Author Topic: Max Cable Length  (Read 10903 times)

weberkr

  • Member
  • ***
  • Posts: 1
Max Cable Length
« on: February 22, 2013, 02:11:41 pm »
I'm trying to understand the 5 meter cable length requirement. What is the failure mode?  Nothing I've found really explains it. Everything I've read says don't exceed this requirement, but I've got an engineer that has it working over 30 ft of (Teflon) cable. I'm not trying to extend with any hubs. I'm just concerned with a host to one device. I’m really looking for the engineering, detailed, technical reason.

I think I've decided that it's really dependent on the 30 nSec one way delay requirement (26 nSec dedicated to the cable). Does the fact that it's not terminated at the receiver in the cable impedance (90 ohm) create reflections/standing waves and it just takes that long for it to settle out at the receiver?

Jan Axelson

  • Administrator
  • Frequent Contributor
  • *****
  • Posts: 3033
    • Lakeview Research
Re: Max Cable Length
« Reply #1 on: February 22, 2013, 02:30:02 pm »
You likely know all of this, but the cable spec is conservative to help ensure that the signals make it to the receiver without error. That doesn't mean you can't use a longer cable and mostly get away with it. Also, a marginal cable might "work" only due to the retries built into the spec.

The 5-meter limit applies only to full and high speeds and low speed traffic upstream from a hub (because hubs uses full-speed's edge rates for low-speed traffic). Low speed device cables are 3m max, and SuperSpeed is approx. 3m max.

Both full and high speed have terminations that reduce reflections at the receiver, but the terminations are at the driver end (driver impedance + additional series termination).

The length limit for full and high speed is due to signal attenuation, cable propagation delay, and voltage drops on VBUS and GND. Which of those is most likely to cause a problem on longer cables I couldn't say. 




Bret

  • Frequent Contributor
  • ****
  • Posts: 68
Re: Max Cable Length
« Reply #2 on: February 23, 2013, 11:00:02 am »
This is actually comparable to what happens with RS-232 technology.  The official RS-232 spec says 20 kbps at 50 feet, but as anybody who has actually used the technology knows, you can actually achieve far greater specs than these.  There are lots of technical factors (impedance/capacitance, frequency-dependent characteristics, voltage, current, shielding, timing, "quality" of the hardware and the design, etc.) that can affect it.

As Jan said, the numbers are conservative to allow for even inexpensive cables to work.  So, even though it may work with a long cable in one situation (especially when things are "new"), doesn't necessarily mean it will work in some other situation.