|
 The
two common USB standards differ greatly in terms of transfer rates. USB
1.1, which has been around for many years, offers transfer rates of up to
12 mbit/s. USB 2.0, which only came out 18 months ago, is growing in
popularity thanks to its transfer rates of up to 480 mbit/s.
Since the USB standard was
designed to be an interface for all kinds of peripherals, it was planned
from the very beginning that devices would be run on USB hubs. All you
have to do is hook up the hub to an USB port on your computer and voila!
You now have several more available ports. The advantage of this set-up is
that you can run more devices than you have connectors. There is a
downside, though: you may encounter problems when using high-performance
USB devices because the bandwidth has to be divided up among the different
devices.
To understand why
performance problems are not uncommon, you need to know something about
the USB protocol. USB devices can access any of the four sub-protocols:
bulk, control, interrupt and isochronous. They help allocate the
"attention span" (read: bandwidth) a device needs to operate.
No matter what else
happens, 10% of the bandwidth is reserved for the control protocol, which
directs all the transfers. Audio or video devices such as webcams or
speaker systems work isochronously ("in real time") because they
always need a minimum amount of bandwidth. Most external storage devices
use the bulk protocol, while USB keyboards or mice avail themselves of the
interrupt protocol.
You've probably already guessed the
problem: whenever you connect two USB devices that both use the
isochronous protocol and take up a certain amount of bandwidth, you'll
have very little left over for any other devices. USB is a
first-come-first-served standard, so if you connect a third or a fourth
device to the hub, you may not have enough bandwidth left over to operate
it.
a USB cable’s peripheral end and
B&B’s product, a USB-to-RS-485 converter. A metal shield surrounds
the cable connector’s plastic housing, which in turn surrounds the two
data lines and two power lines. If you plug the cable into most USB
peripherals, though, part of the outer shield remains exposed.
The exposed metal shield will pass current
from direct ESD hits into both the peripheral device and to the
peripheral’s host PC or USB hub at the other end of the cable. When Banh
injected a ±4 kV contact ESD pulse into that exposed shield, the
resulting current traveled through the cable to the host PC, whereupon the
PC’s software froze. Banh had to unplug the USB device and reboot the
computer to recover from the error.
Engineers at B&B had essentially two
options for increasing a product’s immunity to direct ESD hits. They
could add shields, filters, and grounds to the product or cut off ESD at
the source. Banh and others chose to mechanically isolate the USB cable
connector’s shield from any possible direct ESD hits. They recessed the
USB connector into the black box (Fig. 1), so the plastic case will
protect the cable connector’s shield from ESD.
Any exposed metal makes a product
potentially susceptible to ESD. The black box in Figure 1 has an RS-485
terminal strip at the opposite end (not shown in Fig. 1). The terminal
strip has metal contacts. To eliminate ESD hits on those contacts, B&B
engineers protect the terminal strip by placing it behind the plastic
door.
Indirect Hits
An ESD event radiates energy that can
couple into a circuit through cables, connectors, and PCB traces. Both
Banh and Locke had similar problems on USB cables from indirect ESD; after
an ESD event, their computers would no longer recognize the USB
peripheral. Locke’s product, the custom control panel contains a
USB trackball, a USB keyboard, several USB switches, and a USB hub all
mounted in one housing. The input devices connect to the hub, and a USB
cable connects the panel to a PC. Because the panel’s components were
enclosed in a plastic case, the case protected the panel’s electronics
from direct ESD hits.
Indirect
ESD hits can cause problems, too. During initial prototype testing, Locke
found that a host PC no longer recognized the panel’s components after a
±3-kV discharge into his test setup’s ground plane. The panel uses a
shielded USB cable, as required in chapter 7 of the USB specification.1
Locke had to find out how current from the radiated EMI was entering the
USB cables and traveling to the PC, and then find a way to make the
product immune.
Immunize
Bypass capacitors can help immunize a
product from induced current. Capacitors between a USB product’s D+ and
D– data lines and ground on a PCB will divert high-frequency current to
ground. Unfortunately, Section 7.1.6.1 limits the total capacitance of a
capacitor, the line driver’s output, and the PCB traces between the two
to 100 pF. Capacitors on the D+ and D– data lines can improve ESD
immunity, but too much capacitance may violate the USB spec and compromise
signal integrity. Locke did add capacitors to the data lines in his
product. The capacitors shunted some of the ESD-induced EMI current to
ground, which reduced data errors while keeping the product within the USB
specification’s capacitance limits.
In some applications, ferrite beads around
cables can reduce common-mode currents that disrupt a product’s
operation. The USB spec discourages the use of ferrite beads because they
may slow a data signal’s edges to where a USB device no longer
recognizes bits. Be aware, though, that Intel’s EMI Design Guidelines
for USB Components suggests using ferrite beads as a method for
reducing interference.2 According to Locke, Intel’s design
guideline (which is undated) was written early in USB’s life and you
shouldn’t use ferrite beads.
Guard traces on a PCB can also improve ESD
immunity. A USB product’s guard traces, which connect to ground, isolate
the sensitive data lines from radiated emissions. As rules of thumb, Banh
and Locke also recommend:
• use as much PCB area as possible for
power and ground,
• connect USB shells to ground,
• place oscillators as far away from the
USB data lines as possible, and
• position USB connectors as far as
possible from the peripheral’s USB controller chip.
While the solutions to ESD problems on USB
devices seem easy, both Banh and Locke point out that ESD troubleshooting
is a trial-and-error process that can take weeks to perform. Most products
go through several iterations before they get the right value components
that solve ESD problems. Keep in mind that products need a complete
functional test and an ESD test after each design change. T&MW |
