Power Surge

[David LeBrun]

Greetings
I'm currently restoring a system that was hit with a power surge where
the following were fried:
PSU
Motherboard
Hard Drive
CDROM
CDRW
Network (PCI)
Audio (PCI)
However the following components from the same system survived:
CPU
RAM (all 3 sticks that were installed)
AGP video
Modem (PCI)
Floppy
Speakers
Monitor
Keyboard/Mouse

I just find it curious that components I thought were sensitive to
voltage like the CPU and RAM would survive while other components
wouldn't. The core on the audio card actually popped (I found the
center of the silicon at the bottom of the case). I keep telling
people to get a UPS or a good surge protector but no one listens until
they have to pay $$$

Anyone else have a horror story to tell?

Dave.

 

[Ben]

As for surge protectors, they can do SOME good, but the UPS is the
way to go. Particularly if the UPS does a conversion from A/C to D/C
and back to A/C during normal operation. Many of the cheap ones simply
pass through the normal A/C with a relay and then do a restoration via
the battery when it fails.

If I get one of these UPS surge protectors would that mean I could use
my computer during a thunder storm?

thanks,
Ben

 

[Vanguard]

If I get one of these UPS surge protectors would that mean I could use
my computer during a thunder storm?

thanks,
Ben

Better to install a whole-home surge protector rather than the cheapie
surge protectors you buy at the store. They cost about the same (unless
you need to have an electrician do the wiring for the home surge
arrestor). Then everything in your house is protected. however, surge
arrestors only protect against surges. They don't protect against sags.
Your power supply would have to suck more current to offset the loss in
voltage until the sag became too low (your lights and television will
work but not the switching power supply in your PC). Surge protectors
also do little to condition the line (but then some cheapie UPSes are
just pass-through units, too). I prefer to get a UPS that generates the
power output rather than just pass it through. A UPS with an isolating
transformer is even better but then they get really heavy; mine weighs
60 pounds for the case with transformer and another 60 pounds for the 2
batteries, for a total of 120 pounds (and has true sinusoidal output and
too many other goodies to list, but then it cost as much as a high end
PC). But even a cheapie UPS is usually better than a surge protector.
You could spend $60 on a good point-of-use surge protector (i.e., you
plug it in the wall install of the point-of-entry whole-home surge
arrestor) and hope that it is still working (since many never tell you
when they've gotten fried) or you spend twice that much on a low-end UPS
that protects better. When the power goes out, the surge protector
doesn't do anything. It can be handy to continue computing while the
power is out. You're television don't work so sit at your PC. You can
check the weather, especially if that caused the outage. You can
contact your power company (since calling in puts in the queue with
everyone else trying to report the outage). You can even hook up a
low-power intensity lamp so you can see without having to drain the
batteries in your flashlights. I actually have a small UPS left over
from a prior system used to power the cable modem and a flourscent lamp,
and it's small enough to tote around the house to use elsewhere. Sure
beats sitting in a dark house.

 

[w_tom]

What is this sag that causes problems? Computer power
supplies are some of the most robust electronics in the
house. AC electric can drop to 90 VAC (for 120 VAC) or 190
VAC (for 230). The computer must operate interrupted. Even
with a full load of peripherals, the computer must power up
just fine when AC mains voltage has sagged that low.

Incandescent bulbs would be at less than 40% intensity and
the computer must work just fine. However many buy power
supplies only on on specification - price - and get what they
paid for. Sags must not damage any computer. However if you
are still using disk filesystems such as FAT, then you still
data destruction from excessive brownouts or blackouts can
occur. Again, the UPS is nice, but not necessary for hardware
protection if computer hardware is properly selected and
installed.

I never use plug-in UPSes or any other grossly overhyped and
ineffective plug-in protectors. I go online during every
thunderstorm. Never shutdown for such storms. Never unplug
anything. Suffered some direct strikes and never suffered
damage. No damage because the technique is same well proven
methods demonstrated in 1930s research papers.

Those plug-in protectors fundamentally violate those
principles. Summary is provided in "Opinions on Surge
Protectors?" on 7 Jul 2003 in the newsgroup
alt.certification.a-plus or
http://tinyurl.com/l3m9

BTW, plug-in surge protector and plug-in UPS use equivalent
circuits to provide the same ineffective protection. There is
no such thing as blocking or stopping a surge even with a
transformer. Transformer can help to divert the surge. But
again, it still requires the well proven 'whole house'
protection 'system'. Yes - a 'system' that includes the most
critical 'system' component - earth ground.

Plug-in UPS serves one primary function - data protection.
Surge protector is defined by its most critical component -
that plug-in protectors just don't have and forget to
mention. Read that above, previous summary discussion for
details.

 

[Vanguard]

What is this sag that causes problems? Computer power
supplies are some of the most robust electronics in the
house. AC electric can drop to 90 VAC (for 120 VAC) or 190
VAC (for 230). The computer must operate interrupted. Even
with a full load of peripherals, the computer must power up
just fine when AC mains voltage has sagged that low.

Incandescent bulbs would be at less than 40% intensity and
the computer must work just fine. However many buy power
supplies only on on specification - price - and get what they
paid for. Sags must not damage any computer. However if you
are still using disk filesystems such as FAT, then you still
data destruction from excessive brownouts or blackouts can
occur. Again, the UPS is nice, but not necessary for hardware
protection if computer hardware is properly selected and
installed.

I never use plug-in UPSes or any other grossly overhyped and
ineffective plug-in protectors. I go online during every
thunderstorm. Never shutdown for such storms. Never unplug
anything. Suffered some direct strikes and never suffered
damage. No damage because the technique is same well proven
methods demonstrated in 1930s research papers.

Those plug-in protectors fundamentally violate those
principles. Summary is provided in "Opinions on Surge
Protectors?" on 7 Jul 2003 in the newsgroup
alt.certification.a-plus or
http://tinyurl.com/l3m9

BTW, plug-in surge protector and plug-in UPS use equivalent
circuits to provide the same ineffective protection. There is
no such thing as blocking or stopping a surge even with a
transformer. Transformer can help to divert the surge. But
again, it still requires the well proven 'whole house'
protection 'system'. Yes - a 'system' that includes the most
critical 'system' component - earth ground.

Plug-in UPS serves one primary function - data protection.
Surge protector is defined by its most critical component -
that plug-in protectors just don't have and forget to
mention. Read that above, previous summary discussion for
details.

If the voltage drops on the input side but you still have the same power
demands on the output side and where you must maintain the same voltage
requirements with the same demand for current, how does the power supply
make up for the deficiency in input voltage?

 

[kony]

If the voltage drops on the input side but you still have the same power
demands on the output side and where you must maintain the same voltage
requirements with the same demand for current, how does the power supply
make up for the deficiency in input voltage?

Switching opwer supplies monitor output voltage. They typically
maintain the same switching frequency but as output drops, there will
be increase in the duration of the "on" cycle to keep the voltage up.
So, whether the input changes or the load changes, or both, the
appropriate duration of on-cycle will change, up until the limits of
the heat dissipation (longer on cycle generates more heat) or voltage
drop below minimal allowed, generally due to inadequate sized
transformer per load applied.


Dave

 

[w_tom]

Others have summarized your answer. Some numbers. A
switching power supply first ups the 120 VAC to about 300
volts DC (and yes, those 300 volts did hurt). Then it
oscillates that 300 DC voltage through a transformer,
rectifies it, and creates the +5, +3.3, + 12 etc. Notice the
so many transitions. 120 VAC to 300 VDC to various low
voltage AC to regulated 3.3, 5, and 12 DC.

With all those transitions, it becomes easy to take a widely
varying, AC input voltage to create a noise free, galvanically
isolated, robust, and well regulated DC voltage.

 

[David LeBrun]

Sounds like you (or your customers) need to check into using better
power supplies. When not protected by a UPS, mine have simply popped
the breaker on the backside or blown a fuse (which sometimes requires me
to open the PSU to replace it) or just tripped on the overvoltage and
you wait until it resets and the line voltage is okay to power back on.

By the way, it isn't just surge protectors and, better yet, UPSes for
which customers never listen. They don't listen regarding backups,
either. When they whine about really needing to get their data back,
they look like deer in headlights when you ask about their backups.


--
____________________________________________________________
** Share with others. Post replies in the newsgroup.
** If present, remove all "-nix" from my email address.
____________________________________________________________

I'm not sure how it was connected when the surge happened. I know it
now has an Enermax PSU which I've never had a problem with. I also
know that 2 or 3 other people have worked on the system before it came
to me so no telling how or what caused the problem and what was done
to try to fix it. I opened the case and found remnants of drink
spills, a cpu heatsink with about 5 pounds of dust under a fan whose
bearings are on the way out, wiring nest sitting right in front of the
cpu heatsink, no case fan and the psu vent blocked with another wiring
nest. Anyone that actually knows what they're doing should have fixed
at least one of these problems by force of habit as far as I'm
concerned. First thing (rather second...first I just stood there
shaking my head) was to start cutting a hole in the case for a fan! I
event went as far as lapping the heatsink because I knew heat was
going to be a real problem. Don't know who did the work on it when it
came to me but I know I wouldn't have them work on a system belonging
to my worst enemy! What really bothers me is there are some people
who charge an arm and leg to build/repair systems and what you get is
something that looks like it came out of a meat grinder.
I had installed backup software originally and with the burner
installed there shouldn't have been any problem with the hard drive
failing but like you said...they just never listen.

Dave.
(sorry for the rant)

 

[w_tom]

The plug-in UPS serves one primary function. It protects
data from power problems. It does not protect hardware. Did
you read the long list of exemptions attached to that
warranty? Good luck if you need to have that warranty
honored.

I have a UPS right here for 120 volt operation. To generate
(in battery backup mode) what is considered 120 volts, it
outputs two square waves of 200 volts with a 270 volt
transient between those square waves. This output could be
destructive to some appliances such as small electric motors.
However computers are so resilient that even this harsh UPS
output will not harm computers.

The plug-in UPS is for data protection. Computers already
(should) have internal protection which is why many UPSes do
not damage a computer.

 

[Vanguard]

Switching opwer supplies monitor output voltage. They typically
maintain the same switching frequency but as output drops, there will
be increase in the duration of the "on" cycle to keep the voltage up.
So, whether the input changes or the load changes, or both, the
appropriate duration of on-cycle will change, up until the limits of
the heat dissipation (longer on cycle generates more heat) or voltage
drop below minimal allowed, generally due to inadequate sized
transformer per load applied.


Dave

And that means more current is drawn on the input side, right?

 

[Vanguard]

<snip>
... lapping the heatsink ...

Some folks think I'm nuts for doing this. But I've seen even new heat
sinks with copper contact bases that look like they're used. There's
something really satisfying about getting a mirror finish that makes
excellent. Of course, I've seen where someone decided if a wee bit of
thermal transfer paste is good then lots is better and not realize
they've reduced the rate of heat transfer.

 

[kony]

And that means more current is drawn on the input side, right?

Yes


Dave

 

[David LeBrun]

This one was a definite ameteur...seated the old cpu in the new board
using the existing heat transfer pad that was originally on the boxed
heatsink! Combine that with a ton of dust in it and a fan with a bad
bearing...it soooo sad its actually funny!

Dave

 

[David LeBrun]

That's not nuts... nuts is when you use polishing compound to lap down
that small laser-etched dot pattern in the corner of an Athlon XP
core.



Dave

How about sanding down the small components on top or better yet
the core itself...i mean really if you want good contact grind that
sucker so its flush with the board!!! (I think I need sleep)

 

[David LeBrun]

The plug-in UPS serves one primary function. It protects
data from power problems. It does not protect hardware. Did
you read the long list of exemptions attached to that
warranty? Good luck if you need to have that warranty
honored.

Lets see...
"This warranty does not protect against acts of god (other than
lightning) such as flood, earthquake, wind, war, vandalism, theft,
normal-use wear and tear, erosion, depletion, obsolescence, abuse,
non-authorized program or system, or equipment modifications or
alteration."
"All products must be plugged directly into a power source and must
not be daisy-chained together in serial fashion with other UPS, power
strips, surge protectors or extension cords. Any such installation
voids this warranty."
"This warranty will not cover claims for damage resulting from
telephone or CATV line transients unless the equipment is properly
connected to one of the UPS models that offer telephone line
protection. This warranty will not cover claims for damage to
connected equipment resulting from coaxial line transients."
"This warranty is null and void if the unit was improperly installed,
altered in any way, tampered with, or improperly maintained
(including, without limitation, the replacement of depleted batteries
when indicated)."
"This warranty is null and void if the repair or replacement of the
damaged connected equipment is covered under a manufacturer's
warranty."

Plus the miriad of terms about data loss and who/when the claim is
made and any inspections that may be required at their determination.
The only way into the system not protected by the UPS is via the
speakers which are plugged into a "surge supressor" and via the A/V
dongle which connects my satellite receiver to my vid card...the
receiver has coax running from the dish (which is properly grounded).

I have to assume that these things are ligit considering all the laws
about false advertising and consumer goods etc etc. I don't usually
run my equipment during a t-storm so lightning isn't my concern. Its
more for protection from outages...the "data protection" you
mentioned...and other anomalies from the utility company. But the
just have to better than plugging this stuff directly into the wall
right?

Dave.

 

[w_tom]

Newsgroups have these stories often:
Steve Uhrig on 17 Jun 2003 in the newsgroup
comp.home.automation entitled "UPS for computer and TV"

I lost the modem board in an early generation commercial high
volume fax which was 'protected' by an APC UPS.
I read the terms of their warranty, which I had saved together with
the purchase receipt, and contacted them to submit a warranty claim.
I was nice and polite and had everything documented including photos
of their product installed next to the fax.
They laughed in my face. Almost could not have been more insulting.
I wrote to the executive management of the company, copied customer
service, sent both return receipt to prove they received them, and
never got the courtesy of a reply.

If you believe a warranty proves quality, then a superior
warranty also proves a Hyundai is more reliable than Toyota
and Honda. Warranties never proof anything technical. In
surge protector, the best protectors have no warranty. Proof
of effectiveness is found in basic science. That science has
long been proven repeatedly since the 1930s.

Appliances such as computer power supplies already contain
effective protection as posted earlier with that list of
voltages. But that protection may be overwhelmed if a
transient is not earthed before it can enter a building.
Therefore effective protection is located at the service
entrance. In the meantime, a plug-in UPS connects appliance
directly to AC mains when not in battery backup mode. So
where is the protection? Not from that plug-in UPS. So they
forget to even mention which type of transients they protect
from.

That plug-in UPS serves one primary function - data
protection from blackouts and extreme brownouts.

 

[Vanguard]

In the meantime, a plug-in UPS connects appliance

directly to AC mains when not in battery backup mode. So
where is the protection? Not from that plug-in UPS.

That plug-in UPS serves one primary function - data
protection from blackouts and extreme brownouts.

Really depends on the UPS that you get. If you get the cheapies that
simply shunt the AC from the input to the output, maybe with some line
conditioning, then you are still susceptible to line problems (depending
on how good in the line conditioning). If you get a UPS that generates
the output power then you are further protected. If you get a UPS that
has an isolating transformer (which has to be huge and heavy because of
the current demands and which then by necessity has to generate the
output power) then you are the most protected. I also prefer to get one
that has a true sinusoidal output. You can get a UPS like you describe
just to protect your OS from being shut down hard. You can pay more
that provide surge protection and line conditioning. Or you can pay a
lot for one that isolates, generates, and provides very clean output.
Obviously the best way to determine how well any line protection is
working is to get an oscilloscope that you can connect to your PC to
record the source of your power. Hey, there's the next that I need.

 

[David LeBrun]

Ok...I am not disputing the science behind all this...I couldn't even
begin to...

My intent when I started this thread was to get people's experiences
with power related death to components because I was simply curious
why the cpu, memory and video card (items which I would expect to go
first) survived when other components (which I thought were more
durable) failed. I agree with you 100% that a UPS protects data as
they have saved my stuff on numerous occasions.

The technical details, physics or electrical engineering aspects of
this is all over my head. In a nutshell, would I be correct in saying
that a good UPS would be one where the connected equipment is always
running from battery power and that the battery power is always being
recharged? and that most if not all consumer/retail products do not
provide this function? I thought I had a decent handle on what was
going on with all this but you've proven my ignorance.

Dave.

 

[w_tom]

Described is an expensive UPS system that ties up and uses
most system components constantly without any appreciable
advantage. Better UPS systems simply run off of AC mains when
not in battery backup mode. Battery backup system, that is
more expensive to operate and that may more likely fail when
needed, instead remains in standby. Then when backup power is
required, different (unused) means provide the backup. For
example, the telco switching station uses a battery backup
system (with battery life expectancy now on the order of
twenty years). The batteries in turn provide power until the
backup generators can take over. But system runs directly off
AC power almost constantly. Backup stays in reserve.

Remote concentrators will run directly off of AC until power
is lost, then must run at least 4 hours on internal battery
backup. If power is lost for longer, a mobile generator is
connected to maintain power. Again power directly from AC
mains means less expense and the battery backup system is more
likely to be 100% functional.

Some plug-in UPSes output very dirty power when in battery
backup mode. For example this one creates 120 VAC by
outputting two 200 volt square waves with a 270 volt spike
between those waves. Not very clean, possibility dangerous to
small electric motors, but more than sufficient for a
computer.

Failures come from numerous sources. Too many are so
brainwashed by advertising and urban myths that literally
every failure must be a power surge. Most component failure
is due to internal manufacturing defects. Since CPUs must not
fail, then their manufacturing defects are avoided using
statistical process control to remove the random human
factor. Other requirements include very clean gases (gases
are used to construct semiconductors) on the order of 99.9999%
pure. Some manufacturers may only use 99.99% or 99.999%
purity which accounts for some of their defects. Most
failures are simply manufacturing defects. See the big stink
about defective Taiwan electrolytic capacitors as but one
example.

ICs can be damaged by static electric shock - such as
manufacturing defects in the factory or mishandling in
distribution. A static electric shock today can mean
electronic failure months later. Add a touch of
counterfeiting parts.

Heat is often over hyped as a reason for failure. When heat
is associated with failure, failure really is a defective
component whose problem was made obvious by higher
temperature. But some humans must feel heat was a problem
because that temperature, quite normal for a semiconductor,
was too hot for human skin. Too many blame heat mostly due to
ignorance of what is happening inside that semiconductor.

Another reason for failure, for example, is cost controls.
Transistor selected that was marginal for the task but that
cost less. Insufficient experience with the design because
some bean counter worries about his cash flow. Timing failure
due to one unnoticed characteristic of the design. Noise
because some bypass capacitor was not installed somewhere on a
PC board.

Power supply regulator can fail. But power supplies even 30
years ago had overvoltage protection circuits that made damage
by regulator to motherboard, disk, etc practically
impossible. However because so many today use price as their
only specification, then many failures are directly traceable
to that cost control mentality - the missing overvoltage
protection that all power supplies must include - but that
many do not.

ICs have upper voltage limits. For example a 5 volt IC can
be damaged by 9 volts. However when that same semiconductor
becomes part of a system - ie motherboard - then the 'system'
tends to make that IC more resilient. Components even inside
the IC contribute to IC protection when IC becomes part of a
'system'. And when that motherboard is installed in a
chassis, then that 'system' permits even 20,000 volt static
shocks on case to not damage ICs. Depends on designer's
knowledge. But the point is that ICs, so easily damaged by
almost no voltage can withstand thousands without damage as
they become part of a larger 'system'. Therein explains the
more resilience in parts such as CPU and video controller when
part of a bigger system.

Infant mortality is but another failure mode. Therefore
good system developers use burn-in testing. Many computer
assemblers don't even know what burn-in testing is. The
computer is put in a 100 degree room - upper end of normal
operating temperature - and executes extensive diagnostics.
Then computer is cooled to lowest normal temperature -
typically just above freezing. More diagnostic testing.
Process is then repeated. More extensive testing may also
heat or cool computer to temperatures well above or below
normal operating temperatures - but within spec and without
power. Then temperature is taken back down to upper or lower
operating temperature, powered up, and tested again.

Burn-in testing not performed on every system but on a
statistical processing basis. Defective components identified
before it ships. Defective memory, that works intermittently
at 70 degree room is quickly found by this testing method.

Just some reasons why electronics fail.

 

[w_tom]

Let's assume the $500 UPS that operates always from
the battery. AC mains recharge battery while battery runs
inverter (AC power) into computer. But again, one must first
understand, electrically, what a surge is. Destructive surges
are common mode. Common mode transients cannot be absorbed by
that battery. A common mode transient appears equally on both
sides of battery. Battery sees no transient voltage as
transient continues into and damages adjacent appliance. The
battery would absorb a differential mode transient. But
destructive transient are common mode which means earth ground
is essential to provide protection. Earth ground is why the
building wide UPS provides effective protection and why the
plug-in UPS (any type) cannot provide such protection.

Second, safety ground wire that connects AC mains surge
directly to computer motherboard. Surge protector circuit on
front end of UPS simply shunts that destructive surge onto
green safety ground wire to bypasses the UPS & computer power
supply. Safety ground wire connects transient directly to
computer electronics. This wire that bypasses UPS is a most
common source of transients that damage computer modems.

Third, we need not see how surges damage by following
circuits. Where is the numerical specification for that UPS
that claims common mode surge protection? Please cite that
specification.

In the meantime a long and well proven principle of surge
protection remains. This principle has existed and been
repeatedly proven since before WWII for good reason. Surge
protection is about earthing a surge. A surge protector (even
the one inside that UPS that is actually same circuit found in
power strip surge protectors) is only as effective as its
earth ground.

Four, if that more expensive UPS provides galvanic isolation
from surge damage, then same galvanic isolation that exists in
PC power supplies would provide the protection - making that
UPS protection redundant - providing no additional
protection. Computer power supplies already have effective
protection. But that existing protection assumes the incoming
surge will be earthed before it can enter the building. A
plug-in UPS provides no additional protection because it does
not have that essential 'less than 10 foot' connection to
earth ground.

Many reasons for why a more expensive UPS may not be better
protection. One - common mode transients are not even seen by
the battery. Two - surge can bypass the UPS completely using
safety ground wire. Three - UPS does not even claim to
provide such protection. Four - any protection provided by
that UPS already exists in the power supply.

Bottom line remains. A surge protector is only as effective
as its earth ground. Furthermore, that earthing protection is
so inexpensive. Plug-in UPS has no dedicated connection to
earth ground in contrast to what a serious building wide UPSes
provides. Plug-in UPS avoids any mention of earthing so
that customer will not ask embarrassing questions such as,
"what about common mode transient protection?".

Even dirty (not sinusoidal) power from a step wave UPS does
not adversely affect electronics. The one type of electrical
'dirt' that is so destructive is the common mode transient -
what a plug-in UPS does not even claim to protect from. Using
the 'purer' sine wave provides nothing useful to a computer's
life expectancy.