Re: Did my PSU just die?

Ant wrote:
> I also noticed when the green power light blink/flicker like crazy, it
> takes about 2-3 seconds to turn off/go away after switching the PSU off.

It takes time for the PSU to drain, once the mains switch on the back
cuts off the power flow. So that part would be normal. The LED on
my system, which connects to +5VSB, continues to run for up to 30 seconds,
before it finally goes out. It lasts a lot longer than your 2-3 seconds.

The power LED blinking means you could have a problem with +5VSB on the
supply. The power LED can also be tied into standby logic, but then
the blinking pattern would be very regular (like once or twice a second).
You can probably tell the difference between those two cases, especially
if you've seen the normal S3 standby pattern before.

It really helps, if you have a LED you know for sure is tied into
+5VSB, to use that as a replacement for a multimeter check. For
example, on all my Asus motherboards, there is a single green LED
on the board which is tied directly to +5VSB (through a current
limiting resistor). If that LED blinks, then you know your
power supply is shot. The LED is placed there by Asus,
as a warning it is unsafe to change out hardware. If the LED
is off, it means there should be no remaining power on the chassis
to worry about. It would then be safe to pull the RAM or video etc.
So that LED normally has only two states, on long term or off long term,
and blinking on that one after you've just turned on the switch
on the power supply, is definitely a fault indication.

If you're using another LED, it may have additional logic tied to
it, so the results may be a bit more ambiguous. Some systems
blink the front panel power LED, to indicate a system is in standby. As
long as you're convinced the blink you see is not a "standby" blink,
then you might conclude you're looking at a reliable indication
of a failed +5VSB.

+5VSB is weak and has a low current limit value on it. Anything
which is overloading +5VSB could also be at fault. Try stripping
the junk off the computer first, and do your debugging with
less hardware present.

Actually, I don't believe in switching on a power supply
again, once I've lost faith in it. I wait until I can find
a spare, and test with that, and see what happens. There
have been cases, where the very next time you switch on the
supply, it blows everything up. If you have the skills,
and the patience to take note of where every wire goes
on that supply, you can swap in another one and see how
it goes. That may be slightly preferable to taking
a chance on the supply you have now. Maybe there isn't anything
wrong with it, but time will tell.

You could count your "smelling of smoke" as a second symptom.

There was one guy on here, who when his computer wouldn't start,
he "turned it on and off rapidly about 50 times". Another example
of what not to do. Taking out your frustration on the power switch,
only leads to "a new computer" :-)

If you buy another supply, and that isn't the problem, you can
always use the new supply in some future project. If you have
more than one computer, it won't be long before some other
power supply fails.

If you have someone you trust to test components, let them
test the existing supply. I recommend having a test load
connected to the supply, then jumper PS_ON# to COM to turn
it on. The test load should not be a motherboard you value
for example. I've heard of people taking their gear to
a friend's place for testing and ruining the friend's computer,
which isn't a very nice thing to do. Find someone who knows the
risks, and knows how to test. I have my own load box, so if a
PSU blows, I don't lose anything. The load box uses
cheap power resistors.

http://www.dansdata.com/images/morec...esistor300.jpg

While you can buy a "PSU tester" dongle for an ATX
supply, I don't consider it a very good test. If
the PSU is really bad, it'll probably catch it. If
the supply is just a little flaky, it may say everything
is OK.

Paul

>
>
> On 12/1/2009 9:39 PM PT, Ant typed:
>
>> Hello!
>>
>> I came home to my already powered on old Linux computer desktop (leave
>> it mostly 24/7). I woke it up to log in from its xmatrix screen saver
>> and sleep. I ran a program (VirtualBox) and it hard locked up on me.
>>
>> No lights on my old PS/2 keyboard. No PS/2 mouse cursor movements. I
>> could open and close my CD and DVD drives. I tried to hit case's reset
>> button. Nothing. I held down its power button for over five seconds.
>> Nothing. Tried it again, it finally powered off. I turned it on and
>> nothing. No beeps or anything. I did see the case's green power light
>> blinking like a HDD which was odd. No fans spin, no lights inside (don't
>> remember if it had any lights, but I assume it did like on the
>> motherboard?). I looked inside, and didn't see anything odd like burned
>> outs or anything. Just lots of dusts (cleaned this case almost a year
>> ago!!). I don't have a spare PSU right now to try it and it's already
>> too late to get one. I did try another power AC cable for PSU, but that
>> didn't make any differences.
>>
>> What do you think of the symptoms? PSU-related death? I did smell
>> something burning a few weeks ago, but I never figured out where it came
>> from. Maybe this is related and a sign? If it is the PSU, then I only
>> had this Fortron FSP650-80GLC PSU (650 watts) since 5/14/2007 according
>> to my computer history
>> (http://alpha.zimage.com/~ant/antfarm/about/toys.html).
>> http://alpha.zimage.com/~ant/antfarm.../computers.txt shows my
>> detailed system setup. It's the "Secondary/Backup Computer".
>>
>> Thank you in advance.

Ant wrote:
> On 12/1/2009 10:27 PM PT, Paul typed:
>
>>> I also noticed when the green power light blink/flicker like crazy, it
>>> takes about 2-3 seconds to turn off/go away after switching the PSU off.
>>
>> It takes time for the PSU to drain, once the mains switch on the back
>> cuts off the power flow. So that part would be normal. The LED on
>> my system, which connects to +5VSB, continues to run for up to 30
>> seconds,
>> before it finally goes out. It lasts a lot longer than your 2-3 seconds.
>>
>> The power LED blinking means you could have a problem with +5VSB on the
>> supply. The power LED can also be tied into standby logic, but then
>> the blinking pattern would be very regular (like once or twice a second).
>> You can probably tell the difference between those two cases, especially
>> if you've seen the normal S3 standby pattern before.
>>
>> It really helps, if you have a LED you know for sure is tied into
>> +5VSB, to use that as a replacement for a multimeter check. For
>> example, on all my Asus motherboards, there is a single green LED
>> on the board which is tied directly to +5VSB (through a current
>> limiting resistor). If that LED blinks, then you know your
>> power supply is shot. The LED is placed there by Asus,
>> as a warning it is unsafe to change out hardware. If the LED
>> is off, it means there should be no remaining power on the chassis
>> to worry about. It would then be safe to pull the RAM or video etc.
>> So that LED normally has only two states, on long term or off long term,
>> and blinking on that one after you've just turned on the switch
>> on the power supply, is definitely a fault indication.
>
> I don't know the voltage that well. Does that +5VSB mean a problem with
> the PSU? So PSU is the culprit?
>
> With the PSU switch on and Enlight 7237-ATX mid-tower case's green power
> light blinking, I did not see any lights on my old MSI K8N NEO4-F
> (MS-7125; v1.0) motherboard:
> http://www.msi.com/uploads/prod_b615...3b77f206aa.jpg for
> two photographs/photos. and
> http://www.msi.com/index.php?func=pr...=1&prod_no=249 ...

The power supply is split in two sections.

AC --- rectifiers --- DC, big cap --+-- DC-DC conversion ------ +3.3,+5,+12,-12
|
+-- always_on_conversion --- +5VSB (standby voltage)

The +5VSB powers the motherboard during standby suspend to RAM.
The motherboard may draw one ampere during that time, some of
which powers the RAM refresh, and some for "wake-able" chips.
Even your USB coffee heater could be drawing from that rail
(which is why I suggested removing hardware from the
computer and checking for a symptom change).

+5VSB also powers supervision functions. It powers the logic
that listens for the front panel power or reset buttons. To
turn the power supply on (via the soft power control), the
motherboard logic needs a source of power, and +5VSB provides
that power.

On the power supply itself, the +5VSB circuit may have another voltage
which is used for supervision of the operation of the power supply.
So if that part of the supply winks out, the main part
of the supply may also.

Power supplies have available 2 to 3 amps on +5VSB. There is usually
current limiting (it could be a thermal limiter). The supply doesn't
have any active cooling when providing +5VSB (no fans running). If
an overload is presented, +5VSB protects itself by turning off.

Now, with that short background, we go back to your flashing LED.
The +5VSB can go on and off if there is a circuit malfunction.
The +5VSB could have become weak for some reason. Maybe
it is even overheating internally for some reason. In any case,
the slow oscillation indicates some kind of problem there.

Could the motherboard be causing it ? It might. If you had a
clamp on DC ammeter, you could measure the current flow in
the main power supply cable, and get a quick snapshot of
the condition of things. When you don't have a clamp on
ammeter and a multimeter to check voltage, it means
you're going to do a swap test a lot sooner. Without
some kind of measurement capability, swapping is about
all you can do.

The power supply is one of the most frequent sources of
trouble, which is why it is suggested to swap it with
little evidence. The smell of smoke was probably a hint.
It could be the power supply has had a capacitor failure.
You'd want to find reviews for the supply to see if
that is the case. Several models of Antec supplies
have had problems like that.

It is too bad there are no reviews here, to comment
on whether there have been failures or not. I rely
on the reviews on this site, as an indication of
design flaws. Although the Antec problems would have
shown up, after the product was no longer for sale.
The caps on my Antec failed, even though the supply
was used very little. Basically, they leaked while
in storage. It doesn't even take stress, to make
"capacitor plague" caps fail - they'll do it all
by themselves.

http://www.newegg.com/Product/Produc...82E16817103014

Paul

"Paul" <> wrote in message
news:hf5nvc$mht$...

> The power supply is one of the most frequent sources of
> trouble, which is why it is suggested to swap it with
> little evidence. The smell of smoke was probably a hint.

A practical consider is that PC repair stores (as distinct
from big box stores) will usually test your PSU without
charge and (if it tests defective) instal a new one without any
labour charge. The whole process takes less than 10 min.

--
Don Phillipson
Carlsbad Springs
(Ottawa, Canada)

On Wed, 2 Dec 2009 20:58:54 -0500
"Don Phillipson" <> wrote:

> "kony" <> wrote in message
> news:...
>
> > >A practical consideration is that PC repair stores (as distinct
> > >from big box stores) will usually test your PSU without
> > >charge and (if it tests defective) instal a new one without any
> > >labour charge. The whole process takes less than 10 min.
> >
> > I'm all for supporting local PC shops, but ultimately you
> > can easily end up paying twice as much for the same thing
> > given continual deals on the internet to contrast with, and
> > sometimes all they stock in median wattage unit is overrated
> > generics.
>
> It all depends where the knowledge can be found -- in the
> user's head or behind the counter. This string began with a
> user seeking advice because he lacked the knowledge to
> diagnose what (others thought) was most probably a failing PSU.
>
> Kony is of course right that Internet purchases are usually
> cheaper than store prices for the same item. The difference
> is that users who lack knowledge to specify their purchase
> can get their questions answered quicker over the counter
> than via Internet, i.e. can get more knowledge. We all
> knew beforehand that timely knowledge is worth money.

There is nobody working in a store that sells computer power supplies
who is qualified to test a PSU, or offer advice about a PSU. If they
were qualified to give such advice or test a PSU, they would not be
working in a retail store. So your best bet is to use the Internet for
advice AND purchases. -Dave

"kony" <> wrote in message
news:...

> >A practical consideration is that PC repair stores (as distinct
> >from big box stores) will usually test your PSU without
> >charge and (if it tests defective) instal a new one without any
> >labour charge. The whole process takes less than 10 min.
>
> I'm all for supporting local PC shops, but ultimately you
> can easily end up paying twice as much for the same thing
> given continual deals on the internet to contrast with, and
> sometimes all they stock in median wattage unit is overrated
> generics.

It all depends where the knowledge can be found -- in the
user's head or behind the counter. This string began with a
user seeking advice because he lacked the knowledge to
diagnose what (others thought) was most probably a failing PSU.

Kony is of course right that Internet purchases are usually
cheaper than store prices for the same item. The difference
is that users who lack knowledge to specify their purchase
can get their questions answered quicker over the counter
than via Internet, i.e. can get more knowledge. We all
knew beforehand that timely knowledge is worth money.

--
Don Phillipson
Carlsbad Springs
(Ottawa, Canada)

> > Hmm, seems PSUs aren't doing so well. Who is the best these days
> > even if we have to pay a little more.
>
> You tell me!

BFG or Corsair, either above 600W

On Thu, 03 Dec 2009 22:05:36 -0800
Ant <> wrote:

> On 12/3/2009 5:24 AM PT, Dave C. typed:
>
> >
> >>> Hmm, seems PSUs aren't doing so well. Who is the best these days
> >>> even if we have to pay a little more.
> >>
> >> You tell me!
> >
> > BFG or Corsair, either above 600W
>
> That's interesting. Those are new brands for PSUs for me. Wasn't OCZ
> to be good or something based a few years ago?

OCZ is OK. Still is. Not as good as BFG or Corsair though.

> How are SeaSonic,

If you can afford it, SeaSonic is still a great brand.

> Antec,

Antec doesn't make power supplies. You can buy a power supply marked
"Antec". Many of those used to be made by Seasonic, so Antec used to
be pretty good quality. But Seasonic doesn't make the Antec marked
units anymore. The only decent power supplies marked Antec are marked
Antec "Signature". But those are wicked expensive.

> and Enlight doing for PSUs these days?

Not sure about Enlight, but suspect that they'd still be good.

For a while now, power supply manufacturers have been producing crap
under 650W, and only bothering to TRY to produce something of good
quality if it's rated at 650W or higher. And among the 650W or higher
units, BFG and Corsair are dominating, as far as build quality goes.
That's not to say there aren't other good units. But if you are
looking for a power supply right now, think 650W or higher, and check
BFG or Corsair first. -Dave

Ant wrote:
> On 12/3/2009 2:42 PM PT, Dave C. typed:
>
>>>> BFG or Corsair, either above 600W
>>>
>>> That's interesting. Those are new brands for PSUs for me. Wasn't OCZ
>>> to be good or something based a few years ago?
>>
>> OCZ is OK. Still is. Not as good as BFG or Corsair though.
>>
>>> How are SeaSonic,
>>
>> If you can afford it, SeaSonic is still a great brand.
>>
>>> Antec,
>>
>> Antec doesn't make power supplies. You can buy a power supply marked
>> "Antec". Many of those used to be made by Seasonic, so Antec used to
>> be pretty good quality. But Seasonic doesn't make the Antec marked
>> units anymore. The only decent power supplies marked Antec are marked
>> Antec "Signature". But those are wicked expensive.
>>
>>> and Enlight doing for PSUs these days?
>>
>> Not sure about Enlight, but suspect that they'd still be good.
>>
>> For a while now, power supply manufacturers have been producing crap
>> under 650W, and only bothering to TRY to produce something of good
>> quality if it's rated at 650W or higher. And among the 650W or higher
>> units, BFG and Corsair are dominating, as far as build quality goes.
>> That's not to say there aren't other good units. But if you are
>> looking for a power supply right now, think 650W or higher, and check
>> BFG or Corsair first. -Dave
>
> Interesting. I will have to check those out. I might have to get a new
> PSU if I do have a dead/broken one.

Just remember that the power supply industry uses a lot of contract
builds. Companies like BFG or Corsair won't be making the supply
themselves. They'll contract to someone else. Seasonic makes their
own supplies. OCZTechnology bought PCPower, but it is likely in
either of those cases, that those are contract as well. PCPower has
done pretty well in the past, in picking suppliers for their power
supplies, as well as understanding issues when they arise. Antec is contract,
using in the past HEC, ChannelWell, now Seasonic and Delta, and who knows
who else.

There are web sites which discuss the various contract suppliers.
So if you do some research on your prospective purchase, you
may be able to tell who really makes it. A review here may
mention the actual supplier.

http://www.jonnyguru.com/

With all of this contracting, your reputation means nothing. The
supplies you sell, are only as good as the current contract you
set. Maybe you change suppliers in mid-stream, and don't bother
to change the part number. So while it is fun to pretend that
"brand X has a good reputation", it all depends on how they
handle their contractor. A good company, would do thorough
surveillance of the product being shipped, so goofups don't
get out the door. When you look at the history of Antec,
like supplies that didn't work with certain brands of motherboards,
or the recent issues with UPSes, they don't seem to be checking
their supplies in a practical way. To me, that is the value that
the "brand name" of the supply should be adding, the checking they
do to make sure that the thing they contracted, actually works
in the real world.

There was at least one contracting out type company (who shall remain
nameless), where none of the staff at the company had a foggy clue
how power supplies worked. In their web based tech support forum,
they would berate customers with problems, basically blaming all
the product problems on the customer usage. That is the most extreme
case I've seen, of a "small North American office with four guys in it",
playing the contract game. The things I saw in that forum were pitiful.
None of the brands named above would be quite that bad. Established
companies that don't like dealing with customers, simply make it hard
to contact them. There is at least one power supply company,
where in Canada, you can't reach anyone for warranty claims.
You have to make a stink, before they realize they don't have
any customer service to speak of. That makes it rather easy to
do business, when you never have to handle an RMA...

Paul

Ant wrote:
> On 12/5/2009 12:12 PM PT, kony typed:
>
>> On Sat, 05 Dec 2009 03:02:48 -0800, Ant<>
>> wrote:
>>
>>> We tried a new Antec Basiq BP550 Plus 550W Continuous Power ATX12V V2.2
>>> Modular Active PFC power supply
>>> (http://www.antec.com/Believe_it/product.php?id=NzA= ). Nothing came on
>>> the screen. Not even POST!
>>
>> At that point, clearing CMOS would be the next step.
>
> That didn't work either.

I might have tried some beep tests. For example, if the processor
was able to read the BIOS flash chip, it might be able to "beep"
the computer case speaker, if the RAM and video are missing. The
beep pattern might involve one, two, or three beeps, in a repeating
pattern. To make the beeps requires working BIOS code, which would
be proof that some small portion of the system was working.

The BIOS POST sequence does tests, and if it finds the RAM or the
video missing, it is supposed to beep the case speaker. And you
can use a response like that, as a hardware test case.

In terms of things that fail, from top to bottom they might be

1) PSU
2) Hard drive (some last only two days)
3) RAM (checking the reviews, some don't seem to last that long)
4) Motherboard (bad caps or PCI Express slot blown by static)

The processor hardly ever gets damaged.

When the PSU fails, things that are more "isolated" have a
better chance of survival.

PSU ------ Hard_Drive (direct connection, overvoltage might kill it)
------ RAM_voltage_regulator ----- RAM (some isolation)
------ Vcore_voltage_regulator --- CPU (some isolation)

I can give other examples, like the time I thought I ruined a lab
instrument. I was working in a live chassis (open face power
supply connected to 120V). I'd commented to my managers that
this wasn't a very bright way to do things. And one day, I'm
working in the lab, and the ground of my lab instrument dropped
into the 120V circuit. Bam!

Our instrumentation repair department are miracle workers. In
only two days, my buddy there found a couple chips in the front
end of the instrument that were blown. Everything else was fine.
Only the thing nearest to the fault was ruined, and I was
saved the embarrassment of explaining how a very expensive
instrument was ruined.

You can see the same kind of effects, when there is lightning
damage. Lightning can be induced into power conductors, or can
follow the modem cabling or Ethernet wire from a broadband modem.
The further a component inside the computer is, from those
connections, the better the odds it'll survive.

So if, when the power supply failed, it damaged stuff, you
should look at how the various items get power, to get some
idea of the odds they got damaged. The hard drive, has a
transient suppressor on each rails, and that is intended to
clip inductive transients. It isn't meant to crowbar a
freaked out power supply. So the hard drive is going to
feel it, if the power supply puts out more voltage
than it is supposed to. Things like CPU and RAM, have
regulators in front of them, that, if you're lucky,
provide some protection. The regulators have some max
voltage input allowed, which could be a bit higher than
the normal operating voltage.

I think you could continue to do some testing on your
old components. The beep tests might help you.

Paul