usb device has malfunctioned and exceeded the power limits of its hub

[Bill in Co]

First, a special callout to Paul for providing me some more info on the USB
hubs.

This is what started all of this investigation on the USB ports:

What happened was I had something plugged into a USB port and this error
message came up. However, I know the USB ports are power protected by
design, so this shouldn't really be a hardware problem.

So what I did was uninstall ALL the USB Host Controllers and reboot the
system to let it reinstall all of them again, but the problem still
persists. (At one point in this process after rebooting, it came up with a
warning advising to reset the port, which I did, but the problem still
exists - the port is "dead".

I've researched this online and haven't found any other suggestions, but
still find it hard to believe the hardware (USB port) is truly damaged. So
if uninstalling all the USB Host Controllers so they can be rese can't wake
up the USB port, I don't know what will. Somehow it is not getting reset -
not getting the message.

Is there a chance that reinstalling SP3 could force a port reset during the
install process? Or is there anything else I can try?

[gfre...@aol.com]

I have a flaky 7 port powered hub that dopes the same thing. Usually
just unplugging the hub and plugging it back in (USB and Power) gets
it going again for a while.

[Bill in Co]

Tried that to no avail. BTW, I'm talking about a couple of built in USB 2
ports on the motherboard of the Dell Laptop here, not the external one which
I had removed for all these tests. I probably should have stated that to
avoid any confusion - sorry. This Dell has two USB 2 ports and one combo
USB-eSATA port. (Dell Latitude d830).

[Paul]

You can delete the ENUM key in the CurrentControlSet
and let the OS rediscover all the hardware.

I even tried this on a Win10 install and it worked.

Take a backup first. I did.

*******

If you're out of ideas for registry editors, the
Kaspersky rescue CDs about a year ago, had a registry
editor in addition to the normal offline scan functions.
(The registry editor was released open source
as well, so may be available elsewhere.)
Current ISO downloads may not be as fully functional
as they once were. I had some trouble with the last
attempt, but don't recollect all the details.

If you keep the older KAV ISOs and didn't toss them away,
you may have material you can use.

Paul

[Bill in Co]

Great idea, thanks! But ... Using regedit, I tried deleting the ENUM key,
and it won't let me "cannot delete Enum: Error while deleting key."
Probably due to the fact it's in use, and being protected. Is there any
way to get around this? BTW, there are several ENUM keys, but the ones I
was most interested in deleting were in HKEY_LOCAL-MACHINE and
HKEY_CURRENT_CONFIG. Perhaps in safe mode? But I'm not sure that would
work either.

I restored the registry just in case (using ERUNT), and I'm back to square
one. Maybe I should try safe mode to do it?

[Paul]

My notes that I keep, are always cryptic.

https://s9.postimg.cc/vwioz43f3/WIN10_delete_ENUM.gif

psexec -hsi cmd <=== 32 bit OS, opens SYSTEM cmd.exe window
psexec64 -hsi cmd <=== 64 bit OS, opens SYSTEM cmd.exe window

reg delete HKLM\SYSTEM\CurrentControlSet\Enum /f

This is the OS discovering hardware later.

https://s9.postimg.cc/q8ce88ei7/WIN10_discovers_HW.gif

That suggests the key can be removed with the SYSTEM account.

Paul

[Bill in Co]

I managed to find PowerShell 2.0, which (mostly) successfully installed.
Then I tried the "psexec" command, but it didn't recognize that, but when I
ran "powershell" at the cmd prompt, that worked. OK, good so far.

Then I tried the reg delete ... command line per above, and got "access
denied". So then I went back to regedit and changed the permissions of the
ENUM keys to allow me access, and that seemed to help, but, in the process
of deleting some of the ENUM stuff, rendered the system unbootable, and had
to use Acronis True Image to restore a previous image. UGH.

I guess at this point I have a choice of either being more selective in what
I remove in the ENUM stuff, or living with this, unless you have some other
ideas. :-) But in all cases as I recall, I was never able to completely
delete all the ENUM keys. I was also trying to delete the backup ENUM keys
with the 001, 002, 003 stuff just to cover everything (I didn't want it to
fall back to those backup keys - that way I would FORCE a new ENUM install)

[Paul]

Sorry, perhaps I neglected to provide the
reference to the pstools download. That's where
psexec comes from. You run say, psexec64 from
an existing administrator command prompt,
to make a "system" command prompt. To verify
it worked, try

whoami

The package you want is here.

https://docs.microsoft.com/en-us/sysinternals/downloads/pstools

You unzip the download, and select the matching executable.
If you have a 64-bit OS, you use psexec64.

psexec64.exe -hsi cmd

The window that pops up, runs as SYSTEM.

The reg command issued in that new window,
should also run as SYSTEM.

There's no guarantee that SYSTEM will work. There
is some registry stuff owned by TrustedInstaller.
But a lot of the pesky stuff I've dealt with
in the past, SYSTEM was enough.

Sorry for costing you a disk restore... Like
I said, my notes file is a bit cryptic, and
sometimes I even have to dig around to find
the rest of the details.

*******

As for the Sysinternals stuff, you should be
aware that some programs no longer support WinXP.
If you run into such a situation, you need to
go back in time, to find an older copy of the
utilities. Since the encoding of web site URLS
changed at one point, this is particularly irksome.

You can't just go to archive.org with modern
URL in hand, and go back to the year 2006. Won't
work. Instead, you have to hope that Google has
an old URL you can find in a search. The old URL
is irregular enough in design, to make this
a painful process. If they'd kept the damn
URLs consistent, this wouldn't be nearly
as much of a challenge.

Paul

[Bill in Co]

OK, tried that out and at least got to the "more data is available" error
message state (and not the "access is denied" error (but after also going
into the registry and relaxing the key's access privilege).

Then I discovered a free program called RegistrarHome that allowed me to
delete the keys, with no problems. However, even after doing that and
rebooting, and waiting for all the devices to get reinstalled/reenumerated,
the two stacked computer USB ports on the side of the laptop are still dead.
(there are two stacked ports on the side of the laptop, and one combo
USB/eSATA port on the back of the laptop that still works fine, and that's
the one I apparently will have to use, forever. But I've connected that
port to the external 4-port hub I mentioned before), so that's a workaround
that I guess I will have to live with.

It just surprises me that none of this fixed it. So my conclusion is either
the USB side port is indeed dead at the hardware level, or else is incapable
of actually being reset (even though that error message popped up during
reenumeration, and I diligently clicked on RESET!

But, "it's an ill wind that blows no good". At least I've learned a few
things in the process.

Another idea would be to try and replace the built-in port, but I think
that's a real long shot and would probably be a real PIA, assuming it's even
doable.

But thanks for all your help Paul

Oh, you might want to take a look at that program (free) called
"RegistrarHome" if you're interested (I had to use version 7, as version 8
won't successfully install on my XP system here). It's more powerful than
regedit. (It's called an Advanced Registry Tool Suite, and seems to be just
that).

http://www.resplendence.com/registrar_home

[Paul]

There are a few USB ports with a reputation.

Some of the NEC USB2 ports would blow out, but
they blow out individually.

There were some Intel ICH5 that went into latchup
on the USB ports. A good number of chips
were destroyed. Others burned the bond wires
powering the USB pads, so that no USB worked
any more.

But most other brands of USB2 are OK.

Laptops use 8 pin DIP switches to cut off
power flow. One eight pin chip may be used
to protect two ports. On desktops they use
Polyfuses, and while the fuse is auto-resetting,
sometimes they degrade to the point they have
to be replaced.

The chipset drives D+ and D-.

The power limiting circuit (whatever type)
drives VBUS.

If a powered hub without relay is connected,
that doesn't usually destroy things. I don't know
right off hand, of a failure mode of the
8 pin silicon switch. The switch chip is just
a couple of MOSFETs in a package, with current
sensing used to operate the gate on the MOSFET.

*******

The Kaspersky CD offers an offline registry editor.
I can't see a reason that one would care about
permissions.

My cryptic notes suggest I did the ENUM
delete on CurrentControlSet while the OS
was running. If the OS needs a LastKnownGood,
that's when it reaches for the numbered ones.
I didn't bother to edit those.

Paul

[gfre...@aol.com]

You could just have a hardware problem.

[J. P. Gilliver (John)]

In message <pjoji1$2jd$1...@dont-email.me>, Paul <nos...@needed.invalid>
writes:
>Bill in Co wrote:
[]

>> It just surprises me that none of this fixed it. So my conclusion
>>is either the USB side port is indeed dead at the hardware level, or
>>else is incapable of actually being reset (even though that error
>>message popped up during reenumeration, and I diligently clicked on RESET!
>> But, "it's an ill wind that blows no good". At least I've learned
>>a few things in the process.
>> Another idea would be to try and replace the built-in port, but I
>>think that's a real long shot and would probably be a real PIA,
>>assuming it's even doable.

It may not be too bad. Google for YouTube videos, handbooks, service
manuals and so on for your model of laptop: if you're lucky, the USB
socket (and possibly some of whatever connectors are next to it) won't
be on the main motherboard, but on a separate PCB, with just connectors
- and maybe one or two components - on it. These subsidiary boards
usually connect to the main motherboard by ribbon cables with a plug on
their end. Searching ebay for modelnumber USB _may_ show you them.

(Of course, even if that is the case, getting the back off most laptops
can be a PITA - it's mainly finding where all the screws are, including
under the battery, under the hard drive, under the keyboard, and so on.
(Plus often one of them refuses to budge.) But, if it _is_ a subsidiary
board, then once you've got the back off, replacing the subsidiary board
shouldn't be too difficult.

>> But thanks for all your help Paul
>> Oh, you might want to take a look at that program (free) called
>>"RegistrarHome" if you're interested (I had to use version 7, as
>>version 8 won't successfully install on my XP system here). It's
>>more powerful than regedit. (It's called an Advanced Registry Tool
>>Suite, and seems to be just that).
>> http://www.resplendence.com/registrar_home

Downloaded.
[]

>There were some Intel ICH5 that went into latchup
>on the USB ports. A good number of chips
>were destroyed. Others burned the bond wires
>powering the USB pads, so that no USB worked
>any more.

Presumably externally-powered might still work.
[]
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

If you carry on hating, you're the one who's damaged.
- Sir Harold Atcherley, sent to the Burma/Siam railway in April 1943

[Paul]

J. P. Gilliver (John) wrote:
> In message <pjoji1$2jd$1...@dont-email.me>, Paul <nos...@needed.invalid>

>> There were some Intel ICH5 that went into latchup
>> on the USB ports. A good number of chips
>> were destroyed. Others burned the bond wires
>> powering the USB pads, so that no USB worked
>> any more.
>
> Presumably externally-powered might still work.
> []

Changing VBUS to External_Plus_Five doesn't help.
It's the signal level on D+ and D- which have
both dropped to zero volts DC, that is a problem.
The bond wire melts, because the two Ts have
gone into latchup, and they short together
+5VSB to GND, until the bond wires melt (or
the chip starts to burn). We know the bond wire
melts, because of the burn mark over a USB power
pin (roughly) in the BGA pinout diagram. It will
burn right through the top of the chip, depending
on which fails first. If the bond wire holds,
the plastic above the bond wire burns on the ICH5.
If the bond wire holds, the silicon gets hot enough
to ruin it.

+5VSB
|
<=== bond wire melts
} VBUS
-- T |
| |
+--- D+/D- -------------------- External_Rcvr
| ^ |
-- T \ |
| \ GND
GND Data signal 0 volts

This was supposed to happen on ICH4, but I could
find no evidence to back that up.

Adding external power provides no means
to access the top side of the USB block
output pad totem pole top transistor.

The neat thing is, the logic block to the left
of this diagram still has power. The hardware
state looks good, in terms of status bits. But,
it's the part that drives the output pins on
the ICH5 that is non-operational. And nothing
short of delidding and bonding a new gold
bondwire on some two nodes, will bring it back.

If the silicon receives thermal damage (because
the bond wire didn't break first), the computer
will no longer run. The two end results are
either "all USB ports dead" is one outcome
and the computer still boots, or "burn hole
appears in lid of chip, she's dead Scotty".
Those are your choices when it happens.

I still have a board with an ICH5, but current
disposition is, the board is flaky :-/ Hmmm.
No burn mark. USB works.

Paul