2
\$\begingroup\$

Building a USB 2.0 speed cable is almost trivial. Simply connecting D+ and D- through any sort of wiring seems to allow full 480 MBps speeds.

Now I want to jumper a USB-C connector to a USB3.0 Type-A connector as shown. The distances involved are short as you can tell.

I can achieve USB 2.0 speeds no problem by connecting D+ and D- using jumper wire. Easy.

However, using twisted pair wires to jumper SSTX+/SSTX- to TX2+/TX2-, and SSRX+/SSRX- to RX2+/RX2- fail to allow the device to negotiate up to USB3 Superspeed. Note: The device still connects as a USB 2.0 device and operates properly at USB2 speeds.

If I connect the device directly to the computer, it properly negotiates USB3 superspeed and operates at USB3 speeds.

Yes, the same problem occurs if I use TX1 and RX1 - the orientation of the USB-C doesn't matter.

Help?

enter image description here


Note: The USB-C cable is actually a USB-A to usb-C superspeed cable. Yes, I have verified continuity on the breakout PCBs shown by connecting them to both ends of the same cable. Circuit continuity between the differential pair pins is verified.

\$\endgroup\$
  • 2
    \$\begingroup\$ Does your "twisted pair" have the required differential impedance (\$90\Omega \pm 7\Omega\$)? A mismatch, even if it's only a few inches long, can seriously degrade the waveform at 5 Gbps. Note that even Cat6 Ethernet cable, with a nominal impedance of \$100\Omega\$, falls outside this specification. BTW, you've had better luck than I have with high-speed USB -- I've had all kinds of subtle problems trying to pass it through various board-to-board and board-to-wire connectors. \$\endgroup\$ – Dave Tweed Aug 16 '18 at 13:11
  • 1
    \$\begingroup\$ Thanks. I'll try again with 90 ohm differential impedance wiring for the differential pairs. I am working with Appendix C of the USB3 spec to determine conformance to the spec. usb.org/developers/docs/devclass_docs/… \$\endgroup\$ – hype Aug 16 '18 at 14:26
  • \$\begingroup\$ What is the purpose of this exercise? \$\endgroup\$ – Ale..chenski Aug 17 '18 at 2:20
  • \$\begingroup\$ To jumper these two boards together. \$\endgroup\$ – hype Dec 14 '18 at 14:23
1
\$\begingroup\$

The high transmission line impedance of 0.1" spaced headers ( high L/C ratio) would cause significant group delay distortion within the bandwidth of USB-C 10Gbps that results in "eye-closure" or signal integrity issues.

\$\endgroup\$
  • \$\begingroup\$ Thanks. I'll try again with the 90 ohm differential impedance wiring soldered directly to the board. \$\endgroup\$ – hype Aug 16 '18 at 14:26
1
\$\begingroup\$

Generally you can't "jumpwire" USB 3.0 transmission channel. Even if the link is done using tightly controlled impedance using PCB technology, the results are frequently fairly disappointing, and the final routing requires several board spins and testing. Conformance to USB specs requires special perfectly-designed test fixtures that are difficult to procure, and susperfast (8 GHz - 16 GHz) test equipment that very few can afford.

You might be lucky to have a USB 3.0 device connected once or twice, but any movement/bend of the link will likely cause the handwired link to fail.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.