USB DAC isolation ?

[jkeny]

Originally Posted by wwenze

You already said it's a differential signal, so it does not need a common reference ground.

However i think this is pointless though, since any DAC with its own power supply will be using that to power the USB anyway instead of through the computer. I made the cable because something was causing my USB-powered DAC to crash so I decided to isolate its power from the computer. It still crashed though.

wwenze, I thought you were saying in this quote that your DAC had it's own power anyway (not powered from USB) & later on you said you just cut the USB power & gnd wires to reduce noise?

Anyway, the link above seems to say gnd is needed - it's a pity as I would have liked to isolate the PC completely.

I still ask how is the opticics done - it's galvanically isolated, I believe - does anybody know?

[wwenze]

*Any DAC with its own power supply* - not my USB-powered DAC

[glt]

USB 2.0 spec section 6.4.2: The GND lead provides a common ground reference between the upstream and downstream ports. The maximum cable length is limited by the voltage drop across the GND lead. Refer to Section 7.2.2 for details. The minimum acceptable wire gauge is calculated assuming the attached device is high power.

According to this article you can also use a chip to isolate usb signals: http://www.analog.com/library/analog..._isolator.html

[eclectic2k]

Originally Posted by glt

USB 2.0 spec section 6.4.2: The GND lead provides a common ground reference between the upstream and downstream ports. The maximum cable length is limited by the voltage drop across the GND lead. Refer to Section 7.2.2 for details. The minimum acceptable wire gauge is calculated assuming the attached device is high power.

According to this article you can also use a chip to isolate usb signals: http://www.analog.com/library/analog..._isolator.html

The article points out the AD 'ADuM' isolator chips. These are quite nice. These would be my first guess at something that would work well for USB isolation. They have bidirectional models that provide for a couple of serial data lines. I think they have some that can handle USB edge rates.

[soundcheck]

What chips are Ayre and Naim using? It seems that they have OptoCouplers working?

[soundcheck]

Hi.

Here is an interesting article.

http://www.edn.com/article/CA6347257.html


12MBit is pretty much the maximum optical USB can do. This is also what TI told me a while ago.


There is one solution: ICRON Ranger 442 http://www.icron.com/products/usb/us...er_442_444.php

This one is the only device I am aware of which will handle 480MBit/s. It is USB over IP.

[fmak]

Originally Posted by soundcheck

Hi there.

I was wondering if somebody could share a great idea of how to galvanically isolate an USB 2.0 DAC, just by introducing a selfpowered DIY-adapter.

The Opticis does USB 1.1 only and is pretty expensive.


THX

If you want to isolate galvanically, use a good transformer, with a proper circuit. What makes you think that an opto circuit sounds better? They usually sound worse.

[eclectic2k]

Originally Posted by fmak

If you want to isolate galvanically, use a good transformer, with a proper circuit. What makes you think that an opto circuit sounds better? They usually sound worse.

this isn't digital audio stream in the same sense as s/pdif. It's basically packet data. The OP trying to isolate a USB serial communication stream. Normal transformers won't work for this purpose, as they need to be very high speed to not mess up the digital signals. (which regardless of data rate tend to have fairly fast edges)

The ADuM parts aren't opto-isolators. They are actually very much like tiny little transformers on a chip. (or they actually *are* little transformers) Also with bidirectional buffers built in to each side of the IC. I think they tend to be faster that opto's...

In addition to AD, Silicon Labs (silabs.com) also make some of these style digital isolator. I really think these are the best bet. And since you're isolating the USB stream, they shouldn't crap up the sound any more than uSB already has. -- that is, in the USB path, they should have zero effect on the resulting audio signal - other than that you've now isolated yourself from the computer, which should be a very good thing.

[eclectic2k]

Originally Posted by soundcheck

Hi.

Here is an interesting article.
http://www.edn.com/article/CA6347257.html


12MBit is pretty much the maximum optical USB can do. This is also what TI told me a while ago.

I only skimmed the article. Did give me a thought though. As they are saying with SPI in their example, yuo could isolate the resulting I2S signal before it goes to the DAC chip. (but again there are ADuM parts I think that can do 150Mbit so either way should work)

[soundcheck]

Originally Posted by fmak

If you want to isolate galvanically, use a good transformer, with a proper circuit. What makes you think that an opto circuit sounds better? They usually sound worse.

1. First of all: Nothing "made me think"! --- I am running an Opticis and won't do anything without it. (AFAIK even Wavelength is using it. Ayre runs also optical isolation)
2. See eclectic2k answer. It will still depend on the jitter tolerance of the receiver chip if
the optical transmission and conversion makes things worse.
3. If you run asynchronous (M-Audio Transit) the data/timing quality on the USB link doesn't matter anyhow as long as it stays within the jitter tolerance of the receiver.
4. Ground-loop usually still exists due to shielding if you not go for optical.
5. Through a battery powered receiver (modded Opticis PS ) I can feed BUS powered devices very well -- e.g. my M-Audio Transit is working much better afterwards.
6. I guess you are well aware of the USB-cable discussions at AA, where people spent a couple of hundreds bugs on USB cables to reduce the transmission losses. I don't have that issue at all.


Let us know if you found or if you know or if you've tested by yourself a plug'n play device or a DIY solution based on transformers.

THX