What makes cables suitable or unsuitable for a given purpose?

Question

I've been working on a homebrew set of VR Goggles w/ head tracking, and am just beginning to plan the cable that will connect the parts on the user's head to the electronics on the main unit (which will be mounted under & behind the seat the user will be seated in) I'm expecting about a 4ft cable run between the two units, and the connection will need somewhere between 6 and 10 wires (depending on how many ground wires I share):

Manditory wires:

1. +12v for the video display in the goggles
2. GND
3. I2C SDA (3.3v)
4. I2C SCL (3.3v)
5. Composite Video
6. Mono Audio Optional wires:
7. +3.3V for the headtracker sensors (Could be obviated with an LDO regulator from the 12v
8. GND (paired with +3.3v)
9. GND (Video-)
10. GND (Audio-)

Other requirements:

• the cable needs to be flexible, light weight, and possibly coiled to handle head movement without discomfort or tangling.
• the cable needs to be commodity (this is a personal project)
• the cable needs to be reliably disconnectable/reconnectable

I think I understand the various factors in play (crosstalk, emf, capacitance, etc.) but I don't have any perspective on their importance for a 4ft cable run.

My back-of-napkin guess would be to use an ethernet cable (CAT 5), wired like this: Pair 1: Audio+/- Pair 2: Video+/- Pair 3: I2C SDA/SCL Pair 4: Power +12V/GND

(Use an LDO for the 3.3v on the sensors.)

Is this sensible? What factors should I be most concerned with? Which signals will be most impacted by connectors and wiring choices?

Answer 1

Your main considerations should be as follows for your cable:

• Flex life, i.e. how long the cable can be flexed before it breaks (very important for your application)
• Shielding/grounding for the composite video portion and other low-level signals
• Impedance for signals with content above 16 MHz (see calculation below)

For a cable run of 4 feet and assuming a typical velocity of propagation of ~66% for the cable, the one-tenth wavelength rule says you can ignore transmission line effects below about 16 MHz. I don't know enough about the frequency content of your I2C lines to give a recommendation there, but analog baseband composite video will have frequency content below 8 MHz in all regions of the world.

If you decide to go with a Cat 5E Ethernet product, try to find one designed for continuous flexing. Those types of products are used in industrial robotics products.

For the composite video portion, you have two choices:

1. Make a composite cable (cable made up of multiple elements) yourself with a coaxial element and your Ethernet element.
2. Use a 100-ohm twisted pair from the Ethernet with a balun.

If you decide to make a composite cable yourself, find a 75-ohm video product designed for continuous flexing (long flex life). Combine it with the Ethernet element in a braid.

To maximize your flexibility and the flex life of your finished product, you want to use flexible materials (soft plastic and many strands of copper) and construct things in such a way as to minimize the twisting and untwisting of the internal elements.

I'm a former applications engineer for a wire and cable company. Here is a link to a general-audience presentation about wire and cable: http://www.belden.com/resourcecenter/cablebasics/upload/Cable-101.pdf

And here is a link to industrial products in a wire and cable catalog: https://www.belden.com/docs/upload/Cabling-Solutions-for-Industrial-Applications.pdf

This 2002 book is also a good overview of the audio/video cable industry, even if it is somewhat outdated: http://www.amazon.com/Audio-Video-Installers-Pocket-Reference/dp/0071386211

Answer 2

The video's most likely to pickup noise from the digital signals, audio at headphone power level won't pick up or interfere with any of those signals noticeably. using a separate shielded cable for the video could be a win but you may find that an off-the-shelf video balun works well enough.

Answer 3

To be honest I don't think you will have much of an issue with simply using the Cat 5 other than flexibility. The wires inside are solid copper so they are rigid and prone to wear after too much flexing. The frequencies those signals (like the I2C) will likely be running at shouldn't cause an issue. I'd only be concerned about power draw which I'm assuming will probably be less than 250mA in this case? That's about the max depending on the gauge of the strands in your Cat 5 cable in particular.

Also, the twisted pairs provide isolation because they are somewhat perpendicular, which means less magnetic field is induced in its fellow pair.