6

I'm looking at making my own USB cable for a project. The primary issue I'm having is that I'm getting a lot of capacitance in my cable between D+ and D- (400ish pF) which is enough to throw off the signals being read by my pc. Is there an easy way to fix this? I understand this is an impedance matching issue, but I'm not sure where to begin with this one since the capacitance is between two input lines, not one line and ground which is what I remember from my electromagnetics course.

Shensmobile
  • 135
  • 1
  • 2
  • 6

3 Answers3

2

The impedance of the data line is determined by

  • capacitance between wires and
  • inductance of the pair of wires.

It's a simple formula and I've used it to make custom cables for high speed data links in circumstances where the wires and overall covering needed to be made from PTFE material to suit high temperature applications. The formula is

Zo = sqrt(L/C) - this is the impedance for a lossless line. Here is a good article.

This tells you that to make a higher impedance cable L needs to grow and/or C needs to reduce. One of the methods is to make the actual conductor a smaller diameter whilst keeping the spacing between conductors the same. Or you can make the spacing between conductors bigger. Below is a picture of an on-line calculator: -

enter image description here

Note that the formula given on the page above (Zo) is exactly the same as the one I gave earlier but it uses the more fundamental properties of wires to give L and C. In other words, permeability and permittivity are built into the equations but the 2nd formula explicity lists permittivity because that will be a variable due to the insulation type.

And here is the link to the webpage for the calculator. Please note that there are several calculators on-line so you can double check other ones.

From memory USB cable impedance is 90 ohms.

And if all this seems to much trouble, just make the wires thinner and space them apart a bit more and see what happens. If it is a Zo issue then things should begin to improve as you do this.

Andy aka
  • 456,226
  • 28
  • 367
  • 807
  • Hi, thanks for the great answer. My LRC meter is unable to detect a significant inductance for my cable (keeps giving me negative values!) so I assume that the bulk of my impedance in this cable is capacitance. I have already considered the idea of spacing out my wires, or reducing their gauge but I'm already at 28 AWG and they need to be as tightly packed as possible to reduce size of the cable. My other idea was to have a custom cable manufacturer use an insulating material that can prevent the capacitance from building up. – Shensmobile May 17 '13 at 14:39
  • I was also considering using an inductor to match the capacitor (this is what I remember doing from electromagnetics) but adding in another active component will only further delay my signal in the USB which I fear will cause problems with the signal travel. But aside from altering the physical properties of the cable (of which I'm fairly certain I can't change much more) or putting in an inductor, are those my only solutions? – Shensmobile May 17 '13 at 14:41
  • And USB termination is 90 ohms, which is in the downstream transceiver I believe. – Shensmobile May 17 '13 at 14:42
  • 1
    @Shensmobile you can't put any insulating material between the cores unless it has lower permittivity than the insulation you are currently using - this is the only way to reduce capacitance. It's a laws of physics thing - if you don't get data then fix the impedance by spacing the wires out more, reducing the insulators permittivity or making the wires smaller diameter. – Andy aka May 17 '13 at 15:46
  • Well the insulation I'm currently using is whatever standard plastic sheathing goes over conductor (I'm assuming its PVC) but a custom manufacturer might be able to insulate it with something better.

    Unfortunately I can't space the wires out more, or make the wires much smaller, so I think picking a different insulator would be my only choice.

     – Shensmobile May 17 '13 at 16:12
  • Why would a passive inductor be a problem? If this is for 12 mbit, It should be able to work. Add the same inductor type to both d+ and d- and try it. Another option is active usb amplification on either end. This is used for long usb2 (480 mbit/s) runs. – Jon Watte May 17 '13 at 22:23
  • 1
    @JonWatte It may work - it used to work on old telephone lines to keep the impedance at 600 ohm. From memory 88mH coils were spaced every mile and it linearized the impedance to 600 ohm across the speech bandwidth but above this the impedance fell rapidly. Do you know the length of the cable the OP has in mind and are you able to calculate the inductance value that would suit? USB amplification is a good option but it needs to be bi-directional I think - I'm not 100% on that one. If you have a proposal then make an answer - I'm not comfortable with either of your proposals but they could work. – Andy aka May 17 '13 at 22:37
  • This is gold, man! – Peter Mortensen Dec 12 '13 at 21:05
1

Well, normal USB cables have these shielding to decrease line capacitive effect. My initial guess: Why don't you try to shield them at first to see if the capacitance is the result of the physical cabling. Also having a twisted pair in USB cable might help to decrease the capacitance in between lines, see USB cable construction.

kunguz
  • 51
  • 3
  • Normal USB cables have shielding OUTSIDE the twisted D+/D- pair. That wouldn't help reduce the capacitance in between the two cables. I also have my custom USB cable in twisted pair arrangement right now. I understand that buying a "legit" 4 meter long cable would probably fix my problem but I want to know if there's any engineering principle I can use to apply a fix to this. I will need the custom USB cable for a high-temperature application. – Shensmobile May 17 '13 at 14:35
0

Thinking about it some more, I think it'd be hard to get a large enough inductor to counteract the capacitance, but still keep it small enough to not degrade the USB signal, especially at USB 2.0 data rates (480 Mbit.)

For long USB runs, if you can't alter the physical construction of the cable sufficiently, then active amplification is needed. And, yes, it has to be bi-directional. You can actually buy ready-made cables with built-in amplification, and you can also buy chip sets that you can add to your cable.

Jon Watte
  • 5,670
  • 27
  • 37