Freeeze during boot? Need Help

wrote:
On Mon, 30 Mar 2009 15:06:27 -0400, Paul wrote:


The system has three hard drives, one SATA, two IDE, plus a SATA DVD
burner.
It is a AMD Phenom 9550 Quad-Core 2.6GHz.
ASUS M3A78-CM MOBO.
3.25GB (Supposed to be 4GB) DDR2 RAM.
400W PSU.
XP SP2.
Three hard drives @ 12W each. 36W
DVD burner 12V @ 1.5A, 5V @ 1.5A, (12V drawn when media inserted.) 25W max
Phenom 9550 95W/0.90 = 105.6W at ATX12V 2x2. ( http://products.amd.com/en-us/Deskto...il.aspx?id=400 )
Motherboard plus RAM - allocate 50W
10W for 5VSB and USB loads.
6W for fans
-----------
36+25+105.6+50+10+6=232.6W (and my estimates are still higher than the real system).

So 400W should cover it.

(For RAM, you can get power numbers from Kingston. This 2x2GB DDR2-800 kit, draws 2W per module.)
http://www.valueram.com/datasheets/KHX6400D2K2_4G.pdf

I allocate 50W for motherboard and RAM, to cover a 20W+ Northbridge, a
5-10W Southbridge, and the small amount of power that DDR2 or DDR3
DIMMs now draw.

When a system is sitting idle, it draws much less power than that
estimate. That estimate is for a gaming situation. And with built-in
graphics, you're unlikely to max the power consumption. Maybe Prime95
would be the best loading situation you could achieve on that system.

Power supply analysis, should be done on a per-rail basis. All
the numbers on the power supply label, have a meaning, and should
be checked. But at a first order approximation, the 400W should
be enough.

When the computer starts, the BIOS will be single threaded, and
only one of the four cores will be doing anything. The hard drives
draw more current for the first ten seconds (spinup), but the processor
draws very little power, and the effects cancel out. I do a
detailed calculation of startup power, if there are four or more
hard drives present. If you get enough hard drives in a system,
and they each draw 2.5A during the ten second spinup, eventually
that becomes significant.

Paul

Thank you Paul.

I forgot to include that I am using an ATI Radeon HD3650 512MB PCI
Express X16 video card. That would add to the 232.6W for sure.
The company (Portatech) that sold me this thing is telling me that the
PSU is marginal for what I have, and that I should go for a stronger
PSU. Of course they will be glad to sell me the same.

Your figures tell me that will do no good. Portatech also told me to
restore the defaults to the BIOS to see what happens in the morning
when I next cold start.

I fear the problem is with either the AMD Phenom 9550 or the ASUS
M3A78-CM MOBO. I thought I went for a good combo - but let's face it,
one of two 2GB RAMS was DOA, and the first PSU burned up after two
weeks of use. Doesn't leave me very optimistic about the rest of the
hardwares.

Duke

If the motherboard has integrated video, you can use that instead of the 3650.
But the HD 3650 doesn't draw that much power.

An HD3650 here, is 12W at idle, and 41.7W max during 3D gaming.

http://www.xbitlabs.com/articles/vid...0_4.html#sect0

There is no Auxiliary power connector on the card, and video card designers
seem to limit the amps from the slot, to a little over 4 amps. So 12V @ 4.3A
would be about as much power as they might choose to use (52W from the 12V rail).
They don't generally attempt to go all the way to the max of 75W combined
which can be drawn from 12V and 3.3V. At least, I haven't seen a measured
value go over about 4.3A on the 12V rail (looking at a lot of cards from
Xbitlabs measurements). If they needed more power, then for safety, they'd
include a 2x3 power connector on the card. The 41.7W number seems to be
consistent, from that point of view.

http://www.xbitlabs.com/images/video...b_front_bg.jpg

What is printed on the label of the PSU ? Is there a picture on
the web ? That would save copying the numbers.

HTH,
Paul

On Mar 30, 12:05*pm, wrote:
Looks like 400W is sufficient.

Finish the math. Some power supplies rated by responsible computer
manufacturers at 350 watts are marketed by others to computer
assemblers as 500 watts. Nobody lied. But supplies marketed to
computer assemblers may play specmanship games.

Watts are irrelevant - but simple enough to market to computer
assemblers. Critical are amps for each voltage. A supply of
sufficient watts may also not provide sufficient current on one
voltage. Furthermore, no manufacturer really provides useful numbers
(for watts or current). You have seen ballpark numbers. Only way to
really know is to confirm those specs with a multimeter while supply
loaded - connected to computer.

How to determine if each voltage outputs sufficient current:
solution posted above. Appreciate your problem. An undersized power
supply (insufficient current on one voltage) will still boot a
computer. That may be your original supply. Only a multimeter will
identify an undersized voltage. An undersized power supply that
booted a computer everytime six months ago then may create strange
freezes today. Only the meter would have identified that defect six
months ago.

Whereas ballpark numbers will define what is sufficient, still,
supply specmanship games can leave one voltage undersized. After a
computer boots, best is to obtain voltages as listed above. Of
course, long before replacing a supply, better is to learn if the
original supply was defective - undersized or just failing. Same two
minute method answers both questions. It is that simple. Watt
numbers actually reports little that is accurate due to games of
specmanship. Those watts calculations only suggest what should be
(may be) sufficient.

On Mon, 30 Mar 2009 15:06:27 -0400, Paul wrote:


The system has three hard drives, one SATA, two IDE, plus a SATA DVD
burner.
It is a AMD Phenom 9550 Quad-Core 2.6GHz.
ASUS M3A78-CM MOBO.
3.25GB (Supposed to be 4GB) DDR2 RAM.
400W PSU.
XP SP2.

Three hard drives @ 12W each. 36W
DVD burner 12V @ 1.5A, 5V @ 1.5A, (12V drawn when media inserted.) 25W max
Phenom 9550 95W/0.90 = 105.6W at ATX12V 2x2. ( http://products.amd.com/en-us/Deskto...il.aspx?id=400 )
Motherboard plus RAM - allocate 50W
10W for 5VSB and USB loads.
6W for fans
-----------
36+25+105.6+50+10+6=232.6W (and my estimates are still higher than the real system).

So 400W should cover it.

(For RAM, you can get power numbers from Kingston. This 2x2GB DDR2-800 kit, draws 2W per module.)
http://www.valueram.com/datasheets/KHX6400D2K2_4G.pdf

I allocate 50W for motherboard and RAM, to cover a 20W+ Northbridge, a
5-10W Southbridge, and the small amount of power that DDR2 or DDR3
DIMMs now draw.

When a system is sitting idle, it draws much less power than that
estimate. That estimate is for a gaming situation. And with built-in
graphics, you're unlikely to max the power consumption. Maybe Prime95
would be the best loading situation you could achieve on that system.

Power supply analysis, should be done on a per-rail basis. All
the numbers on the power supply label, have a meaning, and should
be checked. But at a first order approximation, the 400W should
be enough.

When the computer starts, the BIOS will be single threaded, and
only one of the four cores will be doing anything. The hard drives
draw more current for the first ten seconds (spinup), but the processor
draws very little power, and the effects cancel out. I do a
detailed calculation of startup power, if there are four or more
hard drives present. If you get enough hard drives in a system,
and they each draw 2.5A during the ten second spinup, eventually
that becomes significant.

Paul

Thank you Paul.

I forgot to include that I am using an ATI Radeon HD3650 512MB PCI
Express X16 video card. That would add to the 232.6W for sure.
The company (Portatech) that sold me this thing is telling me that the
PSU is marginal for what I have, and that I should go for a stronger
PSU. Of course they will be glad to sell me the same.

Your figures tell me that will do no good. Portatech also told me to
restore the defaults to the BIOS to see what happens in the morning
when I next cold start.

I fear the problem is with either the AMD Phenom 9550 or the ASUS
M3A78-CM MOBO. I thought I went for a good combo - but let's face it,
one of two 2GB RAMS was DOA, and the first PSU burned up after two
weeks of use. Doesn't leave me very optimistic about the rest of the
hardwares.

Duke

On Sat, 28 Mar 2009 11:05:46 -0500, wrote:

On 28 Mar 2009 13:52:56 GMT, Andy wrote:


Can never be 100% sure at this distance from the machine in question.
It's feasable that a fault in the PSU could be causing the problem -
nothing to do with the 400W rating though.

Well said Andy.

The PSU is brand new, and is an identical replacement for a
predecessor that burnt out after a month's usage. Doesn't say much
for the vendor's product, does it? Maybe he is selling defective
stuff?

Duke

Why don't you tell us what make and model PSU it is?

If it is junk, you need to return it for a refund or credit
towards a better PSU and not run the system until then, if
possible, so you don't risk damaging other parts.

On Sun, 29 Mar 2009 16:45:52 -0700 (PDT), westom
wrote:

On Mar 29, 2:35*pm, wrote:
Now when I cold start the thing in the AM, it doesn't get to the BIOS
at all. *It sits there with CPU fan spinning and screen blank. *If I
power off and then on again, it completes the boot fine. *

After all that work, at this point you must have a list of what you
know is good. I suspect you do not because so many recommended by
swapping parts rather than getting numbers that say something
definitively - without doubt.

Until you have established power supply 'system' integrity, then
everything will appear defective. Worse, a defective power supply
system can still boot a computer. The only way to identify a
defective power supply system is with a multimeter while that system
is under heavy load. The meter is even sold where hammers are sold
for about the same price which means you can have that meter and
useful information quickly.


Agreed, although it is possible that a PSU could be of
insufficient capacity to the point that it could be
maintaining voltage under load, but the (transformer
particularly) PSU is still undersized to the extent that
it's rise-time to reach regulation is too slow when cold.
Often to compensate for this, they sabotage the PSU's
feedback for regulation so it doesn't shut off when
undervoltage at the right level making the safety in use
even worse.

Unfortunately such a problem is hard to detect with a
multimeter, logging equipment with a scope is the best way
to see such a problem.

I'm not yet sure it is the PSU but we are only seeing
generic info on these PSU, it would not be surprising for a
badly designed PSU to fail, then be replaced with same bad
design and still have problems, or for the first bad PSU to
have damaged a powered component in the system, and grabbing
an old PSU from the closet may or may not prove anything, if
that old PSU is not in good operating condition, is
similarly overrated, or is an old enough design that even if
it's wattage is high enough it lacks sufficient 12V current
or 5VSB current.

westom wrote:
wrote:

Looks like 400W is sufficient.

Finish the math. Some power supplies rated by responsible
computer manufacturers at 350 watts are marketed by others to
computer assemblers as 500 watts. Nobody lied. But supplies
marketed to computer assemblers may play specmanship games.

Nonsense, as far as I am concerned. A supply rated at N watts
should be able to supply those N watts continuously without
significantly affecting its life. Operated in a typical
environment.

--
[mail]: Chuck F (cbfalconer at maineline dot net)
[page]: http://cbfalconer.home.att.net
Try the download section.

On Mon, 30 Mar 2009 17:28:34 -0400, kony wrote:


Why don't you tell us what make and model PSU it is?

If it is junk, you need to return it for a refund or credit
towards a better PSU and not run the system until then, if
possible, so you don't risk damaging other parts.


OKIA 450W ATX

Yes- I am thinking of trying to upgrade.

Duke

On Mar 30, 5:46*pm, CBFalconer wrote:
Nonsense, as far as I am concerned. *A supply rated at N watts
should be able to supply those N watts continuously without
significantly affecting its life. *

Why discuss ‘life’? Short all outputs from a power supply together
- maximum load - and power supply life expectancy is not reduced.
Power supply must not even be harmed. A standard even 35 years ago.
Even Intel specs defined how big that shorting wire must be because
every power supply is never harmed by the load. Not even harmed when
all outputs are shorted together.

A supply rated by a responsible manufacturer at 350 watts is also
rated by others at 500 watts. Neither is lying. Simply measure
wattage at different points. Obvious among those who did engineering
in a power supply manufacturing house (a little hint).

Furthermore, a supply can have more than sufficient watts - total.
But wattage from one voltage is too small. If the load demands too
much current from the 12 volts, then a supply of sufficient watts
still does not supply sufficient current on that one voltage. System
then freezes sometimes during a boot.

Those who measure power supplies to see failures before replacing
anything have learned this. Many others just know because they
swapped supplies and it worked. The former know from experience. The
latter know only from speculation and shotgunning.

Just because a supply consumes 500 watts does not means it outputs
500 watts. And does not mean is outputs sufficient wattage (current)
on each voltage. But in games of specmanship, computer assemblers who
often don't learn this stuff then insist that a 600 or 700 watt supply
is needed. They cure symptoms without first learning the problem.

Meanwhile, the OP is still doing the 'try this and try that' method
- also called shotgunning. For all we know, the power supply 'system'
is defective. Therefore fixing a BIOS might even complicate the
problem. Always necessary is to build a list of what is 'definitively
known good' and 'definitively known bad' before replacing anything.

On Mar 30, 5:35*pm, kony wrote:
Unfortunately such a problem is hard to detect with a
multimeter, logging equipment with a scope is the best way
to see such a problem.

Most power supply problems were easily identified by voltages – only
when under load. Minimum numbers are intentionally different from
spec numbers due to how voltages are measured and other problems (such
as your examples). If temperature is one reason for failure, well,
that may even be identified only when those numbers are posted here.
Those numbers may contain additional information even if above minimal
values. But only those with better knowledge might see it. How to
take advantage of the few who know more about this? The OP should
provide those numbers and other critical information such as power
supply model number.

Yes, the oscilloscope and logging equipment is how it is done in
professional locations. But the layman use tools that any layman has
and can use. That is the 3.5 digit multimeter even selling in Wal-
Mart for less than $18. Nothing else (including those shotgunning
suggestions) can provide better information.

OP did not list his power supply as if he knows what his help needs
to know. Your demands for information were also appropriate.
Another ballpark indicator of a minimally sufficient supply is a long
list of written manufacturer specs. Power supplies are complex
beasts. Some power supplies missing required functions to sell for a
lower price at a higher profit. Maybe only 1% know what those specs
mean. To keep the informed silent and to sell to assemblers without
electrical knowledge, inferior supplies do not provide those specs.

Of course, Kony knows much of this. However others (maybe a
majority) do not which is the reason for above details.

The OP is still shotgunning when he does not even know if a supply
'system' (more than just a power supply) is functional. Best way for
a layman to determine supply 'system' integrity starts by spending
less than $18 in Wal-Mart for a multimeter. Or in any store where
hammers are sold. Those numbers discover so much that still remains
unknown.

westom wrote:
CBFalconer wrote:

Nonsense, as far as I am concerned. A supply rated at N watts
should be able to supply those N watts continuously without
significantly affecting its life.

Why discuss ‘life’? Short all outputs from a power supply
together - maximum load - and power supply life expectancy is
not reduced. Power supply must not even be harmed. A standard
even 35 years ago. Even Intel specs defined how big that
shorting wire must be because every power supply is never
harmed by the load. Not even harmed when all outputs are
shorted together.

A supply in current limiting has effectively shut itself down to
protect itself, and it is NOT supplying the rated items. Modern
supplies are largely single switcher supplies, with low output
impedances matching the various lines, so load can be moved from
line to line. The magic number is the total output, in watts.

--
[mail]: Chuck F (cbfalconer at maineline dot net)
[page]: http://cbfalconer.home.att.net
Try the download section.