post supply scare - all's well that starts up again!

[Lady Margaret Thatcher]

Here is a problem that solved itself. I thought I would share it with
the group because it is a "good news story," as a contrast to all the
tales of woe that fill these groups.

I was building up a new system using an ASUS board. Antec power
supply. I've had good luck with both of them in the past, so I stay
loyal to the brands.

So I'm putting together this new system, and I use my "diagnostics"
card that gives me a readout of the POST messages. The diagnostics
card has an extension cable with a second display, in case you can't
read the display off the card because it's buried inside the system.

Last night, I booted up the system for the first time. Looks good.
Then I tried to read that extension cable display. But instead, I
yanked a bit too hard and the diagnostics card popped out of its slot.
It is a small card and doesn't have a bracket mount.

Suddenly the system shut down. And wouldn't restart. Even pressing
the front power button didn't work. Switched the power supply off and
then on again didn't help. Unplugging the power supply didn't help.
The system was dead, dead, dead. Not just the motherboard, but even
the case fans didn't spin up.

Geez. What now. Nothing on the Antec web site that could help me.
Maybe 15 minutes goes by. So I try again, and the case fans spin for
about 10 seconds, and then shut down again. Tried again. Same
result.

Ok. I wait about 30 minutes, anxiously. I was so nervous you could
think that I was waiting for a child to be born.

This time, it all works. Case fans, power supply fans, motherboard
green light, and of course the CPU heat sink fan. And the system
boots up perfectly, into POST because there is no hard disk in the
system yet. I have no problem starting up the system BIOS, where I
check the power supply and fan/temp settings.

And three days later, it's still running fine. Motherboard temp
stable. CPU temp up only 2 degrees F from when the system first
started up. Voltages were very close to nominal at boot up (less than
1% off), and no change after three days.

So, all's well that starts up again.

--maggie-

 

[abc]

Here is a problem that solved itself. I thought I would share it with
the group because it is a "good news story," as a contrast to all the
tales of woe that fill these groups.

I was building up a new system using an ASUS board. Antec power
supply. I've had good luck with both of them in the past, so I stay
loyal to the brands.

So I'm putting together this new system, and I use my "diagnostics"
card that gives me a readout of the POST messages. The diagnostics
card has an extension cable with a second display, in case you can't
read the display off the card because it's buried inside the system.

Last night, I booted up the system for the first time. Looks good.
Then I tried to read that extension cable display. But instead, I
yanked a bit too hard and the diagnostics card popped out of its slot.
It is a small card and doesn't have a bracket mount.

Suddenly the system shut down. And wouldn't restart. Even pressing
the front power button didn't work. Switched the power supply off and
then on again didn't help. Unplugging the power supply didn't help.
The system was dead, dead, dead. Not just the motherboard, but even
the case fans didn't spin up.

Geez. What now. Nothing on the Antec web site that could help me.
Maybe 15 minutes goes by. So I try again, and the case fans spin for
about 10 seconds, and then shut down again. Tried again. Same
result.

Ok. I wait about 30 minutes, anxiously. I was so nervous you could
think that I was waiting for a child to be born.

This time, it all works. Case fans, power supply fans, motherboard
green light, and of course the CPU heat sink fan. And the system
boots up perfectly, into POST because there is no hard disk in the
system yet. I have no problem starting up the system BIOS, where I
check the power supply and fan/temp settings.

And three days later, it's still running fine. Motherboard temp
stable. CPU temp up only 2 degrees F from when the system first
started up. Voltages were very close to nominal at boot up (less than
1% off), and no change after three days.

So, all's well that starts up again.

--maggie-

I had something similar happen to me with an Abit board.
I was plugging in some HDDs, the PC wasn't running but was on at the wall
(no on/off switch on the generic PSU). I think I was trying to diagnose HDD
problems at the time.

I think it might be some sort of protection built into the mobo or PSUs. I
didn't short anything (wasn't using tools and I don't wear rings), so the
only thing I can think of is that I may have touched the internals
momentarily, (I'm talking milliseconds), before I earthed myself on the
frame , and some static charge caused this to happen, Even then I just
touched either a SATA cable or HDD power lead.

I could be wrong but I believe most PSUs have a breaker inside them that
acts like a thermostat, and as the breaker closes again you get a small
amount of power-which is why the fans might spin, but nothing else. I'm sure
someone will correct me (even if I am right )

Cheers.

 

[Phil Weldon]

'abc' wrote, in part:
| I could be wrong but I believe most PSUs have a breaker inside them that
| acts like a thermostat, and as the breaker closes again you get a small
| amount of power-which is why the fans might spin, but nothing else. I'm
sure
| someone will correct me (even if I am right )
_____

ATX type switching power supplies have 'crowbar' overload protection. If a
short condition exixts, the power supply shuts down. This is by current
sensing, not heat. The supply will not restart until the short is removed.
Such supplies also have a more conventional overload protector that limits
the total amount of power drawn (a fuse or thermal breaker.)

Phil Weldon

..
..
..

 

[Roger Hamlett]

'abc' wrote, in part:
| I could be wrong but I believe most PSUs have a breaker inside them
that
| acts like a thermostat, and as the breaker closes again you get a
small
| amount of power-which is why the fans might spin, but nothing else.
I'm sure
| someone will correct me (even if I am right )
_____

ATX type switching power supplies have 'crowbar' overload protection.
If a short condition exixts, the power supply shuts down. This is by
current sensing, not heat. The supply will not restart until the short
is removed. Such supplies also have a more conventional overload
protector that limits the total amount of power drawn (a fuse or thermal
breaker.)

Phil Weldon

To meet the ATX specs, the supply should have crowbar _overvoltage_
protection, but not a crowbar on the overcurrent state. Normally if the
crowbar circuit fires, it will blow the input fuse, and will not softly
recover as described. The key to the fan spinning, is that ATX supplies
have two complete powr supplies in the one case. the first is the 'main'
supply, and this uses eletronic on/off control. The second is the
'standby' supply, and this powers the circuit that controls the main
supply. The actual decision whether to power up, when the external supply
is switched on, is done in software on the board itself (hence the BIOS
option on many boards for 'power fail restart'). It is fairly common for
the standby supply to start first, and the circuitry it supplies, to then
momentarily engage full power, while a decision is reached on whether the
system should restart. This gives a fractional 'glitch' on the fans etc.,
powered off this supply.

Best Wishes

 

[Phil Weldon]

'Roger Hamlett' wrote, in part:
| To meet the ATX specs, the supply should have crowbar _overvoltage_
| protection, but not a crowbar on the overcurrent state.
_____

#1. Not according to ATX 12V (March 2005):
[From the document at
http://formfactors.org/developer/specs/ATX12_BSDB_2_2_public_br2.pdf ]

3.4.2. Short-circuit Protection
An output short circuit is defined as any output impedance of less than 0.1
ohms. The
power supply shall shut down and latch off for shorting the +3.3 VDC, +5
VDC, or
+12 VDC rails to return or any other rail. The +12 V1DC and +12V2DC should
have
separate short circuit and overload protection. Shorts between main output
rails and +5
VSB shall not cause any damage to the power supply. The power supply shall
either shut
down and latch off or fold back for shorting the negative rails. +5 VSB must
be capable of
being shorted indefinitely, but when the short is removed, the power supply
shall recover
automatically or by cycling PS_ON#. The power supply shall be capable of
withstanding a
continuous short-circuit to the output without damage or overstress to the
unit (for
example, to components, PCB traces, connectors) under the input conditions
specified in
Section 3.1. The maximum short-circuit energy in any output shall not exceed
240 VA, per
IEC 60950 requirements.

3.4.3. No-load Operation
No damage or hazardous condition should occur with all the DC output
connectors
disconnected from the load. The power supply may latch into the shutdown
state.


3.4.4. Over-current Protection
Overload currents applied to each tested output rail will cause the output
to trip before
reaching or exceeding 240 VA. For testing purposes, the overload currents
should be
ramped at a minimum rate of 10 A/s starting from full load.
**
#2. The 'Power Good' signal is not a 'software' function; it is a hardware
function, and does not depend on the BIOS (depending on the BIOS would
defeat the purpose, as the BIOS is of no use without a CPU, and the whole
purose of the 'Power Good' signal is to prevent operation at incorrect
voltages.)

#3. The 'Standby Power Supply' is always on (as long as AC power is
applied.)

Phil Weldon

 

[Robert Hancock]

'Roger Hamlett' wrote, in part:
| To meet the ATX specs, the supply should have crowbar _overvoltage_
| protection, but not a crowbar on the overcurrent state.
_____

#1. Not according to ATX 12V (March 2005):

A crowbar circuit is not the same as the overcurrent shutdown on the
power supply. A crowbar circuit actually shorts out the power supply's
output in order to either force the overcurrent to trip or blow the
input fuse. This prevents excess voltage from damaging components if the
output voltage rises too high due to some malfunction.

 

[Phil Weldon]

'Robert Hancock' wrote:
| A crowbar circuit is not the same as the overcurrent shutdown on the
| power supply. A crowbar circuit actually shorts out the power supply's
| output in order to either force the overcurrent to trip or blow the
| input fuse. This prevents excess voltage from damaging components if the
| output voltage rises too high due to some malfunction.
_____

Your description is of older, linear power supplies, not of modern switching
power supplies such as ATX specification power supplies. The classical
crowbar function is not necessary or desirable for switching power supplies;
the function is provided by overvoltage sensors and short sensors (now
usually part of the controller chip) that remove the drive to the input
switching transistors, shutting down the power supply. There is no need to
force an AC supply side fuse to blow since the same results are much better
obtained by removing the drive. The fuses/breakers on on an ATX power
supply protect against excessive current draw on the line voltage side. On
the low voltage, high current side, sensors protect each of the rails, one
or more of which might go into a over current condition without causing
excessive AC current draw.

As an experiment you could use a bare ATX supply, shorting one or more of
the outbut rails. The supply will shut down, but no fuses will ever be
blow, nor any breakers tripped.

What you describe just doesn't happen for switching power supplies.

Phil Weldon

 

[Roger Hamlett]

'Robert Hancock' wrote:
| A crowbar circuit is not the same as the overcurrent shutdown on the
| power supply. A crowbar circuit actually shorts out the power supply's
| output in order to either force the overcurrent to trip or blow the
| input fuse. This prevents excess voltage from damaging components if
the
| output voltage rises too high due to some malfunction.
_____

Your description is of older, linear power supplies, not of modern
switching power supplies such as ATX specification power supplies. The
classical crowbar function is not necessary or desirable for switching
power supplies; the function is provided by overvoltage sensors and
short sensors (now usually part of the controller chip) that remove the
drive to the input switching transistors, shutting down the power
supply. There is no need to force an AC supply side fuse to blow since
the same results are much better obtained by removing the drive. The
fuses/breakers on on an ATX power supply protect against excessive
current draw on the line voltage side. On the low voltage, high current
side, sensors protect each of the rails, one or more of which might go
into a over current condition without causing excessive AC current draw.

As an experiment you could use a bare ATX supply, shorting one or more
of the outbut rails. The supply will shut down, but no fuses will ever
be blow, nor any breakers tripped.

What you describe just doesn't happen for switching power supplies.

It does.
What I was taking offence to, (if you read the original post) was the
suggestion that there would be a crowbar on an _overcurrent_ situation.
This as you point out, would be self aggravating, pointless, and even
potentially dangerous, being much simpler just to shut off the drive.
Crowbars are still used on switch mode supplies, being applied across the
capacitors on the incoming rail, as one of the few ways to get rid of the
excess energy stored here, in the event of a potentially dangerous output
situation, which has not been controlled by the normal regulation. These
are required to meet the specifications in certain countries, especially
for office/industrial applications, and are common on current generation
supplies.

Best Wishes

 

[Flash]

I'm sorry, but I wouldn't want to throw a crowbar into my PS. It might
let the smoke out!

Did that once with a 286 computer. The fuse blew and I replaced it
with a "slow blow" fuse. The mushroom cloud was impressive. Git 'er
done!!

Cheers...