Shortest period to switch off electronic equipent for before switching on

"Roderick Stewart" wrote in message
...
In 24 years I've had more punctures than power cuts, more broken
glassware, more burnt fingers, more job redundancies, more tooth
fillings, just about every adverse event you could name. Ive even won
the lottery more often than I've had power cuts (regrettably only a
few quid but wins nonetheless).

Gosh. You are very lucky that you've only had one power cut. I think
everywhere I've lived (various houses in suburbs of towns in
Buckinghamshire, Berkshire, Oxfordhire, and now our house in a village near
a market town in Yorkshire) we've had power cuts (either many minutes/hours
or else a long enough glitch to reboot everything) on average a couple of
times a year - usually after a thunderstorm.

Probably more power cuts than punctures, though maybe about the same number
as blown headlight bulbs on my present car, which went through Halfords
bulbs at an incredible rate, though the Osram ones that I now use last a lot
longer. I reckon it's the mechanical shock from all the potholes in the
roads round here. The garage can't find any electrical problem such as
alternator voltage going too high.

On Sun, 9 Apr 2017 12:21:16 +0100, "NY" wrote:

wrote in message
.. .
You can't just lump everything into the same bracket - a lot of people
just
turn the TV off "cold turkey", its a lot better than leaving it unattended
on standby.

There is absolutly nothing wrong with "standby".

Anything with a hard drive can trash a file if you power down in the
middle
of a write operation.

I suppose anything with disk-like storage (including flash drives) can be
trashed if the power goes off during a write, though the time during which
the files is in an inconsistent state between data sectors and sector map
may be less with solid state than a mechanical rotating disk.

Nonsense. There are many ways to design a device that will survive a
power failure during disk write. A bit of reserve power is probably
the best solution but a JFS works, too.

It depends what filesystem they use. NTFS is pretty robust but it requires a
licence to be paid to the inventors which is why a lot of PVRs use/read only
FAT/FAT32. If they are Linux-based they will be able to use filesystems
which are more robust (I'm not very clued up on Linux).

That's an economics decision. There is nothing cast in jello that
says they have to cheap out, or even use an existing FS. JFS
techniques are well known. Even if they do want to use a vulnerable
FS, a proper power system design can guarantee a safe shut down.

But I'd say that a design which trashes the file that is being written to or
the whole filesystem when the power goes off (eg due to a power-cut, which
is not totally unexpected) is a bad design: you need some form of resilience
in terms of battery-backed supply until the file write is complete and
consistent.

Exactly so. This is common with flash-based systems. The power
system has to guarantee at least the current block has been written.

I use my Windows 7 PC as a PVR (either Windows Media Centre or NextPVR) and
I've occasionally had power cuts during recording. I've never yet lost the
recording that was being made or the filesystem of the recording HDD (NTFS).
OK, Windows may do a chkdsk repair after the PC reboots, but it seems to
sort itself out.

NTFS?

A UPS would be useful for graceful shutting down during power cuts, but it
might be more trouble that it's worth. At present my BIOS is set to reboot
the PC automatically after the power is restored. If the UPS keeps the power
up and initiates a graceful shutdown, the PC may not boot back up once the
power comes back.

My wife bought a 3kVA UPS with her big Dell PC about 10 years ago. We didn't
get round to setting up the UPS for a couple of years. By the time we did
(and after the warranty had expired, inevitably) we found that it had
virtually no battery capacity: the battery monitoring software (monitored by
USB connection) showed the battery accepting charge and gradually charging
up and eventually showing as fully charged, but as soon as the mains was
removed, the battery discharged within about 10 seconds with a nominal 40 W
lightbulb as the load. Utterly useless, and it wasn't even worth buying a
replacement battery for it because the fault may have been in the charging
circuit rather than the battery itself. APC didn't want to know when we
asked what a repair might cost, so it went in the skip - a waste of money as
it was never even used.

Always test your emergency system. When I was in IBM, they tested the
emergency generators once a month. I have no idea why they had such
in the building I was in (just offices and a few labs at that time)
but they were there, so the were tested.

It's a really good idea to test your system backups from time to time,
too. You may find the restore doesn't work. :-(

Moral of the story: always try any new hardware during the warranty period!

Anything that's important has to be tested.

On 09/04/17 10:01, Brian Gaff wrote:
If that
fails, then hold in the power button, NOT the reset button as this normally
protects the hard drive from corruption.

As I understand it, a long press is handled by the hardware and forces a
dirty shutdown. If it were handled by the firmware, it could be
difficult to shut down a battery powered device which had locked up.
RESET should be more gentle than a long press.

Typically, power supplies will not have decayed before any sector write
in progress has completed and disks should park. However, you are
basically relying on fault recovery mechanisms to protect the physical
media, and the low level formatting.

Journalling file systems may protect the basic file structure, but data
buffered in user space, and write back caches (OS and device) will not
get written and very few application designers will have designed them
to be completely safe against partially written files.

Consumer devices have to be tolerant of such abuses, but whenever you
start relying on fault recovery mechanisms, for normal operation, you
are on dangerous ground.

On Sun, 09 Apr 2017 12:21:16 +0100, NY wrote:

====snip====


My wife bought a 3kVA UPS with her big Dell PC about 10 years ago. We
didn't get round to setting up the UPS for a couple of years. By the
time we did (and after the warranty had expired, inevitably) we found
that it had virtually no battery capacity: the battery monitoring
software (monitored by USB connection) showed the battery accepting
charge and gradually charging up and eventually showing as fully
charged, but as soon as the mains was removed, the battery discharged
within about 10 seconds with a nominal 40 W lightbulb as the load.
Utterly useless, and it wasn't even worth buying a replacement battery
for it because the fault may have been in the charging circuit rather
than the battery itself. APC didn't want to know when we asked what a
repair might cost, so it went in the skip - a waste of money as it was
never even used.

Moral of the story: always try any new hardware during the warranty
period!

Well, Duh!!!

JOOI, had you even unpacked this BFO UPS to plug into a mains socket so
as to at least maintain the SLAs? You may have gotten away with waiting a
year before allowing it to refresh charge its battery pack but leaving it
as long as two years (plus whatever time it may have spent on the shelf,
awaiting sale) would almost certainly result in a sulphated battery pack
and the very symptoms you witnessed.

Mind you, APC, in the interest of maximising the headline 'As New' OOTB
autonomy, set the per 12v SLA float charge voltage to the maximum of
13.8v rather than the more sane (less corrosive) 13.5v which trims the
autonomy figure by some 5 to 10% for a given cost of battery pack so even
if you had unpacked and plugged it into a mains supply straight away to
prevent sulphation, you may have suffered the same symptoms regardless.

Mind you, one might expect to see a good 3 to 5 years service life out
of a typical APC battery pack under such continuous float charging
conditions: two years seems a bit on the short side even for an APC
battery pack to become 'worn out' from such abuse.

--
Johnny B Good

wrote in message
...
On Sun, 9 Apr 2017 12:21:16 +0100, "NY" wrote:
My wife bought a 3kVA UPS with her big Dell PC about 10 years ago. We
didn't
get round to setting up the UPS for a couple of years. By the time we did
(and after the warranty had expired, inevitably) we found that it had
virtually no battery capacity: the battery monitoring software (monitored
by
USB connection) showed the battery accepting charge and gradually charging
up and eventually showing as fully charged, but as soon as the mains was
removed, the battery discharged within about 10 seconds with a nominal 40
W
lightbulb as the load. Utterly useless, and it wasn't even worth buying a
replacement battery for it because the fault may have been in the charging
circuit rather than the battery itself. APC didn't want to know when we
asked what a repair might cost, so it went in the skip - a waste of money
as
it was never even used.

Always test your emergency system. When I was in IBM, they tested the
emergency generators once a month. I have no idea why they had such
in the building I was in (just offices and a few labs at that time)
but they were there, so the were tested.

The office building where I worked had huge diesel-powered generator to cope
with power cuts. My office was near it, and I know it wasn't tested during
office hours for five years. One day there was a power cut that affected the
whole town, and the diesel engine fired up, amid a tremendous roar and lots
of filthy black smoke. Power returned for about a minute and then there was
a huge explosion and a jet of flames. Apparently the generator couldn't
handle the load, even with the engine working as hard as it could (a good
diesel grunt!) and something caught fire, igniting the diesel tank. Turned
out that it had been specced back in the days when there wasn't a PC on
every single person's desk, and it was woefully underpowered for the modern
demands that were placed on it.

On Sun, 09 Apr 2017 09:27:41 +0100, alan_m wrote:

On 08/04/2017 10:33, pamela wrote:
As a rule of thumb, what's the shortest period to switch off electronic
equipment for a reset before switching on?

I mean equipment like televisions, PVR, satellite or cable boxes, PCs,
cordless phones, clock radios, etc.

My teenage relative reset our Humax PVR by switching it off and almost
INSTANTLY switched it on again. I told him to be patient but how long
should he wait for in general?

Twenty years ago 5 seconds would have been enough but nowadays some
electronic devices (like my Samsung tv) shows its power LED for 10 or
20 seconds after the mains is disconnected. Also hard drives in
Playstations and PVRs will spin on. Telling impatient kids to go and
make a cup of coffee isn't going to work.


Twenty years ago 5 seconds may have caused a problem due to the
possibility of surges due to mains input filters.

Often if the switch off is to cold boot crashed software then the
internal PSU voltages have to decay. My rule of thumb is 30 seconds
after any front panel LED has gone out.

If you're referring to an "IBM PC" (real deal or a clone), internal
voltage decay is not an issue. The IBM PC PSU and all of its successors
incorporate a PG circuit which basically asserts a reset state on the
reset line when the PG signal is negated by the PSU detecting the initial
departure from the lower tolerance limit[1] on its most critical supply
rail (typically the 5v line back in the day - now possibly any one of the
lot with a modern ATX supply).

What this means is that if a brief drop out of the mains doesn't cause a
reboot of the PC, you can rest assured that the PSU voltages did not
suffer a transient dip. Otoh, if such a brief 1 or 2 second drop out did
result in a reboot, you knew the PC was not trying to cope with out of
tolerance voltages and, therefore, unlikely to start acting
unpredictably. A definite "Hard Reset" event is much better than random
operations due to decaying voltage rails.

Any hard disks will react to the reset signal in such a way as to
eliminate sector corruption. In the case of non-critical read operations
it simply aborts the current read and goes into standby. In the case of a
reset during a write operation to a sector (whether 512b or a 4Kb
sector), it will complete the current sector write to avoid creating a
false bad block before going into standby.

Such responses will affect any file writing ops adversely but, apart
from some exceptions (DataBases?) any processing or editing of disk files
is done by saving the edits or transcoded streams as an independent file,
leaving the original intact, or at least not deleted straight away
(typically some renaming takes place to get around the "No Duplicate File
Names!" rule). In the case of text editors and word processors, the
original version lands up named as type .BAK or similar.

Most software that is vulnerable to the consequences of a power outage,
use a strategy that minimises *total* disaster such as a complete loss of
the original file when editing it directly 'on disk'. Also, the File
System itself is designed to minimise data loss due to unplanned power
outages. This robustness was designed in from the very beginning back in
1981 since 'unplanned power outage' events were expected to be far more
likely in the case of a personal desktop computer than for a mainframe or
mini computer.

[1] This also includes the startup/switch on phase when the voltage rails
are rising from zero volts to their designated levels where the PG signal
remains negated until a short time (a few tens to a hundred or so
milliseconds) after *the* critical supply rail or rails have stabilised
at their specified levels.

The negated PG signal (an alternative /reset signal) effectively acting
as a "Handbrake" on the operations of the whole PC, only letting it loose
after the supply rails are well and truly established so as to completely
avoid the risk of unpredictable behaviour in the initial stages of the
boot process. As little as is humanly possible being left to mere chance
when it comes to both power up and power down phases of a PC's operation
(or, for that matter, *any* microprocessor controlled gadget).

--
Johnny B Good

In article ,
says...

The office building where I worked had huge diesel-powered generator
to cope
with power cuts. My office was near it, and I know it wasn't tested during
office hours for five years. One day there was a power cut that affected the
whole town, and the diesel engine fired up, amid a tremendous roar and lots
of filthy black smoke. Power returned for about a minute and then there was
a huge explosion and a jet of flames. Apparently the generator couldn't
handle the load, even with the engine working as hard as it could (a good
diesel grunt!) and something caught fire, igniting the diesel tank. Turned
out that it had been specced back in the days when there wasn't a PC on
every single person's desk, and it was woefully underpowered for the modern
demands that were placed on it.

Conversely the possibly apocryphal story I heard had the generator
tested regularly. When eventually needed it ran for a few minutes and
then ran out of fuel. The protocol didn't include top-ups!

Mike.

On Sun, 09 Apr 2017 14:33:58 GMT, Johnny B Good
wrote:

Such responses will affect any file writing ops adversely but, apart
from some exceptions (DataBases?) any processing or editing of disk files
is done by saving the edits or transcoded streams as an independent file,
leaving the original intact, or at least not deleted straight away

Unfortunately for most popular filesystems this isn't correct. What
you're describing is a feature called "copy on write" where changed
blocks are written to elsewhere on the disk, then pointers updated.
Rollback would be possible.

But yer normal FAT/NTFS (Windows), jHFS+ (Mac), ExtFS/XFS (Linux) are
all "smash on write" filesystems which write directly over previous
blocks. You have to go to fancy filesystems like BTRFS, ZFS, APFS and so
on to get the safe version.

Or, bizarrely, to iThings, which with iOS10.3 have gone to APFS and so
have copy-on-write. Which isn't much help because you can't see the
filesystem

Anyway. That's a long stray from the original question, to which I'd
answer "power it down for at least as long as it takes for the power LED
to go out".

The only super-slow device I've got right now is a monitor which
occasionally loses track of the idea of DisplayPort and needs a full
power out to do it. With the monitor in standby this can take seven or
eight seconds for the orange LED to fade off. With it powered up, it
only takes a moment.

Cheers - Jaimie
--
Tomorrow (noun) - A mystical land where 99 per cent of all human
productivity, motivation and achievement is stored.
-- http://thedoghousediaries.com/3474

On 09/04/2017 15:34, MJC wrote:
Conversely the possibly apocryphal story I heard had the generator
tested regularly. When eventually needed it ran for a few minutes and
then ran out of fuel. The protocol didn't include top-ups!

The one I know of is where the fuel pump was on the non-maintained
mains. All the tests were fine. But when the power went off for real so
did the fuel pump...

BTW I use 30 seconds too. 30 odd years of computers here, quite often
dealing directly with the HW designers so I could write the tests.

Andy

---
This email has been checked for viruses by AVG.
http://www.avg.com

"NY" wrote in message
.. .
"Roderick Stewart" wrote in message
...
In 24 years I've had more punctures than power cuts, more broken
glassware, more burnt fingers, more job redundancies, more tooth
fillings, just about every adverse event you could name. Ive even won
the lottery more often than I've had power cuts (regrettably only a
few quid but wins nonetheless).

Gosh. You are very lucky that you've only had one power cut. I think
everywhere I've lived (various houses in suburbs of towns in
Buckinghamshire, Berkshire, Oxfordhire, and now our house in a village
near a market town in Yorkshire) we've had power cuts (either many
minutes/hours or else a long enough glitch to reboot everything)

Just had one of those - it happens every once in a while.


---
This email has been checked for viruses by AVG.
http://www.avg.com