Cheers

[hagtech]

So if you were going to use your USB->SPDIF converter in a single box with a wall powered DAC, you would still use the 3.3V supply from the USB line to power that chip?

Depends on how it is done. I would not power the USB chip with any of the supplies in the DAC. Better to make an independent supply. The reason again is to prevent coupling from a computer system ground to the audio ground.

There are two ways to power the USB chip. Easy method is to use the 5V bus power supplied by the computer. The problem here is that this is a noisy supply. It is powered from a dc-dc converter that has both HF and LF noise. It is not low impedance and varies with other loads within the PC such as when a hard drive kicks in or a fan. Some PCs are better than others. This is NOT the best raw 5V power you can get. The other approach is to add a supply running from AC mains. However, that power line is just as noisy and full of garbage as the PC power.

So what to do? Use dirty dc power or dirty ac power?

The answer is that you can use either. It's just a matter of addressing the problems proactively. For dc, you use passive filtering to clean up the HF noise on the 5V bus. Since the USB chip runs at 3.3V, you have quite a bit of voltage headroom to play with for the filtering. I use a ferrite bead and low impedance electrolytic. That cleans the HF. Then use the built-in linear voltage regulators on the USB chip. That cleans up the LF. You end up with pretty decent clean supplies. Most importantly, there are four separate linear regulators, which you keep separate. This helps to decouple noise between the analog, digital, PLL, and other circuit sections within the USB chip.

Using an ac supply can be better. The reason is that you can clean up the dirty input and create a nice dc output. In fact, a regulated and already clean 5V dc output is best. This is in comparison to the dirty 5V dc from the USB bus. So obviously, it can be an improvement. Then the rest is the same. What you have to watch out for with the ac supply, is that it remains grounded to the PC side, not the audo side. And then you still use a transformer output. You see, if you had borrowed 5V from the DAC, then you would have to tie the PC and audio grounds together. That would introduce noise back into the system.

The other mistake you can make here is by supplying a common regulated 3.3V to the USB chip. The reason that is not good is because it ties together the internal sections. You don't want to do that. Keep the digital supply off the PLL off the analog. You can, however, use 3 or 4 separate 3.3V supplies. In fact, you can improve a notch further by running these at 3.6V. Just be careful that if you try this, sequence the outputs to make sure they all come up at the same time.

jh

[imperfectcircle]

Why not just get a high end soundcard like the Lynx l22 , then you dont have to worry bout usb, spdif or i2s from the pc to dac. Ive heard alot of people like the sound better then the dddac.

[soundcheck]

Originally posted by imperfectcircle
Why not just get a high end soundcard like the Lynx l22 , then you dont have to worry bout usb, spdif or i2s from the pc to dac. Ive heard alot of people like the sound better then the dddac.

The DDDAC, especially the USB variant, needs proper source material, a highly optimised PC setup and some tweaking to sound best. I think a lot of people just miss this out.

If you manage to get this done, all the NONOS DAC incl. passive output stage advantages will apply!

The DDDAC is loosing, IMO, at least 30% of its potential on the way from PC to PCM2707. If you manage to improve the front-end of the DAC, Lynx will have a hard time.

E.g. Just feeding the DAC with 48kHz material (offline upsampled with Shibatch ssrc under Linux) lifts it up in a different league!


Cheers

[simcap]

Originally posted by dublin78
Newbie question: Please can you explain the function and advantages of the two capacitors connected to the analogue outputs?
Michael

Hello dublin78,

They are the output capacitors.
In this picture of the dac scheme:

http://www.simonecapretti.eu/Temp/Image1.jpg

you can view that output capacitors are not on the circuit board.
In the kit are already included the 100uF + 100nF but it's so funny to try other values and brand :-)
So actually I'm using the Mundorf MCap Supreme Silver Oil.

Ciao!

Simone

[goon-heaven]

Hi Soundcheck,

Though I've thrown out my coupling caps, since I connect the DAC directly to the AMP ( I am running a modded Charlize module).

Is this because Charlize has input caps?
* If so, what caps are they? And what did you modify on her?
* If not, explain how it works?

Thanks
Steve

[soundcheck]

Originally posted by goon-heaven
Hi Soundcheck,



Is this because Charlize has input caps?
* If so, what caps are they? And what did you modify on her?
* If not, explain how it works?

Thanks
Steve

Correct.
Charlize input caps are sufficiant.
If you don't catch DC on the DAC output, forget about the output caps. The best cap is no Cap!
I used small value input caps, running as low-cut in conjunction with the 20k input resistor, on the Charlize input.
I just run my mids and highs over Charlize, crossed over at
250Hz. That way you take off low freqeuncy load from Charlize.

I use BG non polars as coupling Cs. They sound as good as my Mundorf Silver/Gold I used before. The small values do make a
huge difference.

This tweak heavily speeds up the whole chain.

Further mods on Charlize: All BG caps and Mundorf Air Coils and some other minor things.


Cheers

[goon-heaven]

Thanks Klaus.

I'm running foobar into RME96 SPDIF into DDDAC Mk1 into Gilmore Headamp. I have (only) this one pair of Auricaps, at the output of the DDDAC, that I wish were not there (for all those reasons you give). Looks like they will survive just a little longer.

Different to your USB input: I would like to know how to build a clock output circuit from the Tent clock in the DDDAC that the RME would slave to. Who can spell it out to me?

I am also considering how to do a Mk2 PC audio, so I read all this with much interest.