0

I have a board design that uses an Amphenol rugged USB-C socket to feed a Microchip USB5744 USB3 hub. When laying out the tracks, it struck me that the 'unused' Superspeed inputs are like stubs on the used pins and that this sounds like a very bad idea. As I can't see an alternative to these tracks, can someone assure me that they don't cause a problem?

JYelton
  • 32,302
  • 33
  • 134
  • 249
Currents
  • 1
  • 1

1 Answers1

1

The Superspeed pins are only 'unused' if your application has no intention of using them, i.e. you're happy with USB 2.0 data rates. If that's the case then you can just leave them unconnected without a problem - the Host will determine they're not being used and never assert them.

td127
  • 2,932
  • 6
  • 13
  • I'm referring to the short transmission lines that join socket pins like SSTX side A to SSTX side B. If the input cable is using SSTX side A, it will see the open circuit lines to SSTX side B as stubs. Surely these O/C lines can cause a reflection problem? – Currents May 30 '23 at 13:14
  • 1
    "If the input cable is using SSTX.." - but you said your application isn't using Superspeed, so these lines will never be asserted by the Host, so there is no reflection problem because there is no active signal on them. – td127 May 30 '23 at 18:32
  • I don't recall saying that I wasn't using SuperSpeed - I think that was an unfortunate way of translating the 'unused pins' statement. I only meant that two TX pins are not being used when the other two are active. i.e SSTX side A is running signals while SSTX side B is floating free, except for the transmission line stubs from SSTX side A. – Currents May 30 '23 at 19:48
  • Ah, I see - thanks for the clarification. Well, I think the answer is that the SS A side and SS B side are not physically tied together so there shouldn't be any stubs. On the device side they come into separate pins on a multiplexer (which is switched based on cable orientation). The SS drivers should only be driving one set of the wires and the other set should not only be floating but physically separate from the driven pair. – td127 May 30 '23 at 21:29
  • On all the board layouts that I have found, the A side and B side pins are tied together by transmission lines - otherwise you couldn't drive a device with a single USB3 input from either side of the socket. It just looks like a 'bad idea' to have these lines going nowhere. I assume that the cable doesn't terminate them with a resistive matched load? – Currents May 30 '23 at 21:42
  • That doesn't sound right. The USB2 pins ARE tied together top to bottom - A6 to B6 and A7 to B7. But the Superspeed pairs have separate connections. When you flip the cable over a hardware multiplexer is invoked to tx/rx the other pairs. Since the USB2 lines are doubled up top and bottom no multiplexer is required for those. – td127 May 31 '23 at 02:56
  • This is a layout that I found on this web site. https://electronics.stackexchange.com/questions/299196/properly-route-through-hole-usb-c. You will see that the SSTX lines are joined on opposite sides of the socket, as are the SSRX lines (the opposite side connections are through vias at the top). – Currents May 31 '23 at 07:03
  • 1
    Ugh, no wonder you have concerns. That's a completely bogus implementation. Check out the datasheet for a superspeed multilplexer, which has the correct wiring diagram for USB 3.0. Do a web search for OnSemi's NL3HS3124A datasheet. – td127 May 31 '23 at 15:57
  • Many thanks for the suggestion - it does look like a good option. – Currents May 31 '23 at 18:22