[Prev][Next][Index][Thread]

Re: Cable Extension

Armitage wrote:
> 
> On Wed, 9 Jul 1997, Hanno Mueller wrote:
> 
> > I doubt that 100 feet is possible with any kind of cable extension.
> > Connectix says that 3 feet is allowed, I was able to use a 6 feet
> 
> <Dredging up old knowledge>
> 
> This is what I remember for Circuits 1 in college, so... [By +/- 1, I
> mean 1 * the max voltage, not 1v]
> 
> The problem with using long cables (especially with a digital signal)
> is that the lengths of wire develop capacitance, and the switching
> time degrades. So your signal which should look like:
> 
> +1       ______    ______
>          |    |    |    |
>          |    |    |    |
> -1 ______|    |____|    |_____
> 
> Looks like:
> +1         __        __
>           /  \      /  \
>          /    \    /    \
> -1 _____/      \__/      \____
> 
> If your system receiving reads +1 and -1, then the periods at which
> the voltage is high enough to register is much shorter. Then longer
> the cable, the greater the capacitance, and the shorter the time
> period. If the cable gets long enough, you won't even be getting +1 or
> -1, rather +0.9, +0.5, or whatever. If your receiver registers at +/-
> 0.1, you'll be ok with a longer cable, but most hardware won't
> register that low.
> 
> This is the same reason that your SCSI, 10BaseT, serial, monitor, etc.
> cables shouldn't be too long.
> 
> -B
> 
> If there are any facts wrong, lemme know. I haven't dealt with this
> for years.
> 
> Brandon High                          http://users.ccnet.com/~bjhigh
> bjhigh@ccnet.com
> --
> A power so great, it can only be used for Good or Evil!

Actually the cable looks like series inductors and resistors and
parallel capacitors - elementary electronics/electrical engineering.

This causes phase shifting at higher frequencies as the impedence is not
necessarily uniform at all frequencies. Phase shifting is the real
limiting factor on long cables. After a while the third harmonic gets
far enough out of phase to look like a separate signal. Balancing the
line fixes this to some extent (no - parallel cables aren't balanced),
as do a variety of other tricks. Plenty of current will also help
(charges the capacitance(?) faster) so the separate power supply will
extend the available cable length, but eventually the phase shift will
get you.

DEC had a better solution with their parallel interface, but it cheated
by using proper line drivers with lots of current. You could run those
cables a long way.

Have a look on a CRO. Long cables have messy signals at the end.

Anyway this starting to remind me of the joke about Microsoft employees
and light bulbs, so I'm going back to software. See next e-mail for my
real interests.

RJM
References: