2
\$\begingroup\$

Are standard USB connectors Impedance controlled? Do they follow differential characteristic impedance of 90 Ohm for DM & DP lines?

Does standard Type A, Mini / Micro - USB Type follow same impedance match from PCB? Or Characteristic impedance changes at connector pins?

If they follow, how they manage to do so? As there is no ground / reference plane on USB connector to get this impedance.

\$\endgroup\$
4
  • 1
    \$\begingroup\$ I certainly hope so. You don't need ground planes to get a certain impedance, cables don't have a ground plane either. \$\endgroup\$
    – PlasmaHH
    Commented Oct 25, 2018 at 13:46
  • 1
    \$\begingroup\$ Pretty sure most USB cables have a ground shield surrounding the cable, though \$\endgroup\$ Commented Oct 25, 2018 at 13:54
  • \$\begingroup\$ I doubt it. Look at this SAMTEC datasheet on a thru-hole USB-B connector - no mention of impedance at all. That is likely why many parts (ATMega16U2) require a 22 ohm series resistor on DM+/. FT232 seems to address that internsally and does not require a resistor. In both cases, the IC data sheet suggests a short run for the traces, and equal lengths. I've always tried to lay out my boards that way, with ground plane under & around the traces, and one via maximum as the runs to an FT232 always seem to need to cross. suddendocs.samtec.com/prints/usb-b-s-x-x-th-x-mkt.pdf \$\endgroup\$
    – CrossRoads
    Commented Oct 25, 2018 at 14:12
  • \$\begingroup\$ @CrossRoads, cheap ATmega parts need the 22 Ohm series resistors (and some caps to ground) not because of transmission line or connector problems, but to align their internal FS (full-speed) cheap driver 25-ohm impedance and too-fast edge rate with USB standard. Some internal USB PHYs do have slew and impedance control, some cheap/old don 't. \$\endgroup\$ Commented Oct 25, 2018 at 21:00

2 Answers 2

2
\$\begingroup\$

The spec is not exactly 90Ω for connectors, but has a more loose requirement for impedance. Connectors that are built to the USB spec will have an impedance of the standard. The data lines inside of USB are twisted (or should be) which also makes an transmission line.

The way the manufacturers probably keep the impedance within spec is by modeling the parasitics and making a simple circuit. The pins to shell would form a capacitor and the pins have resistance and a small amount of inductance. If you do have a termination (like a connector, it is important to match closely but it doesn't have to be exactly matched, the goal is to minimize reflections and if you have an impedance close to 90Ω then that will be sufficient to maintain signal integrity.

The USB spec calls for a mated connector impedance of 75Ω to 105Ω:

Universal Serial Bus 3.1 Legacy Connectors and Cable Assemblies Compliance Document

Mated Connector Impedance (Differential)

75 Ω minimum, 105 Ω maximum. 50ps (20-80%) rise time of a differential TDR. (Mated connector includes cable termination areas). SuperSpeed pairs only.

Source: https://www.usb.org/sites/default/files/CabConn_Legacy_3_1_Compliance_Rev_1_1.pdf

\$\endgroup\$
2
\$\begingroup\$

Yes, USB connectors are designed to maintain 90 Ohm differential impedance. USB founding consortium includes lead engineers from several connector manufacturers (Foxconn, Hirose, JAE), who actively participated in definition and overall construction of connectors, trading off the cost and electrical quality.

However, any non-coaxial connector introduces noticeable imperfections into transmission line. The impedance depends not only on relative configuration of conductors and their crossection, but also on quality of surrounding plastics, like dielectric constant and loss. So not all USB connectors are made equal even if they maintain the same geometry of conductors. USB connectors don't have reference planes, but they do have "signal return ground" in near proximity to D+/D- and Rx/Tx +- wires. In extreme case, a twisted pair of conductors does have a defined differential impedance even if there is no ground plane around, look at UTP CAT5/6 network cables.

Good thing is that the actual mating length across the entire connector is under ~20 mm only, while the signal wavelength even at Super-Speed rate (5Gbps, 2500 MHz) in surrounding dielectric is about 60 mm (and ~600mm for USB 2.0 rates). So the effect on actual waveform is somewhat alleviated if the connector impedance is managed within the 70-105 Ohms range given in USB specifications.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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