Author Topic: Isolated USB 3.0  (Read 6169 times)

0 Members and 1 Guest are viewing this topic.

Offline wholderTopic starter

  • Regular Contributor
  • *
  • Posts: 72
  • Country: us
    • Wayne's Tinkering Page
Isolated USB 3.0
« on: January 30, 2019, 12:27:47 am »
I just ran across a new USB isolator that claims to run at USB 3.0 speeds.  Would love to know what chipset is used inside to accomplish this.  Also, if anyone has purchased and used this device, please pass on your thoughts how well it works.

  https://www.saelig.com/product/adq-usb-iso-ps.htm

Wayne
 

Online coppice

  • Super Contributor
  • ***
  • Posts: 10030
  • Country: gb
Re: Isolated USB 3.0
« Reply #1 on: January 30, 2019, 03:02:59 am »
There are several USB3 5Gbps isolator gadgets available. I don't know what devices any specific gadget may use, but things like http://www.advancedphotonics.co.jp/en/usb3-isolator.html exist for this purpose,
 

Offline DaJMasta

  • Super Contributor
  • ***
  • Posts: 2392
  • Country: us
    • medpants.com
Re: Isolated USB 3.0
« Reply #2 on: January 30, 2019, 05:25:53 am »
A lot of high bandwidth options are actually on-chip optical RF links.  They modulate the incoming data onto the optical signal, then demodulate it a few mm away where it's received on the opposite side.  You can get a lot of data through a single channel that way, and the isolation is as high as you put the transmit and receive diodes.
 

Offline Doctorandus_P

  • Super Contributor
  • ***
  • Posts: 4004
  • Country: nl
Re: Isolated USB 3.0
« Reply #3 on: January 30, 2019, 06:43:30 am »
The block diagram on page 2 of the datasheet suggests a capacative barrier.
This seems a logical choise. Capacitors work better at high frequencies.

Capacitive barriers are also already being used in similar applications (Was it in the ADUC.... signal isolators?) and in MOSfet drivers such as IR2102.

Optical fibres (with their electronics) go upto much higher bandwidts, but might not be available for the right price point :)

Curios if some RF link would be doable. A low power transmitter and receiver in a metal can could be very low cost. Look at the multitude of 2.4GHz WiFi and other transceiver chips. A few of those links could be used in parallel. With short distance and shielded metal can these transceivers could be very simple. Even some pulsed theme similar to the principle of the very old spark-gap transmitters.
 

Offline SiliconWizard

  • Super Contributor
  • ***
  • Posts: 15794
  • Country: fr
Re: Isolated USB 3.0
« Reply #4 on: February 02, 2019, 04:28:45 am »
I think TI's (and SiLabs?) digital isolators typically are capacitive, whereas AD's ones (ADuM series for instance) use micro-transformers.

The fastest I could find so far from AD is this one: https://www.analog.com/en/products/adn4654.html (1.1Gbps).

One alternative possible approach could be to use 10G ethernet transformers such as this: https://www.haloelectronics.com/products/lan/ethernet/discrete-transformers/10g/
of course with appropriate signal conditioning... (not necessarily trivial).

Or yet another one could be to use USB 3.0 transceivers such as this: http://www.ti.com/product/TUSB1310A and isolating at the PIPE interface-level, which is "much" slower at 250MHz. That would probably be pretty expensive.

The Advanced Photonics isolator seems great. Didn't know these products. Would be a likely candidate if the diagram didn't suggest a capacitive isolation (and also the size of the box which give hints about the circuit's size, although it may be deceptively empty...)
 

Offline jpb

  • Super Contributor
  • ***
  • Posts: 1771
  • Country: gb
Re: Isolated USB 3.0
« Reply #5 on: February 02, 2019, 12:01:03 pm »
The audio hifi company ifi do one but it is pretty pricey:
https://ifi-audio.com/products/nano-igalvanic3-0/
And intona do a superspeed USB 2 which obviously not as fast is much faster than most of the standard usb 2 isolators:
https://pro.intona.eu/en/products/
They use the Si 8660 B chip:
https://www.silabs.com/documents/public/data-sheets/Si866x.pdf
 

Offline IDEngineer

  • Super Contributor
  • ***
  • Posts: 1950
  • Country: us
Re: Isolated USB 3.0
« Reply #6 on: February 03, 2019, 01:38:21 am »
I can't find a spec for its isolation capacitance. They only include the default "isolation voltage", e.g. what the device can withstand without damage. Without isolation capacitance, you don't really know what "isolated" means.

I see they imply a DC/DC converter is used for power isolation. The spec sheets for those modules almost always specify their isolation capacitance, and it's generally not very good (the value is too high), so the effective isolation isn't as good as you'd expect. So if they've started with a module, this thing's actual isolation can't be better than a typical DC/DC module, which ain't all that great.

Disclaimer: I have a USB isolation unit made by Linear Tech, and it does provide some isolation, but nowhere near what any small signal Engineer would consider "isolated". You can definitely tell when it's in the circuit (though it's better than a direct USB connection).

There's still room for improvement in this area, for sure.
 

Offline Doctorandus_P

  • Super Contributor
  • ***
  • Posts: 4004
  • Country: nl
Re: Isolated USB 3.0
« Reply #7 on: February 03, 2019, 09:10:29 pm »
For both low capacitance and low EMI there's no substitute for Fiber I know of.

The simplest / cheapest is the S/Pdif audio at around 1.5Mb/s.
The fastest is over 400Gb/s
https://en.wikipedia.org/wiki/Fiber-optic_communication

In between there are many combinations of speed / price.
10Gb/s, 40Gb/s Ethernet and higher all can use optical.
SCSI sometimes uses optical.
There is a thing called Fibre channel.

I've seen optical fibers used in industry for programming PLC's with laptops.
I've seen separate IR emitter and receptor diodes seen in low noise equipment because regular optocouplers had too much capacitive coupling.

But I haven't done further research and have no idea of the price / performance ratio of those things, but there are a lot of generations of standards out there.


« Last Edit: February 03, 2019, 09:50:18 pm by Doctorandus_P »
 

Online Hydron

  • Super Contributor
  • ***
  • Posts: 1085
  • Country: gb
Re: Isolated USB 3.0
« Reply #8 on: February 05, 2019, 01:56:11 pm »
Ironically the USB2 (i.e. 480MBps) part of the USB3 connection is probably the trickier part to isolate - USB 1/2 is not purely differential plus is bi-directional and thus needs quite complicated circuitry to split up into isolate-able parts.
The USB3 bit itself is just 2 differential pairs - one in each direction - based (AFAIK) on the PCIE phy spec. You can probably feed them almost directly into something like a SFP+ (10Gbit) fibre transceiver, as it should already be scrambled (or 8b/10b encoded) to ensure DC balance and sufficient transitions for clock/data recovery.

If you see any USB3 _only_ isolators, this is probably why - the fastest interface is by far the easiest to deal with! I'm not really sure if most/none/all USB3 devices enumerate properly without the legacy USB1/2 connections present too - would need some experimentation.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf