spdif vs usb

[dublin78]

P.s. Yeo's has an optional 12s output if needed in the future.

[pupsik46]

Originally posted by hagtech
Of course, I2S could also be coupled via trannys (like DSL ones), but I don't believe anyone does it.

Why not to use ISO or ADUM for galvanic isolation?

[Radian]

Originally posted by pupsik46
Why not to use ISO or ADUM for galvanic isolation?

That's what I wondered also. I have a few ISO150 from TI that I had ordered for that purpose, but after I had so many problems with the PCM2707 dieing on me, I gave up on it.

Here is the description from TI:

"The ISO150 is a 2-channel, galvanically-isolated data coupler capable of data rates of 80M Baud, typical. Each channel can be individually programmed to transmit data in either direction.

Data is transmitted across the isolation barrier by coupling complementary pulses through high voltage 0.4 pF capacitors. Receiver circuitry restores the pulses to standard logic levels. Differential signal transmission rejects isolation-mode voltage transients up to 1.6 kV/µs

The ISO150 avoids problems commonly associated with optocouplers. Optically isolated couplers require high current pulses and allowance must be made for LED aging. The ISO150's Bi-CMOS circuitry operates at 25 mW per channel."

Man it would be great if they could be used for USB. With my old PCM2707 USB Dac I had terrible hum by using the laptop's power supply. On batteries there was no hum.

I got another PCM2707 USB Dac in the works and I really want to
solve that problem before going on.

I will probably first try the approach John of EC-designs used. He seems to not even care about the quality of the signals to much till they reach the dac. He uses a very smart approach by just going differential with buffers on all the I2S signals from the PCM2707. These differential signals are not referenced to ground. Than he goes back to SE using the same chip with a different 5v supply. Now the I2S lines are just referenced to the dac's ground and voila, the two grounds are isolated and common noise is rejected.

He still uses the PCM in bus powered mode though which I don't want to do, but it could be that it is not as critical because of the way he proceeds with the I2S signals.

Maybe someone who has experience using ISO's can chime in and tell us if it can be used to isolate the USB lines.

Greets,
Klaus

[pupsik46]

Originally posted by Radian

Maybe someone who has experience using ISO's can chime in and tell us if it can be used to isolate the USB lines.

You don't isolate USB lines, but put ISO's after PCM2706/7 I2S output. I have made several USB DACs (AD1852, TDA1543, AD1851) with and without oversampling.

[Gasho]

How about wireless data transfer like Squeezbox? No ground loops, no cables from PC to DAC....

[Radian]

Originally posted by pupsik46
You don't isolate USB lines, but put ISO's after PCM2706/7 I2S output. I have made several USB DACs (AD1852, TDA1543, AD1851) with and without oversampling.

Your right of course, the PC wouldn't even detect the device.

Wireless USB (CWUSB) was announced last year but there is still no products available for end users.
http://www.usb.org/developers/wusb/

Using the approach of EC-designs seems to be a better solution
then going the ISO route.

Greets,
Klaus

[analog_sa]

Originally posted by pupsik46
Why not to use ISO or ADUM for galvanic isolation?

I have asked myself the same question. Using these isolators is obviously a compromise between added jitter (ISO has a jitter spec, ADUM's datasheet only quotes pulse width distortion and skew) and reduced noise.

Depending upon the situation one may choose the lesser evil.

Isolating the USB bus itself is probably a safer option. At least the PLL in the USB chip will probably reduce the added jitter.

Why are commercial optically isolated USB cables so expensive?

As for ECdesign's solution it does not offer the same degree of isolation. As for jitter, he solves this with an asynchronous reclocker

Hagtech's solution is quite elegant but obviously puts a lot of faith in the secondary PLL. Sadly these devices are not really diy-friendly. Yes, i can buld one. But how would i know whether it really works?

[Radian]

Originally posted by analog_sa


Hagtech's solution is quite elegant but obviously puts a lot of faith in the secondary PLL. Sadly these devices are not really diy-friendly. Yes, i can buld one. But how would i know whether it really works?

...... with your ears.

Why do you think asynchronous re-clocking can not yield good results?

I am really fortunate in that I have no clue about electronics.
That way my brain does not get in the way of my ears.

It is amazing how most people hold on to a technology when they are convinced of it and how reluctant they are to try anything else.
I try to have an open ear and mind after experiencing how a mediocre implemented AK4396 dac on a soundcard walked all over a well respected 8X stacked TDA1543 dac that was highly tweaked. It took only 8 caps and the removal of the whole active stage to do that.


I know better now than to blindly believe those who say: that way of doing it will not work or, this reclocking can not yield good results........

Best thing we can do is building what people claim is good and comparing it with the other designs we have.

Greets,
Klaus

[t-head]

astouffer,

Peter Daniels offers a nice NOS DAC with USB as a pcb, kit or completed unit. I like the sound. I use EAC (Exact Audio Copy) to rip CDs and the result is very good...

tea

[analog_sa]

Originally posted by Radian


I am really fortunate in that I have no clue about electronics.
That way my brain does not get in the way of my ears.

You make a pretty good point here. I draw most of my conclusions from listening but conducting proper listening tests is both exhausting and time consuming.

Originally posted by Radian

I try to have an open ear and mind after experiencing how a mediocre implemented AK4396 dac on a soundcard walked all over a well respected 8X stacked TDA1543 dac that was highly tweaked.

Now this is amusing. Respected by whom? Any half-assed EE with no ears will predict that the AK4396 will walk over an ancient, low-performance, budget DAC. One of those rare moments when theory and practice produce identical results