4TB external HD

On Wed, 27 Jun 2018 21:48:33 +0100, Brian Gregory
wrote:

On 26/06/2018 23:25, Paul wrote:
Brian Gregory wrote:
On 26/06/2018 03:20, Paul wrote:
Brian Gregory wrote:
On 25/06/2018 23:44, Paul wrote:
Brian Gregory wrote:
On 25/06/2018 22:01, Ed Cryer wrote:
Ed Cryer wrote:
Paul wrote:
Stephen Wolstenholme wrote:
On Mon, 25 Jun 2018 12:53:03 +0100, Ed Cryer

wrote:

I have a Seagate 4TB external HD, powered, like this;
https://goo.gl/q1z7RC

It works fine with Win7. No sign of trouble.
But when I plug it into Win10 the system dies. Strangely,
when I boot with it plugged in, everything goes fine.

There's nothing relative in the Windows Log, other than info
that the system didn't shut down properly before the reboot.

Has anybody seen this? Or, any hunches as to what's causing it?

Ed

I had a similar problem with a Verbatim drive. It was fixed by
a new
driver that was applied by an automatic update. I assume
Seagate have
some driver fixes.
Steve


It should be using a regular USB Mass Storage driver.

The drive is 3.5" and has an external 12V @ 1.5A adapter.
The question would be, where does the +5V come from.

A lot of drive enclosures, convert some of the 12V to 5V to
run the logic board on the hard drive. So rather than the
wall adapter having four wires like a Molex, only +12V comes
from the adapter, and the +5V is produced on site.

(Wall +12V ------------ Adapter ---- +12V ------ Hard drive motor
* Wart) GND ------------ Board** ---- GND* ------
******************************** ---- GND* ------
******************************** ---- +5V* ------ Hard drive
logic board
(not all pins shown)
***** +5VSB ------------ USB to****** TX+ ------ Hard drive
****** D+*************** SATA******** TX- ------ data
****** D-*************** Converter*** RX+ ------ connector
****** GND ------------- Chip******** RX- ------ 7 pin (3 GND)

The USB cable by comparison, has a source of +5VSB at
up to 900mA on USB3. If the ATX power supply doesn't
have sufficient +5VSB rating, and you actually
overload +5VSB, it can cause the motherboard to shut
off. Instead of doing a restart, it would appear to
spontaneously shut off.

Maybe they made it like this, but... unlikely.

***** +12V ------------ Adapter ---- +12V ------ Hard drive motor
****** GND ------------ Board** ---- GND* ------
******************************* ---- GND* ------
******************** +-------------- +5V* ------ Hard drive
logic board
******************** |
***** +5VSB ---------+-- USB******** TX+ ------ Hard drive
****** D+*************** SATA******* TX- ------ data
****** D-*************** Converter** RX+ ------ connector
****** GND ------------- Chip******* RX- ------ 7 pin (3 GND)


But even though that looks bad, it's not possible to
tell from the outside, exactly how the external storage
box was designed.

If it was a driver issue, you'd expect to see some
error info in the Reliability Monitor related to the
driver.

The drive might not be spinning when it is plugged
in during a session, but spinup current comes from
the 12V 1.5A adapter, rather than the USB bus cable.
The USB bus cable should only power the adapter
board inside the enclosure. Or in a more exotic
design, the bus power could also be used to run
the hard drive logic board.

For a driver issue, you'd expect to see a BSOD on the
screen for a short time. If you disabled automatic
restart, then the BSOD would stay on the screen longer.
If it really is a power issue (crushing of +5VSB) then
changing the automatic restart setting would not change
the symptoms.

If the computer shuts off and doesn't restart, that
sounds more like a +5VSB issue.

And while placing a USB3 powered hub between the
computer box and the external drive would sound
like fun, those aren't always designed properly
either. Sometimes there are undesired interactions
between the downstream +5V on the powered hub,
versus the +5VSB on the host side of the hub box
(the "backfeed" problem).

If an iPad was being charged off the PC, at the
same time the HDD enclosure had the USB data
cable plugged in, that would represent a pretty
heavy load on +5VSB (several amps).

*** Paul

Hi Paul.

I've just tried it on another Win10 box and it works perfectly
there.
I took it back to the faulting box, booted with it in,
uninstalled it, rebooted ...... problem still there.

I've not had any other USB problem with the box. I've plugged in
other HDs, memory sticks, external bluray writer. All fine.

Ed


The main difference that leaps out at me is that the 4TB drive
has a power adaptor; none of the other devices do.

Ed


My theory:
The power supply on the drive is cheap Chinese rubbish and has no
earthing and measurable mains voltage leakage and the PC shuts
down when a small spark jumps from the USB cable to the PC as you
insert the USB connector.

If the drive's power supply is earthed through to it's output then
maybe the PC isn't earthed or an earth loops is causing current
flow that crashes the PC.


I would be alarmed if I saw sparks fly between the PC USB
connector and a powered USB 3.5" drive.

The ground shell touches first, and we'd be talking about
a spark-into-ground event, with the possibility of
crosstalk between the shield and internal conductors
to upset other electronics. This tends to be worse on
*front* panel connectors. Due to the front connectors
never being earthed properly (typically set in plastic,
and only the shield on the cable provides ground). The
back connectors have obvious physical paths to chassis,
and a slightly lesser chance of induced upset.

The wall adapter has 230V on one side. The HiPot test
is supposed to verify 1100V withstanding for some
period of seconds. But the sticker or proof of test
might not be visible from the outside. ATX supplies
sometimes have a visual indication on them that
they've been HiPot tested. It's best to apply the
test after final assembly, as if anyone makes a dopey
mistake, the test might catch it.

The transformer(s) themselves could be HiPot tested,
and that's probably done on the transformer manufacturing
line. But that's not likely to be the last test.
You want a final test, to verify no creepage & clearance
problems, or solder splashes where they don't belong.
Visual inspection (even robotic) isn't enough for
safety purposes. If someone files a law suit
against you, you can show the HiPot stamp on the
unit, show the court your test station for the
procedure and so on. For whatever that's worth.

If you do see a spark, well, stop using it :-)
If the design is intended to float, it should
not throw sparks. And there's no reason for
that test case to violate the HiPot max.
If you raised the PC chassis thousands of volts
above the rest of the house, yes, you'll see
a spark on the resulting multiple HiPot failures.

*** Paul

I said small spark.

I've seen this happen on an HP printer. A Deskjet 720 if I remember
correctly. Luckily I could leave it connect all the time so it
wouldn't crash the PC.

I still don't see you properly describing exactly what happens, a
crash or a power off or what.


If you short +5VSB to ground, you get a power off.

Never heard of crowbar protection?

I have.

I designed it into a power supply at home, at
the age of 18.

I don't think that's what you mean though.

You probably want OCP (Over Current Protection), which
can be implemented thermally (regulator overheats at
a certain current flow level and removes output).

There is also foldback current protection, where
the device reduces current flow, protecting both
the load and the supply.

There is also "put-put" mode, which early PCs
used, leading to more successful boots, even
if a PC "wasn't feeling well". That mode is suited
to the older computers running 30-35W loads or
so for the processor.

There are several approaches to overload.

*******

Crowbar is for OVP. OverVoltage Protection.
What it seeks to do, is clamp the output to
zero volts, causing the OCP to cut in.

Crowbar was intended for cases where the failure
modes of the power supply, made it really easy
for the supply to be in a permanent "high voltage
output" state. For example, a series pass linear
which is jammed in an "ON" state, would leave the
DC output permanently too high. In some cases,
a supply with a fast-blow fuse located elsewhere
in the circuit, application of the crowbar would
blow the fuse.

Modern SMPS attempt to remove switching drive
from the primary side, on a fault. So there's no
need to crowbar them. As soon as the switching
action stops on the primary (even if one transistor
is permanently jammed on), it takes alternation
to pass an AC signal through the isolation
transformer to the secondary.

DC to DC converters (like the secondary power
board in 80+ supplies), have different fault
modes than the main 12V section, and may require
a different approach. (They could be buck regulators
for example, as they don't need isolation as a feature.)
While some of the chips used may remove primary drive
to the buck transistor (causing MOSFET to go to OFF state),
that's not sufficient to handle all fault scenarios.

It would be an unusual design today that uses
a crowbar. There are plenty of options in modern
supplies. For example, if the secondary board
in an 80+ fails, you can kill the switching
action on the primary side of the main transformer,
or you can disable the PFC front end (the part
that power-factor-corrects the AC load caused
by the ATX supply). The advantage of killing
the switching action, is protection is closer
to instantaneous. Killing at the PFC requires
16mS to a second or so before the output stops.

** Paul

So YES a pulse getting on a power rail can cause a power supply to cut out.

I have a Clamp Leaker 140.