5

On PC motherboards you commonly see 0.100" headers used for USB 2.0 connections, often for a cable running to ports on the front of the PC. Here's an example.

https://www.digikey.com/en/products/detail/tripp-lite/U024-003-5P-PM/7243242

USB cables are supposed to have a differential impedance of 90 ohms +/- 15%. Maybe those headers just happen to be close to 90 ohms, but I doubt it. Even if they are, terminating the wires to the headers requires untwisting the D+/D- twisted pair for some short length, and I'm sure that is a significant deviation from the impedance requirement.

Does this ever lead to signal integrity problems, or is it not an issue for USB 2.0 as long as the rest of the path is well designed?

user44627
  • 53
  • 4
  • This topic has been discussed here at significant length, https://electronics.stackexchange.com/questions/387223/standard-2-54-1-27mm-pinheader-impedance/387226#387226 – Ale..chenski Jun 26 '23 at 04:50

1 Answers1

4

The uncoupled or mismatched distance is small compared to the pulse time (for USB High Speed, a hair over 2ns), so doesn't affect signal quality much.

More importantly, the shield does not surround the pins in such a header. The uncoupled distance means external noise (that would otherwise be carried on the outside of the shield) is permitted into the pairs (or wires individually). How much, depends on the frequency and intensity of the interference, and the uncoupled length. Typically, a few inches is enough to fail ESD immunity, for example (give or take how well the comm channel auto-recovers from interruptions like that).

Inside a PC, the metal enclosure deflects the brunt of interference such as ESD, and mostly the noise environment is that generated by the motherboard and peripherals. So the tolerable unshielded length can be larger.

Tim Williams
  • 22,874
  • 1
  • 20
  • 71
  • Very helpful. Thanks. To clear one thing up, you mentioned ESD (Electrostatic Discharge) immunity, but I would have thought that the shield would be more relevant for blocking EMI (Electromagnetic Interference). Could you comment on that? – user44627 Jun 22 '23 at 13:36
  • ESD is a kind of EMI. While susceptibility is tested with a modulated CW waveform, ESD and EFT are impulsive and much higher in amplitude (~kV, vs. 3V conducted or 3V/m radiated for commercial levels). Even a highly effective shield say 60dB (1/1000th ratio) won't prevent 10kV ESD from interfering with the signal (USB has a common mode range of, at most, the 3.3V supply). Either very good shielding is required, or a protocol capable of tolerating or recovering from such interference. – Tim Williams Jun 22 '23 at 14:30
  • For the sake of accuracy, USB2 FS mode has the bit-time of 83ns. 2ns is for high-speed (480 Mbps) mode. – Ale..chenski Jun 26 '23 at 04:46
  • Oops, thanks for the correction @Ale..chenski ! – Tim Williams Jun 26 '23 at 05:08