Thread: Head-Scratcher
View Single Post
  #7  
Old April 17th 18, 06:47 PM posted to alt.comp.periphs.mainboard.asus
Bill Anderson
external usenet poster
 
Posts: 249
Default Head-Scratcher

On 4/13/2018 12:45 AM, Paul wrote:
Bill Anderson wrote:
On 4/12/2018 4:06 PM, Paul wrote:
Bill Anderson wrote:
Golly this group has gone quiet lately.Â* This has always been my
go-to place for help with my computers, but now...Paul, are you
still here?

About three years ago I built a desktop machine around an old Asus
P5Q Pro Turbo MBO I had lying around and donated it to a local
nonprofit.Â* I set it up with Win10 and two accounts -- an
administrator account I use as a volunteer to reproduce their CDs
and DVDs and to print their photos on a large-format printer, and a
guest account anybody can use when I'm not around, which is most of
the time.Â* I think it's seldom used that way, but who knows?Â* I'm
not around most of the time.

The computer has been working fine since 2015, but the other day I
was hoping to burn some CDs on it and the computer's power kept
cutting out.Â* I couldn't get it to power up at all at first, so I
fiddled with the power button and the power cord and power supply
switch and plugged and unplugged and suddenly it started.Â* And as
Win10 was loading, the power cut off.Â* Then back on, thenÂ* off...so
I brought the thing home to work on it.

At first it powered right up, then the power cut off.Â* And then I
noticed the case power button was sitting cockeyed...aha!Â* So I
fiddled with it for a minute and the power came on and voila!
Problem solved. Or so I thought...because the weirdnesses continued.

As long as I had the thing at home I figured I ought to go ahead and
update the OS to the latest version as Win10 kept nagging me to do,
so off to the races I went, only to be told after an inordinately
long (I thought) download and update process, Win10 reverted to its
older version with an apology that the update had failed.

So I tried again and after a few hours I got the same Windows
apology. Just for the heck of it, I ran a Memcheck test and found no
problems on 32 gigabytes of RAM.

All right, but before I tried anything else I figured I'd better
back up what I had.Â* Since I would be backing up a C: drive, I
plugged a spare SATA drive onto the MBO (to hold the backup) and
booted from my Acronis True Image 2014 boot disk.Â* Didn't boot
Windows--booted from CD and so far, so good.Â* But after setting up
the backup parameters and starting the backup process, just as files
began copying -- snap, the computer power went off. Dead.Â* Just like
that.

But the case power button turned things back on and I tried again
and at exactly the same point in the backup process...snap, the
power went off again.Â* It was a sudden total shutdown with no warning.

So...desperate for a backup now, I booted into Win10 and just copied
all my important files onto that backup drive, and with them safely
stored away I re-installed Win10 on the computer.Â* It was a fresh
install on the original hard drive, with the old partitions deleted
-- nothing left over on the hard drive.

And Win10 (ver. 1709) installed in a breeze and I copied all my
important files back onto it and I began installing all the programs
I needed and they installed without a hitch and things were going
great last night when I shut it all down to go to bed and pick up
this morning where I left off.Â* Just a few more programs to get
right, and I can take this thing back to the nonprofit.

And then this morning when I booted the computer by pressing the
spacebar (I did NOT touch that power button) the power came on for
at most a second and a half, then went off.Â* And as I stood there
looking at it, the power came on again, then off. And again and
again and again...this wasn't going to stop.Â* Just cycling
on/off/on/off. So I pushed the power button and held it in and soon
the cycling stopped with power off.Â* Then I pushed the button again
and the power came on and the cycling started again.Â* Pushed the
button...power went off.Â* Pushed the button, power came on and
stayed on.Â* And it has remained on with no problems at all for about
the past hour.

So what's the problem? Can't be Win10 as the power cycling is taking
place well before the computer even posts.Â* Case power button?Â* I no
longer think so, as it was untouched after shutdown last night and
power up this morning.Â* I used the Asus boot with spacebar feature
in BIOS...not that button.Â* Motherboard?Â* Possibly.Â* Loose cable?
Possibly, though I've checked connections throughout the system.
Power supply? Probably, I guess, though that computer is running
steadily over there right now with no apparent issues at all.

I suppose I'll try to track down a power supply in town today,
though things like that are getting harder to find locally these
days.Â* I may have to use Newegg or Amazon.Â* But I'd hate to buy a
new power supply only to learn it was something else.

Ideas?Â* Ever experienced anything like this yourself?


In a traditional era, computers had two buttons on the front.

A NO (Normally Open) RESET button which momentarily closes.
A NO (Normally Open) POWER button which momentarily closes.

In an emergency, you can go to the PANEL header, remove the
two wires of the POWER button and move the RESET button wires to
where the POWER button would normally go.

Then, use the RESET button to start the machine. (Leaving
you for the moment, with no functional RESET button.)

Alternately, if you keep a couple spare normally open switches
with twisted wire pair on hand, you can plug one of those in,
to take the place of the POWER button.

This covers cases where the POWER button got crushed.

You can also start or reset a motherboard, by removing the
sets of wires, and using a screwdriver tip, *but* that takes
a steady hand and very good eyesight. It's fine to be
doing that if a motherboard is sitting flat on your bench
for testing. Not a good idea if the motherboard remains
inside the case. Too hard to see in there.

*******

THERMTRIP will turn off the power. I don't think it's supposed
to turn on like that again right after a THERMTRIP event. It
should take a button push.

However, there is a BIOS setting on some machines, which "restores
power state to previous" after a power loss. Perhaps the restored
power after an event, is caused by the choice of BIOS setting.
I usually leave my machine set to "remain OFF after power is
restored".

If the motherboard VCore that powers the processor trips on
OCP (over current), the motherboard should definitely stay
off until you turn the power switch on the back to OFF for
30 seconds, then turn it ON again.Â* And then the front power
button will attempt to work again. Those don't seem to trigger
all that often. A "famous" one used to trip, when a certain
motherboard model was combined with a certain Antec with
"slow" 12V power rail. But other than that, the motherboard
is just as likely to "char" all around the socket, as it is
to properly shut down on a VCore short.

Don't forget to inspect for bad caps. Asus is not known for
epidemic proportions on those, but it still happens in a
small percentage of cases. As a counterpoint, there was one
Dell machine, where the motherboard was virtually guaranteed
to fail (getting close to 100% failure rate after a number of
years). And for a situation like that, it's not even safe
to get a used motherboard off Ebay (at least, unless the seller
claims it's been re-capped, which would add $50 or more
to the price in effect). Use the "normal level of suspicion"
about electrolytic caps (the ones with the "X" or "K" cut in the
top). The polymer caps with the "solid dome" and no X or K,
don't have quite the same failure modes.

*******

You can look through the posts in the VIP forum for evidence
of a "themed failure".

http://vip.asus.com/forum/topic.aspx...nguag e=en-us


Â*Â*Â* Paul


Thanks for the ideas, Paul.Â* I've ordered a new PS from Amazon --
nothing too expensive, not a PC Power and Cooling model, but something
more suited for the MBO than what I have in it now.Â* The P5Q Pro Turbo
has an 8-pin 12v socket which is only half used by the current PS,
which has only a 4-pin 12v plug.Â* But it worked when I first tried it
in 2013 and it's worked ever since, so I haven't worried about the
empty spots on the socket.Â* But I figured if I were going to get a new
PS I might as well get 8 pins.

It was the way it was behaving this morning that persuaded me the case
power button wasn't the culprit.Â* No amount of tapping around the
button can set off the problem.Â* In fact, I have no idea how to
duplicate the problem.Â* But I suppose if I install a new PS and the
problem recurs, I'll be looking at that button next.Â* And this not
being the good old traditional days, there is no "reset" button on the
case.

In fact, the computer has been running basically nonstop all day today
and after the initial misbehavior this morning, I've had no power
problems whatsoever.

I did manage to get the fresh new Win10 installation backed up today
using my Acronis boot disk.Â* There were no problems at all with that
today.Â* If it was the contents of the HDD I was backing up yesterday
that was causing the power to die during a backup, well, clearly I
know a lot less about how all this works than I thought I did.Â* Why
would Acronis True Image care what 0s and 1s it was backing up?Â* Makes
no sense to me still, but the backup ran fine this afternoon so I
guess I should just be happy.

I don't see any swollen caps, and I couldn't find anything in BIOS
regarding restarts after power failure.Â* So I'm setting it all aside
now that it's backed up, and I'll wait for Amazon Prime to deliver the
new PS on Monday.Â* I'll come back here to report the results.Â* Many
thanks as always.


To do a test of a PSU, you can get yourself a "PS_ON adapter",
which shorts #PS_ON to GND and that causes the PSU and fan to
run. Normally, I connect some electrical loads while testing.
This is for standalone PSU testing, without any electrical
connection to a motherboard.

If it ran for a couple hours, without shutting off, I might
conclude it's OK.

I use cement resistors for loading the PSU.

https://www.amazon.com/Uxcell-a11101.../dp/B01149WHOM


They consist of a ceramic shell, and the resistor element inside
is covered in some sort of fusible material. I think if it gets
hot enough, it might melt. I usually point a fan at them
to reduce their surface temperature.

You have to work out the resistance value you want, before shopping,
and decide how many resistors to use, how many watts to dump and so on.
And that's a basic way of loading down a PSU.

For example, a 5V rail, with a 2 ohm resistor, is 5/2 = 2.5amps
5V*2.5A = 12.5W. If I get four 2 ohm 10W resistors and do a
series parallel network, I can handle the 12.5W over four
resistors with ease. That's the kind of simple Ohms law
arithmetic involved.

Â*Â*Â*Â*Â*Â* +--- 2 ----+Â* +--- 2 ----+Â*Â*Â*Â*Â* A 2 ohm 40W resistor made
Â*Â*Â* ---+Â*Â*Â*Â*Â*Â*Â*Â*Â* +--+Â*Â*Â*Â*Â*Â*Â*Â*Â* +---Â*Â* out of four 2 ohm 10W resistors.
Â*Â*Â*Â*Â*Â* +--- 2 ----+Â* +--- 2 ----+

I could also do the same thing with 1/4W carbon composition resistors,
but I'd need boxes of them and a large roll of solder. It would
look pretty though.

And if you like to recycle or use old heatsinks from your
collection, this format of load resistor can be used, with a
little thermal paste on the bottom to sink into some old
aluminum heatsink you've got.

https://www.mouser.com/search/refine...&Ntt=177068861

On modern dual forward conversion supplies ("80+ type"), you might
want to load a minimum of two rails. When converters have multiple
outputs, you can save a little money on cement resistors by not
loading everything. I think I have at least one resistor for each
rail on mine.

Â*Â* ------- main -----+------ +12VÂ* === load here
Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â* |
Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â* +--- 3.3V/5V converter --- 5VÂ*Â*Â* === load a rail
here
Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â* Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â*Â* --- 3.3V

Just to give some ideas how the newer 80+ generation
might be treated.

Rails like -12V are weak, so I don't put a huge load on them.
I think I used a 50 ohm cement for that one, which is 0.25A
and close to the 0.3A limit printed on the label on the side
of the supply. On modern motherboards, the actual load is
the loading of a single TI 75232 RS232 chip for a serial port.
So probably 0.05A or less.

The time constant of your on-off behavior, doesn't
sound like thermal cutoff on the PSU, but what do I know.
Usually a PSU gets blazing hot if there's trouble, and it
takes a bit of time before it cools enough for more "antics".
Regular cycling in the 1-2 second range, suggests a control
problem somewhere. Not a thermal limiter.

Â*Â* Paul


I may or may not have solved the problem with the new PS, but there have
been no failures since I installed it on Saturday. I will continue to
keep fingers crossed, but apparently it's fixed. We shall see...

Thanks for the help, Paul!

--
Bill Anderson

I am the Mighty Favog