dead USB ports too frequent for my liking

I have had 20 PCs since USB came out, and at least half of them had 1
dead USB port after a while. In only one case could I see bent contacts
in the socket. So I wonder is this the usual problem, or do the USB
chips get zapped?
I presume the sockets 5 V pins are fed from a common rail, so it
would be the data lines that died.

wrote:
I have had 20 PCs since USB came out, and at least half of them had 1
dead USB port after a while. In only one case could I see bent contacts
in the socket. So I wonder is this the usual problem, or do the USB
chips get zapped?
I presume the sockets 5 V pins are fed from a common rail, so it
would be the data lines that died.

Buy yourself a pair of those conductive ESD shoes ?

I've lost zero USB ports here.

One of my secrets, is I never wire up USB front
panel connectors. I use the rear connectors
exclusively. That's why they don't fail.

Paul

On Sun, 14 Jun 2015 02:48:18 -0400, Paul wrote:

| wrote:
| I have had 20 PCs since USB came out, and at least half of them had 1
| dead USB port after a while. In only one case could I see bent contacts
| in the socket. So I wonder is this the usual problem, or do the USB
| chips get zapped?
| I presume the sockets 5 V pins are fed from a common rail, so it
| would be the data lines that died.
|
| Buy yourself a pair of those conductive ESD shoes ?
|
| I've lost zero USB ports here.
|
| One of my secrets, is I never wire up USB front
| panel connectors. I use the rear connectors
| exclusively. That's why they don't fail.

I connect front panel ports to use for USB flash drives and such and have never had
one of those fail.

Larc

On 14/6/15 11:59 AM, wrote:
I have had 20 PCs since USB came out, and at least half of them had 1
dead USB port after a while. In only one case could I see bent contacts
in the socket. So I wonder is this the usual problem, or do the USB
chips get zapped?

Are they branded PCs? What brand?

--
@~@ Remain silent. Nothing from soldiers and magicians is real!
/ v \ Simplicity is Beauty! May the Force and farces be with you!
/( _ )\ (Fedora release 21) Linux 4.0.4-202.fc21.i686+PAE
^ ^ 22:42:01 up 7 days 2:15 0 users load average: 0.00 0.01 0.05
�借貸! ��騙! ��交! �打交! �打劫! �自殺! 請考慮綜� (CSSA):
http://www.swd.gov.hk/tc/index/site_...sub_addressesa

On Sun, 14 Jun 2015 10:10:46 -0400, Larc wrote:

I connect front panel ports to use for USB flash drives and such and have never had
one of those fail.

Neither have I. An uncle of mine /did/, but I haven't encountered in on
other PCs.

--
s|b

On 14 Jun 2015, "s|b" wrote in
alt.comp.hardware.pc-homebuilt:

On Sun, 14 Jun 2015 10:10:46 -0400, Larc wrote:

I connect front panel ports to use for USB flash drives and such
and have never had one of those fail.

Neither have I. An uncle of mine /did/, but I haven't encountered
in on other PCs.

There's no real reason the front ones should fail any more often than
the rear ones, except that they're more exposed to being knocked
around. USB connectors are a bad design in the first place and they
don't resist physical abuse very well.

On Sat, 13 Jun 2015 20:59:20 -0700 (PDT), wrote:

I presume the sockets 5 V pins are fed from a common rail, so it
would be the data lines that died.

Measure them or use a flashstick w/ a data activity LED. If it lights
up could be more involved, possibly a PWR and not a data strobe feed.
Then measure the backplane pins or MB USB headers for divergence.

If not, buy a same gender extender or a PCI USB card, be more gentle,
or check your ancillary USB devices for physical compliance if
possible.

Nil wrote:
On 14 Jun 2015, "s|b" wrote in
alt.comp.hardware.pc-homebuilt:

On Sun, 14 Jun 2015 10:10:46 -0400, Larc wrote:

I connect front panel ports to use for USB flash drives and such
and have never had one of those fail.
Neither have I. An uncle of mine /did/, but I haven't encountered
in on other PCs.

There's no real reason the front ones should fail any more often than
the rear ones, except that they're more exposed to being knocked
around. USB connectors are a bad design in the first place and they
don't resist physical abuse very well.

There is a reason, but I didn't want to write it up.

It has to do with the plastic front on the computer.

When you dump ESD into a front panel USB, it can't
"drain" into the chassis directly. The ESD discharge
goes down the front panel USB cable, down the ground
of it. And while doing so, the wire is in parallel
with D+ and D-, inducing high voltage in those leads
too.

Whereas the rear connectors are beautifully designed.
The area around the connector has metal, the metal makes
contact with the chassis. The spring fingers on the I/O plate,
provide a conduction path between the chassis and the connector
body. The "parallel" run of wires in that case, is very short.

If you're lucky enough to have a PC with a metal fascia
on the front, then there would be an opportunity for some
of the ESD energy to go into the chassis, instead of
near the wires.

Using plastic on the front of the machine, allows the
manufacturer to make complex aesthetically pleasing
shapes with minimal work. If the front was
fabricated out of metal, the box would likely
look more ugly that it already is.

ESD protection for equipment is a complex subject. Some
designers, their first approach is to insulate everything
on the front, if possible. (So the user can't zap anything.)
If you're putting I/O connectors on a surface, that idea
is ruined right away. The next approach after that, is to
"drain" to chassis. Even this can be problematic in terms
of upset. We had a system, where every time there was the
least contact between computer metal and metal chairs or
objects in the room, the box would reset. And there was an
actual reset wire inside the computer, picking up energy by
induction. Once that was improved, the problem stopped. I
got a free plane flight to a customer site, to observe that
one happening, and it was plenty freaky. The amount of signal
required to upset the computer was amazingly small. You could
fart, and that computer design would reset itself, and all
because some clever person left a long run of reset wire
(no twisted pair) inside the box.

We test systems like that at work, with an ESD probe
modeling the Human Body Model. It includes an RC time
constant, and you charge to "X" volts and do your test.
The equipment will have "upset" and "damage" ratings,
at two different voltages. So maybe the spec says
it cannot reset if hit with 5kV of discharge, or
it cannot be damaged if hit with 10kV. The probe has an
insulated handle, and a rounded tip to avoid corona.
And then you bring it up to the equipment and zap it,
and test whether your protection design is working.
We had *many* failures with the "metal draining" method,
requiring analysis of where the discharge was going. In
some cases, spring fingers being added to some metal things,
to encourage high voltage discharge to go down a
certain path.

I never had to get involved in that stuff for what I was
building (never ran the HV probe myself), but I was around
while the other staff were working on it (multiple projects).
So got to discuss the approach they were using. Fun stuff.
And we never had incidents of staff zapping one another with
the probe. True professionals :-) Just to prove the probe
was actually modeling HBM, and gave a human amount of
static-like discharge.

Intel USB ports are rated for around 5-6kV of ESD (written up
in the Intel Journal). RS232 ports (good ones), are rated
for 15kV, and those were the first chips to be recognized as
having a problem, and the redesign of those was excellent.
Regular chips (where the wires don't go near humans) are
rated for 1kV or so. For third party USB, like NEC brand,
I have no idea what their rating is, but it's probably not
as high as Intel. Some PCI Express slots are extremely sensitive,
and could well be 1kV or less (there were some early PCI Express
designs, where customers blew out the PCI Express slot during
their equipment build). On those systems, once alerted to that
fact, you'd want your ESD strap while assembling computers
based on those boards. Since I'm no longer hearing that
complaint, I guess the chipset makers have figured that out.

It's possible to add protection networks to electrical items
externally. But this topic is more art than science, and the
vast majority of available solutions for it are crap. I
worked with a gentleman, who tirelessly tested this stuff.
Even when not tasked with a design, he would get samples
of components like this and test them. The critical parameter
is stray capacitance, and good components (1pF or less) are
hard to find. Some of the good solutions, were coming
from small companies (startups), and not from the
larger suppliers. You cannot put these on high
speed signals, unless the stray capacitance is
extremely low. Which rules out most of the ones
you see advertised.

http://www.eetimes.com/document.asp?doc_id=1275127

Paul

On Sunday, June 14, 2015 at 10:51:19 PM UTC+8, Mr. Man-wai Chang wrote:

Are they branded PCs? What brand?

A few were HP or Dell. Mostly they were small PC shop jobs: Intel, Asus or Gigabyte boards in Antec, Coolermaster or Thermaltake case.

I have 2 cases with metal fronts: Silverstone and LIANLI. The LIANLI
has a dud USB port, the Silverstone has been trouble free.

On 14/6/15 11:59 AM, wrote:
I have had 20 PCs since USB came out, and at least half of them had 1
dead USB port after a while. In only one case could I see bent contacts
in the socket. So I wonder is this the usual problem, or do the USB
chips get zapped?

I have never fried a USB port before, NOT ONCE.

But then my PC is not in public space.

--
@~@ Remain silent. Nothing from soldiers and magicians is real!
/ v \ Simplicity is Beauty! May the Force and farces be with you!
/( _ )\ (Fedora release 21) Linux 4.0.4-202.fc21.i686+PAE
^ ^ 23:09:01 up 7 days 19:57 0 users load average: 0.00 0.01 0.05
�借貸! ��騙! ��交! �打交! �打劫! �自殺! 請考慮綜� (CSSA):
http://www.swd.gov.hk/tc/index/site_...sub_addressesa