PC said:
Is there some reason that a panel-mounted USB cable like this:
http://m4.sourcingmap.com/photo_new/20090106/c/ux_a09010600ux0042_ux_c.jpg
couldn't be made with ribbon cable?
Assuming that there are female USB connectors designed for press-fit
ribbon cable?
You could make up a panel-mounted connector solution in 2 minutes using
ribbon cable.
?
There is a 650 page spec for USB2.
"High-/full-speed cable consists of one 28 to 20 AWG non-twisted
power pair and one 28 AWG twisted data pair with an aluminum metallized
polyester inner shield, 28 AWG stranded tinned copper drain wire,
65% tinned copper wire interwoven (braided) outer shield, and PVC
outer jacket."
That means the power wires are a different gauge than the signal wires.
Good USB cables use thicker power wires, to meet the VBUS drop budget
at 500mA current flow.
If you used ribbon cable, you'd need to use something like teflon ribbon,
to meet a roughly 90 ohm differential impedance on D+/D-. They do make
custom ribbon cables, including SCSI ribbons with things like twisted
pairs right in the ribbon. Any cost savings would go out the window,
when you price a roll of cable like that. With teflon, you might be
able to avoid the need for twisted pair, and just run a pair flat.
Setting the number of twists per inch, helps tune the impedance value,
and twisting also helps if you have true differential receivers on
the end (which the USB ones won't be - electronics are notorious for
not using amps with good common mode rejection).
That's what I'm looking at when I talk about "lead dress". The
D+/D- pair are supposed to remain 90 ohms, and won't be when
untwisted and sitting in free space. The "uncontrolled"
part of the cable (with shield removed, and the shield
changes the impedance) should be as short as possible, to
reduce reflections.
A guess would be, yes, you could make a cable which is flat in
nature. They even make ribbon cable with external shielding.
So the USB cable could be "flattened", if there was
a requirement for it (round cable wouldn't fit through a
hole, say). But as a "quick and dirty", it might not be
close enough, and might cause a detectable error rate.
The cabling spec has diff TDR tests to meet, which you could
subject the test ribbon cable to, to determine whether it
is close enough or not.
On USB3, the shield is apparently not enough, and an Intel
document shows USB3 cables throw off enough microwaves,
to kill adjacent Wifi signals. Peripherals require careful
shielding, to stop that from happening. Shielding that
you know peripheral makers aren't using, or testing for.
The only consequence, is a "surprise" when the laptop
Wifi doesn't work, and you have to "move the cable this
way and that way", until Wifi or an adjacent wireless
keyboard/mouse start working. I don't know if Intel
wrote this document as FUD, to push Thunderbolt, or
it's an honest effort to warn the industry about
more care being needed with 5Gbit/sec USB3 signals
and their associated broadband spectrum. USB3 has
even more extreme requirements, a higher allowed VBUS
current. The three diff pairs would probably be good
candidates for the teflon ribbon (with a ground wire
between each pair), but again, you'd need to check
for crosstalk and reflections with TDR and other tests.
The ribbon would likely "radiate like a bastard". You
can get doubly shielded ribbon, with rubber on the
outside, and that is thick enough, you might as well
go back to the round cable again
As for insulation displacement connections, for the
higher speed cases you'd want to investigate how
that affects lead dress, and how much unshielded
cable area is present as a result. If the ribbon is
shielded, you have to figure out how to connect
(terminate) the shield/drain at the end. Overmolded
cables hide all those details, so we can't see how
it is done.
And most of this stuff is outside my pay scale. "If I
made your cables from scratch, they'd probably suck."
But I have worked with people who do this stuff.
They have the TDR, network analyser, test head,
to do the characterization.
Paul
