USB device shuts down PC

[Neil]

Hi,
I have a PC running Windows XP home. I have both front & rear USB ports.
If I plug a printer into any of the USB ports, the PC shuts down with out
warning.
If I plug a USB memory stick into any of the ports, the same thing happens.
I have a USB extension lead, if I plug the extension lead into any USB port
and then plug the memory stick into the extension it works OK.
I have recently installed a PCI USB card, the computer still shuts down when
devices are plugged in.

Any suggestions?

Thank you.

 

[Rich Barry]

Neil, have you done a check on your Power Supply voltage values??
Don't know the age of your computer but Everest Software and Motherboard
Monitor gave a fairly
accurate read out of your PSU voltages.

 

[Leythos]

Hi,
I have a PC running Windows XP home. I have both front & rear USB ports.
If I plug a printer into any of the USB ports, the PC shuts down with out
warning.
If I plug a USB memory stick into any of the ports, the same thing happens.
I have a USB extension lead, if I plug the extension lead into any USB port
and then plug the memory stick into the extension it works OK.
I have recently installed a PCI USB card, the computer still shuts down when
devices are plugged in.

Any suggestions?

If you mean, just powers down, no warning, no sign of it doing a proper
shutdown, then you have a Power Supply problem.

You don't mention your computer vendor/model, PSU size, number of
components running off the power supply...

Go buy a quality 600W PSU and you should, based on the limited info
you've provided, be fine.

 

[Paul]

Hi,
I have a PC running Windows XP home. I have both front & rear USB ports.
If I plug a printer into any of the USB ports, the PC shuts down with out
warning.
If I plug a USB memory stick into any of the ports, the same thing happens.
I have a USB extension lead, if I plug the extension lead into any USB port
and then plug the memory stick into the extension it works OK.
I have recently installed a PCI USB card, the computer still shuts down when
devices are plugged in.

Any suggestions?

Thank you.

Modern computers sometimes choose to power the motherboard USB
connectors from +5VSB. If you read the label on the side of the
supply, the 5VSB output is rated for about 2 amps. That is only
enough current, to power four USB devices loading to the max
of 500mA each. (And there are other loads in the design, on the
+5VSB rail, so the 2 amps cannot be used entirely for USB.)

The PCI USB card, could power its ports from the regular +5V port.
If so, its ports may behave differently.

You may get some idea, which rail is used for the ports, if a
USB device with a LED on it, continues to operate the LED when
the computer is in sleep state. (A USB LED reading lamp is another
device you could use to check it.)

If +5VSB is momentarily overloaded, that can be enough to cause
the PS_ON# control signal to be deasserted, which in turn causes
the power supply to go off.

A second reason for +5VSB to glitch, is if the proper precautions
are not observed in the design. Intel recommends that a bulk capacitor
be placed near the USB port. The purpose of that capacitor, is to
provide current, to compensate for the "inrush" current that flows
into a USB device freshly plugged in. The USB device itself, should
not use gobs of its own bulk capacitance, as that makes inrush at
its input terminals, worse. In situations where the power supply
is otherwise healthy, and yet there is still a problem when a
USB device is plugged in, it could be a transient problem caused
by inrush at the device, or insufficient bulk capacitance at the
connector.

There are limits as to how much bulk capacitance can be used
in any design, because the ATX power supply likely has a
"stability limit" with respect to total capacitance on the
output. I mention that, because there are people out there who'll
reach for the biggest capacitor they've got, and slap it into
their PC. The following table is from a formfactors.org standard
for ATX.

"3.2.8. Capacitive Load

The power supply should be able to power up and operate normally
with the following capacitances simultaneously present on the DC
outputs. This capacitive loading should be used to check stability
and should not be included for noise testing.

Table 12. Output Capacitive Loads

Output Capacitive load (µF)
+12 V1DC 5,000
+12 V2DC 3,000
+5 VDC 6,000
+3.3 VDC 6,000
-12 VDC 350
+5 VSB 350

I suspect some motherboards probably use more than the limit, for
things like the +5VSB. But the above numbers would be used for
guidance, so someone doesn't slap 100,000 uF on one of the outputs.

So, the problem could be

1) Bad power supply - +5VSB is weak and dropping out, causing
PS_ON# to indicate shutdown.
2) Excessive total +5VSB loading from a DC load perspective.
3) A transient problem caused by a particular USB device. A
different device might not show the same characteristics.
A bad motherboard design, on the other hand, might do it
consistently for just about anything plugged in.
4) A miswired USB port, something getting shorted to ground.
Typically, miswired ports are on the front of the computer,
while the rear ports will always be correct (as they're routed
via copper tracks in the motherboard PCB).

HTH,
Paul

 

[Ian D]

Modern computers sometimes choose to power the motherboard USB
connectors from +5VSB. If you read the label on the side of the
supply, the 5VSB output is rated for about 2 amps. That is only
enough current, to power four USB devices loading to the max
of 500mA each. (And there are other loads in the design, on the
+5VSB rail, so the 2 amps cannot be used entirely for USB.)

The PCI USB card, could power its ports from the regular +5V port.
If so, its ports may behave differently.

You may get some idea, which rail is used for the ports, if a
USB device with a LED on it, continues to operate the LED when
the computer is in sleep state. (A USB LED reading lamp is another
device you could use to check it.)

If +5VSB is momentarily overloaded, that can be enough to cause
the PS_ON# control signal to be deasserted, which in turn causes
the power supply to go off.

A second reason for +5VSB to glitch, is if the proper precautions
are not observed in the design. Intel recommends that a bulk capacitor
be placed near the USB port. The purpose of that capacitor, is to
provide current, to compensate for the "inrush" current that flows
into a USB device freshly plugged in. The USB device itself, should
not use gobs of its own bulk capacitance, as that makes inrush at
its input terminals, worse. In situations where the power supply
is otherwise healthy, and yet there is still a problem when a
USB device is plugged in, it could be a transient problem caused
by inrush at the device, or insufficient bulk capacitance at the
connector.

There are limits as to how much bulk capacitance can be used
in any design, because the ATX power supply likely has a
"stability limit" with respect to total capacitance on the
output. I mention that, because there are people out there who'll
reach for the biggest capacitor they've got, and slap it into
their PC. The following table is from a formfactors.org standard
for ATX.

"3.2.8. Capacitive Load

The power supply should be able to power up and operate normally
with the following capacitances simultaneously present on the DC
outputs. This capacitive loading should be used to check stability
and should not be included for noise testing.

Table 12. Output Capacitive Loads

Output Capacitive load (µF)
+12 V1DC 5,000
+12 V2DC 3,000
+5 VDC 6,000
+3.3 VDC 6,000
-12 VDC 350
+5 VSB 350

I suspect some motherboards probably use more than the limit, for
things like the +5VSB. But the above numbers would be used for
guidance, so someone doesn't slap 100,000 uF on one of the outputs.

So, the problem could be

1) Bad power supply - +5VSB is weak and dropping out, causing
PS_ON# to indicate shutdown.
2) Excessive total +5VSB loading from a DC load perspective.
3) A transient problem caused by a particular USB device. A
different device might not show the same characteristics.
A bad motherboard design, on the other hand, might do it
consistently for just about anything plugged in.
4) A miswired USB port, something getting shorted to ground.
Typically, miswired ports are on the front of the computer,
while the rear ports will always be correct (as they're routed
via copper tracks in the motherboard PCB).

HTH,
Paul

Nice explanation of the +5VSB, but it doesn't address
the OP's symptoms. Here's my interpretation of what
he said.

- Plug any USB device, including self powered devices,
(printer), into any USB port and the PC shuts down immediately.

- Plug a USB device into a USB extension cable connected
to any USB port on the PC and it works normally.

- Plug a USB device into a PCI USB card and the PC
shuts down immediately.

I was thinking of a software/driver issue, but the fact that it
works with a USB extension cable negates that. One other thing
could be static, and the length of the extension cable dissipates
the discharge. The problem with that is that is unlikey a static
discharge would occur every time a USB device is inserted
into the PC. The fact that the instant shutdown also occurs
when a device is plugged into the PCI USB card eliminates
the cause being in the motherboard USB controller.

 

[Leythos]

Nice explanation of the +5VSB, but it doesn't address
the OP's symptoms. Here's my interpretation of what
he said.

- Plug any USB device, including self powered devices,
(printer), into any USB port and the PC shuts down immediately.

- Plug a USB device into a USB extension cable connected
to any USB port on the PC and it works normally.

- Plug a USB device into a PCI USB card and the PC
shuts down immediately.

I was thinking of a software/driver issue, but the fact that it
works with a USB extension cable negates that.

The length of the extension cable could also indicate that it's added
resistance is decreasing the load.

 

[Ian D]

The length of the extension cable could also indicate that it's added
resistance is decreasing the load.

--
- Igitur qui desiderat pacem, praeparet bellum.
- Calling an illegal alien an "undocumented worker" is like calling a
drug dealer an "unlicensed pharmacist"
(e-mail address removed) (remove 999 for proper email address)

I had thought of cable resistance, and capacitive or inductive
reactance, but that doesn't explain why a shutdown occurs
with the PCI USB card, which is is a separate system from
the motherboard USB controller. Also, a printer only puts a
10 or 15ma load on a USB hub, and it would probably have 6
foot cable.

It would be interesting to see what happens when the USB
extension is plugged in with a USB device already attached.
If it happens then, I'm again thinking of a static discharge
as the cause.

 

[Paul]

I had thought of cable resistance, and capacitive or inductive
reactance, but that doesn't explain why a shutdown occurs
with the PCI USB card, which is is a separate system from
the motherboard USB controller. Also, a printer only puts a
10 or 15ma load on a USB hub, and it would probably have 6
foot cable.

It would be interesting to see what happens when the USB
extension is plugged in with a USB device already attached.
If it happens then, I'm again thinking of a static discharge
as the cause.

I wonder if the machine is OK, if the printer is plugged in,
before the computer is booted. Does the computer run forever
in that way ? That would suggest more of a transient problem,
if the computer isn't bothered by a steadily present printer.

Paul

 

[Ian D]

I wonder if the machine is OK, if the printer is plugged in,
before the computer is booted. Does the computer run forever
in that way ? That would suggest more of a transient problem,
if the computer isn't bothered by a steadily present printer.

Paul

I'm wondering if it could be grounding problem. The only
electrical connections that would be common to both the
motherboard USB, ans the PCI USB card are the +5v,
and ground. Through the years in the hardware groups,
I've seen posters with weird problems that were eventually
traced to poor or missing grounding on the electrical supply.

 

[w_tom]

I had thought of cable resistance, and capacitive or inductive
reactance, but that doesn't explain why a shutdown occurs
with the PCI USB card, which is is a separate system from
the motherboard USB controller.

Actually it does. USB extension cords are a violation of the USB
standard. With a longer USB connection, the USB device may not
request high speed mode; therefore will not draw excessive power on
the +5VSB.

Paul explanations are a possible explanation. The recommendation of
replacing the power supply with a 600 watt supply is expected from
those who don't have even basic electrical knowledge.

Of course, a 3.5 digit multimeter measures the purple wire (+5VSB)
voltage, then that and many other possible reasons for failure could
be identified or eliminated immediately. Immediately as in not
wasting time and money shotgunning the power supply.

Excessive load on +5VSB may also result in the same failure if using
a larger supply. Or the computer may still boot even though the +5VSB
load is still excessive for that new supply. The only answer for
Paul’s reasonable explanation – and answer that is explicit and clear
- means measuring +5VSB both when the computer boots and when the USB
device is connected to cause a failure. Those numbers that may not
even tell the OP what has happened can be very informative to those
who learned this stuff.

Only way to solve the problem the first time is to first learn what
the problem really is - without shotgunning. Appreciate what happens
when a USB extension cable is used, why it could explain Paul's
scenario, and why that cable is a violation of the USB standards.

 

[Leythos]

Actually it does. USB extension cords are a violation of the USB
standard. With a longer USB connection, the USB device may not
request high speed mode; therefore will not draw excessive power on
the +5VSB.

Paul explanations are a possible explanation. The recommendation of
replacing the power supply with a 600 watt supply is expected from
those who don't have even basic electrical knowledge.

And by the time you follow all of W_TOM's purchase a DVM advice you
could have installed a quality PSU and been up and running. The DVM
won't tell you anything unless you hack apart a cable, breaking it in
the process, since you're not getting power for the USB ports from any
place easy to measure.

 

[w_tom]

And by the time you follow all of W_TOM's purchase a DVM advice you
could have installed a quality PSU and been up and running.

Multimeter measurements can be taken five times over. And a power
supply still could not be swapped. Learn basic power supply
concepts. A defective power supply can still boot a computer.
Swapping a supply reports nothing useful (definitive). OP's supply
could have been defective months or even a year ago. And that meter
could have identified a defect even that long ago - in thirty seconds.

Leythos claims the meter says nothing. Leythos does not know how to
use a meter - and does not have the basic electrical knowledge to
understand what those numbers report. Ignore the usual insults from a
naive naysayer.

Thirty seconds with the meter also provide numbers so that the few
who actually know how electricity works will reply - can provide
further assistance. Simply power a meter on 20 VDC range, touch leads
to where wires connect to the motherboard, then read a number. Post
those numbers here. Get a useful reply that can actually define the
problem - without doubt.

Doubt: a defective or insufficient power supply can still boot a
computer.

bg - if connectors caused a short, well, USB functions then report
that short as a "power surge" and disable that USB port. Nothing in a
normal USB port (if provided sufficient voltage) can crash the OS.

Why would that USB device cause a power problem? Paul offers a
plausible reason. If in high speed mode, a USB device draws more
power. Numerous other reasons exist. With insufficient information -
such as voltage numbers - no responsible answer is possible. A
definitive answer happens after important numbers are provided -
especially DC voltage from the purple wire (+5VSB) and from other
important wires such as any one red, orange, and yellow wires.

 

[Leythos]

Leythos claims the meter says nothing. Leythos does not know how to
use a meter - and does not have the basic electrical knowledge to
understand what those numbers report. Ignore the usual insults from a
naive naysayer.

Now you see why W_Tom is not respected in these groups - I've not once
said that a multimeter says nothing - in fact, quite the contrary. What
I have said is that in order to test the voltage at the END of the
second cable, you would have to cut into the cable, and that's now good.

If the problem is with the power provided to the card via the
motherboard connection at the PCI slot, a multi-meter is not going to
provide any help at all.

So, the quickest way to determine the reason for THIS problem is to get
a higher wattage PSU, a new one, and see if that resolves your issue.

 

[Ian D]

Now you see why W_Tom is not respected in these groups - I've not once
said that a multimeter says nothing - in fact, quite the contrary. What
I have said is that in order to test the voltage at the END of the
second cable, you would have to cut into the cable, and that's now good.

If the problem is with the power provided to the card via the
motherboard connection at the PCI slot, a multi-meter is not going to
provide any help at all.

So, the quickest way to determine the reason for THIS problem is to get
a higher wattage PSU, a new one, and see if that resolves your issue.

--
- Igitur qui desiderat pacem, praeparet bellum.
- Calling an illegal alien an "undocumented worker" is like calling a
drug dealer an "unlicensed pharmacist"
(e-mail address removed) (remove 999 for proper email address)

It seems like we're flogging a dead horse here. It's been 3 days
since the original post, and the OP has not responded with any
further information. If the PC shuts down, but could be restarted
with the front panel button without cycling the rear PSU switch,
then the +5v STBY has not tripped off, as it is needed to start the
system. That's the kind of info we're lacking about this problem.

 

[Leythos]

It seems like we're flogging a dead horse here. It's been 3 days
since the original post, and the OP has not responded with any
further information. If the PC shuts down, but could be restarted
with the front panel button without cycling the rear PSU switch,
then the +5v STBY has not tripped off, as it is needed to start the
system. That's the kind of info we're lacking about this problem.

I've got time to flog a dead horse.

 

[SC Tom]

I've got time to flog a dead horse.

--
- Igitur qui desiderat pacem, praeparet bellum.
- Calling an illegal alien an "undocumented worker" is like calling a
drug dealer an "unlicensed pharmacist"
(e-mail address removed) (remove 999 for proper email address)

Besides, they're the best ones to flog; they don't kick.

 

[Paul]

Plugging in a usb device would cause the computer to reset. Even with
extreme care to insert it straight in, I got a reset. I had a spare set of
connectors with the cable for an ASUS so I just swapped it, and all was
fine. I didn't bother to troubleshoot it any further.
The op said his usb worked ok when he used an extension cable. This sounds
very similiar to what I experienced.

To analyze this problem, you'd want

1) A storage oscilloscope, connected to +5VSB (assuming +5VSB
powers the USB port).

2) Start trace, just as the USB device is plugged in. Set the trigger
condition, on a dip in the voltage. Even under normal circumstances,
the USB bus voltage dips when a USB device is plugged in.

3) Capture a second trace, where the USB device is connected
by means of an extension cable.

If it works with the extension cable, and fails when directly
connected, it could be that the length of cable presents
enough series resistance, to prevent a full amplitude collapse
of +5VSB. The cable is functioning as an "inrush limiter".
Inrush currents on USB, have been known to go as high as
5 amps (at one time, I had an Intel document, with an
oscilloscope picture of this).

If +5VSB is ever "flattened", that can cause the power supply to
go off. The question is, what is the mechanism - badly designed
motherboard, badly designed USB device (spec violation on insertion),
or whatever. A simple multimeter is not going to do a good job of
highlighting a transient problem. The transient could be quite
short. A Polyfuse takes time to open, so it will also
pass the transient.

If the motherboard designs had the old option, of jumper
selectable powering of the USB port, you could switch a stack
of two ports over to +5V. That is the regular supply of +5V, which
has a rating of 20 amps or more. And if that one is flattened for
some reason, the power supply won't necessarily go off. The
computer could crash, bur the fans might continue to spin.

They switched over to this "+5VSB only" philosophy a couple years
ago, and it removes a degree of control for the end user. One
of the reasons to leave some degree of control to the end user,
is to work around stupid design issues like this. That is one
of the reasons I won't buy a motherboard that doesn't allow me
to adjust Vcore or Vdimm or timing or whatever. And one reason
I cannot buy a Dell/HP/Acer/Gateway etc, because it doesn't
have enough control for me.

The thing is, if I had a problem on my current motherboard (which
has no header to choose +5V versus +5VSB), I know how to fix it.
I unsolder the Polyfuse feeding the USB dual stack. I run a lead
over to +5V, placing the Polyfuse in series with the lead. That
protects me against prolonged overcurrent conditions (i.e. a real
short and not a transient). But running from +5V, also removes the
ability of that USB device to wake the computer, which might be a
desirable feature for a USB keyboard for example.

If the symptoms weren't as stated, I would also suggest a
powered external hub as a solution. The hub would have
to "eat" the transient, which would protect the computer.
(The hub would have its own +5V @ 2A DC supply.) But since
a simple length of extension cable is fixing this,
using a powered hub is not curing anything. If the computer
was failing, even with an extension cable, then using an
external powered hub would be another way to address the
problem.

Paul

 

[w_tom]

If +5VSB is ever "flattened", that can cause the power supply to
go off. The question is, what is the mechanism - badly designed
motherboard, badly designed USB device (spec violation on insertion),
or whatever. A simplemultimeteris not going to do a good job of
highlighting a transient problem. The transient could be quite
short.

A transient that short would never cause a problematic voltage
reduction. And increases wire would not avert that transient.

However the extension cord may cause a USB device to not enter High
Speed mode. +5VSB must be so low as to be defective even without the
USB device. Would still boot the computer. Would appear defective
only on the multimeter. So close to the edge that a USB device in
High Speed mode finally causes the crash.

There exists a wide area between good voltage and a crashed
computer. In that wide region is a working computer and a voltage too
low. A condtion found using a multimeter.

Again, a USB transient cannot be that fast and still crash a
computer. Extension cord wire cannot diminish that current. At least
half the posts here are immediately eliminated if simply measuring the
+5VSB with a multimeter. Without those voltage numbers, we are doing
nothing but wild speculation.

 

[Leythos]

At least
half the posts here are immediately eliminated if simply measuring the
+5VSB with a multimeter. Without those voltage numbers, we are doing
nothing but wild speculation.

And you can't measure that value without cutting into the cable/device,
since you have to measure it at the device to be sure.

A power supply is cheap, does not require any cutting, and is most
likely the cause of the problem.

 

[john]

A transient that short would never cause a problematic voltage
reduction. And increases wire would not avert that transient.

However the extension cord may cause a USB device to not enter High
Speed mode. +5VSB must be so low as to be defective even without the
USB device. Would still boot the computer. Would appear defective
only on the multimeter. So close to the edge that a USB device in
High Speed mode finally causes the crash.

There exists a wide area between good voltage and a crashed
computer. In that wide region is a working computer and a voltage too
low. A condtion found using a multimeter.

Again, a USB transient cannot be that fast and still crash a
computer. Extension cord wire cannot diminish that current. At least
half the posts here are immediately eliminated if simply measuring the
+5VSB with a multimeter. Without those voltage numbers, we are doing
nothing but wild speculation.

I believe that that you are incorrect. I have experienced the 'puter
shutting down because of a static charge when I plugged a a USB drive
in. The solution to my problem was as simple as raising the humidity to
about 25%. Have not had any problems since.

John