USB isolating question

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Seeing that many people have success with different USB isolating gizmos, I
want to find out if it is possible to make a cheap usb isolator.
Please chime in and share your knowledge if you think my suggestion is
technically impossible:
I want to hook up a cheap xmos the latest generation device so the computer
sends out audio via usb. Parallel to this device I want to put a USB (one way)
isolator followed by a good DDC with the same Xmos chip as the cheap one.
The question I have is this: will the isolated xmos work since it receives data,
or does it also need to send back data to the computer to work properly?

Any input is greatly appreciated.

Klaus
 
Cheap isolators are limited to full-speed USB (12Mbits/s). Going to high speed gets significantly more expensive - presumably its high speed you're wanting with an XMOS interface. Seeing as that's going to be an async interface, your proposal to 'piggyback' off the transmission isn't going to result in synchronization with the isolated interface. They'll have different clock rates.

The TL;DR answer is yes, the isolated one does need to talk back to the PC in order to work as an async peripheral.
 
Yes, I also think that's the normal meaning - but here I guess its referring to the DAC running async from the PC (and other DACs too). The opposite is 'isochronous' where the PC determines the sampling speed of the DAC through the timings of the packets.

With two async DACs they're going to each run at their own individual clock rates which will be apart by a few tens of ppm typically.
 
I got this out of a computer science Page about asynchronus data transfer:
"........It doesn’t need a clock for integration—rather, it utilises the parity bits to tell the receiver how to translate the data.
It is straightforward, quick, cost-effective, and doesn’t need 2-way communication to function."

Now, wouldn't this statement mean that my isolated xmos chip with its local clock just
procsesses the data that was actually send by the computer to the non-isolated xmos chip?
 
Last edited:
USB communication by principle is bi-directional. It really does not matter what protocol is being used.

As of isolation - audio devices have the advantage of typically using simple-to-isolate buses behind the USB receiver - I2C and I2S. Therefore most isolated USB high-speed audio devices use isolation behind the receiver chip (XMOS).

For the actual USB isolation - high-speed USB isolators start at 150EUR.

There is another option - using ethernet (galvanically isolated by principle, standard requires 1.500Vrms for 60 secs) and an inexpensive SBC running linux with USB-IP server Galvanically-Isolated USB (+ longer distances) - USB over Ethernet

For the USB audio async protocol - there is a small FIFO in the USB soundcard (the smaller, the lower latency and wider usability the device offers). The device monitors fill level of its FIFO and reports back accordingly whether the host should be sending more or fewer samples in each USB frame - hence the control feedback.

Adaptive audio devices just PLL their clock from the incoming stream of the USB samples - i.e. the audio pace is timed by the USB host controller.
 
USB communication by principle is bi-directional. It really does not matter what protocol is being used.

As of isolation - audio devices have the advantage of typically using simple-to-isolate buses behind the USB receiver - I2C and I2S. Therefore most isolated USB high-speed audio devices use isolation behind the receiver chip (XMOS).

For the actual USB isolation - high-speed USB isolators start at 150EUR.

There is another option - using ethernet (galvanically isolated by principle, standard requires 1.500Vrms for 60 secs) and an inexpensive SBC running linux with USB-IP server Galvanically-Isolated USB (+ longer distances) - USB over Ethernet

For the USB audio async protocol - there is a small FIFO in the USB soundcard (the smaller, the lower latency and wider usability the device offers). The device monitors fill level of its FIFO and reports back accordingly whether the host should be sending more or fewer samples in each USB frame - hence the control feedback.

Adaptive audio devices just PLL their clock from the incoming stream of the USB samples - i.e. the audio pace is timed by the USB host controller.

Thanks for your clarification.
I would really like to isolate the Computer from the xmos chip and buy a
DDC like the Singxer F1 that does the I2S isolation part.

So what would happen in a cenario like that in my picture?
If the slave xmos talks back into nirvana will it still process the incoming data
send by the computer?
If it receives the data with neglectable delay because of the isolator
will the small buffer compensate for this?
 

Attachments

  • Twin-Xmos-Isolator.png
    Twin-Xmos-Isolator.png
    12 KB · Views: 235
Please help me to see my error in this configuration.

a "write" token is generated by the host
both master and slave are then reading the data packet.
since both devices are working at the exact same speed
the handshake packets by the master should work for the
slave as well I guess.
 
First you'd better have a closer look at state-of-the-art quality USB DACs.

These exhibit negligible noise, jitter and distortion levels on standard USB connections, without the need of adding
any kind of USB gadgets. Surprise? ;)

You might want to read this first.

The ASR guy IMO simply got his conclusion wrong though!!!

It should be IMO phrased this way:

The iFi USB filter does a good job on the filtering part, it lives up its promises.
It very well cleans the USB side.

However. If you use a excellent low noise DAC, such as the Topping D90,
a filter, pretty much any filter, wouldn't have any impact.
Because a DAC in that league usually has such a filter feature already built-in!

Under certain conditions a filter can make things even worse. No filter is lossless!
And these filter associated losses might add to the overall result.

E.g As seen on the chain under measurement , if you don't review your grounding scheme
after introducing a new gadget, which e.g. isolates just one leg, you might catch other
issues through the backdoor.


Bottom line. A USB filter, a reclocker, a regenerator, an isolator, asf. can have a positive impact,
if the DAC is of rather low quality.
(And this has pretty much nothing to do with XMOS btw.)
If the DAC designer has designed a DAC with all the challenges and weaknesses that come with USB in mind,
any kind of filter gadgets will and should be useless.

If you run rather low-Q or med-Q DACs (in terms of lacking a USB cleansing effect), filters might very well be a way to improve your audio experience.
Note: This can happen on DACs that cost several thousands of $!

Conclusion:
Better, neutral or even worse results by adding filters are possible. :spin:
And fun-fact: The experienced results solely depend on your very own chain.

So, If you hear improvements by adding a filter it simply means that your DAC can't cope
properly with the USB challenges. And from here you can go back to the first line of the post. :D

Instead of poking around in the USB gadget jungle, better have a look at state-of-the-art USB DACs and your setup first.
And stay away from DACs , which exhibit clearly audible differences under changing (USB) conditions.


Enjoy.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.