failures on boot

John B. Smith wrote:
On Mon, 25 Dec 2017 10:32:48 -0500, John B. Smith
wrote:

On Sun, 17 Dec 2017 16:15:17 -0600, VanguardLH wrote:

My computer has booted without failure for a week. Thinking about all
your replies something occurred to me: some time back I turned off the
power supply with the switch at the back to do something or other.
When I turned it back on it didn't work. Eventually it did, but since
that time I've been unplugging the supply every time I wanted to mess
around inside the case, then plugging back in when I wanted to turn
on. So I'm thinking maybe the flaky on switch caused the supply to
drop out in mid-boot, then come immediately back on. Without an
orderly shutdown maybe the machine would fall back to the C1 BIOS
display which occurs at turn-on but was prevented from starting up
again by the state of things.

So I started thinking about a new power supply. This one interests me
https://www.newegg.com/Product/Produ...82E16817703038

My old power supply is this one:
https://www.newegg.com/Product/Produ...82E16817703005
I have it mounted in the TOP of my Gigabyte case
https://www.newegg.com/Product/Produ...82E16811233013

My motherboard is this one:
https://www.newegg.com/Product/Produ...82E16813127050

Here is one of the questions a NewEgg poster posed about the newer
supply I'm considering:

"Just bought one of these for a computer I'm putting together (first
one). However, upon opening the box, I discovered that the 20pin ATX
connector seems to be missing the third pin. Not only that, but that
whole wire is missing. Is this how it's supposed to be? Looks sort of
like that in the picture on Newegg, but it's really hard to tell."

The answer was:
" That is the -5v line. It was used in very old machines (ancient by
aging standards)
Manufacturers leave it out because there is no reason for the
additional cost since people aren't building computers with 15 year
old hardware."

Well, I built my baby in 2008 so it's not quite 15 years old BUT do I
need that -5 volt line? And does anybody see anything else that would
prevent the power supply I'm considering from working in my set up?


Wups, I see a compatibility problem. The new ps I picked out has its
fan on top. Since my current supply has a fan in back and is located
in the top of my case, the new one, if it would even fit up there,
would have the fan pressing right against the top of the case.
Further, ALL the new supplies seem to have the same problem. The only
solution I see is to try and cut a hole for the fan in the top of my
case. And I ain't got that big a can opener.

You've compared the hole pattern on the supply,
to your mounting plate ? Maybe the picture is not
right-side up ?

I think what's happened to the PCPower one, is the
product label is on upside-down, for those situations
where the supply is put in the bottom of the computer
case. For an inverted layout PC, the supply looks "normal".

When used in a legacy case, the label will be upside-down,
and the fan will be facing downwards (like it should).

Inspect the hole pattern. My two machines here,
there's a screw just below the AC plug, and because
that screw isn't on a rectangular layout, I'm using
that as my "key" to figuring it out.

Looking at the Newegg pictures more closely, it looks
like there are product labels on two sides, and they're
not oriented the same way either.

*******

And yes, there definitely have been ATX supplies with
the fan on the *wrong* side. You did a good job to check
this, because it has happened. I've had reports before,
of people who mounted the thing that way (mounted according
to hole pattern), and then posted as to "how is this
supposed to work exactly" :-) An eyeball check, before
buying, is a good idea.

Just a guess,
Paul

John B. Smith wrote:

On Sun, 17 Dec 2017 16:15:17 -0600, VanguardLH wrote:

My computer has booted without failure for a week. Thinking about all
your replies something occurred to me: some time back I turned off the
power supply with the switch at the back to do something or other.
When I turned it back on it didn't work. Eventually it did, but since
that time I've been unplugging the supply every time I wanted to mess
around inside the case, then plugging back in when I wanted to turn
on. So I'm thinking maybe the flaky on switch caused the supply to
drop out in mid-boot, then come immediately back on. Without an
orderly shutdown maybe the machine would fall back to the C1 BIOS
display which occurs at turn-on but was prevented from starting up
again by the state of things.

So I started thinking about a new power supply. This one interests me
https://www.newegg.com/Product/Produ...82E16817703038

My old power supply is this one:
https://www.newegg.com/Product/Produ...82E16817703005
I have it mounted in the TOP of my Gigabyte case
https://www.newegg.com/Product/Produ...82E16811233013

My motherboard is this one:
https://www.newegg.com/Product/Produ...82E16813127050

Here is one of the questions a NewEgg poster posed about the newer
supply I'm considering:

"Just bought one of these for a computer I'm putting together (first
one). However, upon opening the box, I discovered that the 20pin ATX
connector seems to be missing the third pin. Not only that, but that
whole wire is missing. Is this how it's supposed to be? Looks sort of
like that in the picture on Newegg, but it's really hard to tell."

The answer was:
" That is the -5v line. It was used in very old machines (ancient by
aging standards)
Manufacturers leave it out because there is no reason for the
additional cost since people aren't building computers with 15 year
old hardware."

Well, I built my baby in 2008 so it's not quite 15 years old BUT do I
need that -5 volt line? And does anybody see anything else that would
prevent the power supply I'm considering from working in my set up?

I've gotten where I prefer PSUs with modular wiring. That is, instead
of having to deal with one monstrous tanglement of cabling from the PSU,
the PSU has individual connectors for cables. I can add just as many
cables as are currently needed for the components inside. I don't have
to deal with tying off and making sure unused connectors don't short,
get into fans, or the unused cabling getting in the way of air flow. If
I add something later, I can add another cable.

I cannot speak to the reliability and quality of the PC Power PSU.
Never used them. I'd have to check who actually manufacturers the PSU.
I've tended to stay with Fortron (FSP = Fortron Source Power) or Sparkle
(same company, different price points, FSP/Fortron is USA, Sparkle is
Taiwan, but Sparkle was first and Fortron was later). It's has been a
long time since I last checked but back then FSP did their own
manufacturing and QC (i.e., they didn't contract out to slap their
sticker on someone else's product). FSP has several brands under which
they sell PSUs and several other brands contract their PSUs from FSP, so
there a Sparkle, Fortron (changed to FSP), EVGA, Antec, OCZ,
Silverstone, Thermaltake, Nexus, and Zalman (maybe more). However, you
can never be sure with rebranding until you look inside since purchasing
agents can change their minds of where to get parts.

The PSU (its load capacity with low ripple) are the lifeblood of your
computer so I never skimp there. The PSU put into a salvaged PC used at
home has a Corsair unit. Corsair doesn't make anything. They spec out
what they want a manufacturer to build. When I last looked, they used
CWT (Channel Wall Technology, Taiwan, perhaps the largest PSU maker) to
build thier PSUs; however, like any manufacturer, they build to the spec
of their customer. The same manufacturer could produce crappy units
because that's what the customer ordered and paid for. Like FSP, I've
found Corsair to be accurate on their specs. Alas, it didn't have
modular wiring but I used an unused 5.25" drive bay to shove in the
extra cables. That Corsair PSU has been working great since 2009. If
you can stand not having modular cabling, Newegg has a sale (8 hours
left) at http://tinyurl.com/y7vfwnrx for $70 with a $20 rebate card
(prepaid card you can use anywhere). Can't say how good is Corsair's
tech support since I've never had to use them for anything labelled
Corsair.

Have you used a wattmeter (plugs into an outlet, plug the PC in the
wattmeter) to see how much load your current setup consumes? The *new*
and claimed efficiency rating was 83%, so divide the wattmeter's rating
by 75% (it's old now) to see how much load the PSU is currently putting
out. Then remember that you might want to put in a better and more
power hungry video card, or a second one, more HDDs, more memory, etc.
You sure 750W will do for the next 10 years?

Regarding orientation of the PSU, just flip it over to match its
mounting position. Typically the bottom-mounted PSU cases have a square
opening in the backpanel. However, many top-mounted PSU cases have a
larger cutout for the fan position and a smaller cutout where is the
switch(es) and the cord enters. Sometimes that reduced cutout can come
very close to the on-off switch. If flipping over the PSU so it fits in
a top-mounted case and if the reduced cutout is in the way of the
switch, you could use a nibbler to chew away to make the opening larger.
Just be sure the nibbling tool can handle the gauge of case sheeting.
The typical ones handle 18 gauge sheets. Some are rated only for soft
steel and aluminum while others just say "steel". Cheap ones are $20,
good ones around $50. Your old Gigabyte case looks like it probably
won't need any nibbling.


But before you go to a new PSU ...

Did you yet replace the CMOS coin cell battery (discussed in the other
subthread) on the mobo to make sure that wasn't the cause of your flaky
boot problem? If the battery dates back to 2008, it is WAY too old.

On Mon, 25 Dec 2017 15:06:07 -0500, Paul
wrote:

John B. Smith wrote:
On Mon, 25 Dec 2017 10:32:48 -0500, John B. Smith
wrote:

On Sun, 17 Dec 2017 16:15:17 -0600, VanguardLH wrote:

My computer has booted without failure for a week. Thinking about all
your replies something occurred to me: some time back I turned off the
power supply with the switch at the back to do something or other.
When I turned it back on it didn't work. Eventually it did, but since
that time I've been unplugging the supply every time I wanted to mess
around inside the case, then plugging back in when I wanted to turn
on. So I'm thinking maybe the flaky on switch caused the supply to
drop out in mid-boot, then come immediately back on. Without an
orderly shutdown maybe the machine would fall back to the C1 BIOS
display which occurs at turn-on but was prevented from starting up
again by the state of things.

So I started thinking about a new power supply. This one interests me
https://www.newegg.com/Product/Produ...82E16817703038

My old power supply is this one:
https://www.newegg.com/Product/Produ...82E16817703005
I have it mounted in the TOP of my Gigabyte case
https://www.newegg.com/Product/Produ...82E16811233013

My motherboard is this one:
https://www.newegg.com/Product/Produ...82E16813127050

Here is one of the questions a NewEgg poster posed about the newer
supply I'm considering:

"Just bought one of these for a computer I'm putting together (first
one). However, upon opening the box, I discovered that the 20pin ATX
connector seems to be missing the third pin. Not only that, but that
whole wire is missing. Is this how it's supposed to be? Looks sort of
like that in the picture on Newegg, but it's really hard to tell."

The answer was:
" That is the -5v line. It was used in very old machines (ancient by
aging standards)
Manufacturers leave it out because there is no reason for the
additional cost since people aren't building computers with 15 year
old hardware."

Well, I built my baby in 2008 so it's not quite 15 years old BUT do I
need that -5 volt line? And does anybody see anything else that would
prevent the power supply I'm considering from working in my set up?


Wups, I see a compatibility problem. The new ps I picked out has its
fan on top. Since my current supply has a fan in back and is located
in the top of my case, the new one, if it would even fit up there,
would have the fan pressing right against the top of the case.
Further, ALL the new supplies seem to have the same problem. The only
solution I see is to try and cut a hole for the fan in the top of my
case. And I ain't got that big a can opener.

You've compared the hole pattern on the supply,
to your mounting plate ? Maybe the picture is not
right-side up ?

I think what's happened to the PCPower one, is the
product label is on upside-down, for those situations
where the supply is put in the bottom of the computer
case. For an inverted layout PC, the supply looks "normal".

When used in a legacy case, the label will be upside-down,
and the fan will be facing downwards (like it should).

Actually, I was concerned about the ps fan blowing air INTO my case
area. My old one has the fan facing 'outside' and I always thought it
was blowing OUT. However, putting a slip of paper in front of the 80mm
fan I'm unable to tell which direction the air is blowing. The supply
sits on a little shelf up there, which cannot be adjusted for up/down.
Measures like I got about 4" of height to play with, maybe a bit
under.

I'm relieved to have you tell me it's ok to face the fan downward. I
guess I'm not the only one who's confused about which way the ps fan
blows.
http://www.tomshardware.com/answers/...-air-fans.html

Inspect the hole pattern. My two machines here,
there's a screw just below the AC plug, and because
that screw isn't on a rectangular layout, I'm using
that as my "key" to figuring it out.

Looking at the Newegg pictures more closely, it looks
like there are product labels on two sides, and they're
not oriented the same way either.

*******

And yes, there definitely have been ATX supplies with
the fan on the *wrong* side. You did a good job to check
this, because it has happened. I've had reports before,
of people who mounted the thing that way (mounted according
to hole pattern), and then posted as to "how is this
supposed to work exactly" :-) An eyeball check, before
buying, is a good idea.

Just a guess,
Paul

On Mon, 25 Dec 2017 17:40:22 -0600, VanguardLH wrote:

John B. Smith wrote:

On Sun, 17 Dec 2017 16:15:17 -0600, VanguardLH wrote:

My computer has booted without failure for a week. Thinking about all
your replies something occurred to me: some time back I turned off the
power supply with the switch at the back to do something or other.
When I turned it back on it didn't work. Eventually it did, but since
that time I've been unplugging the supply every time I wanted to mess
around inside the case, then plugging back in when I wanted to turn
on. So I'm thinking maybe the flaky on switch caused the supply to
drop out in mid-boot, then come immediately back on. Without an
orderly shutdown maybe the machine would fall back to the C1 BIOS
display which occurs at turn-on but was prevented from starting up
again by the state of things.

So I started thinking about a new power supply. This one interests me
https://www.newegg.com/Product/Produ...82E16817703038

My old power supply is this one:
https://www.newegg.com/Product/Produ...82E16817703005
I have it mounted in the TOP of my Gigabyte case
https://www.newegg.com/Product/Produ...82E16811233013

My motherboard is this one:
https://www.newegg.com/Product/Produ...82E16813127050

Here is one of the questions a NewEgg poster posed about the newer
supply I'm considering:

"Just bought one of these for a computer I'm putting together (first
one). However, upon opening the box, I discovered that the 20pin ATX
connector seems to be missing the third pin. Not only that, but that
whole wire is missing. Is this how it's supposed to be? Looks sort of
like that in the picture on Newegg, but it's really hard to tell."

The answer was:
" That is the -5v line. It was used in very old machines (ancient by
aging standards)
Manufacturers leave it out because there is no reason for the
additional cost since people aren't building computers with 15 year
old hardware."

Well, I built my baby in 2008 so it's not quite 15 years old BUT do I
need that -5 volt line? And does anybody see anything else that would
prevent the power supply I'm considering from working in my set up?

I've gotten where I prefer PSUs with modular wiring. That is, instead
of having to deal with one monstrous tanglement of cabling from the PSU,
the PSU has individual connectors for cables. I can add just as many
cables as are currently needed for the components inside. I don't have
to deal with tying off and making sure unused connectors don't short,
get into fans, or the unused cabling getting in the way of air flow. If
I add something later, I can add another cable.

I cannot speak to the reliability and quality of the PC Power PSU.
Never used them. I'd have to check who actually manufacturers the PSU.
I've tended to stay with Fortron (FSP = Fortron Source Power) or Sparkle
(same company, different price points, FSP/Fortron is USA, Sparkle is
Taiwan, but Sparkle was first and Fortron was later). It's has been a
long time since I last checked but back then FSP did their own
manufacturing and QC (i.e., they didn't contract out to slap their
sticker on someone else's product). FSP has several brands under which
they sell PSUs and several other brands contract their PSUs from FSP, so
there a Sparkle, Fortron (changed to FSP), EVGA, Antec, OCZ,
Silverstone, Thermaltake, Nexus, and Zalman (maybe more). However, you
can never be sure with rebranding until you look inside since purchasing
agents can change their minds of where to get parts.

The PSU (its load capacity with low ripple) are the lifeblood of your
computer so I never skimp there. The PSU put into a salvaged PC used at
home has a Corsair unit. Corsair doesn't make anything. They spec out
what they want a manufacturer to build. When I last looked, they used
CWT (Channel Wall Technology, Taiwan, perhaps the largest PSU maker) to
build thier PSUs; however, like any manufacturer, they build to the spec
of their customer. The same manufacturer could produce crappy units
because that's what the customer ordered and paid for. Like FSP, I've
found Corsair to be accurate on their specs. Alas, it didn't have
modular wiring but I used an unused 5.25" drive bay to shove in the
extra cables. That Corsair PSU has been working great since 2009. If
you can stand not having modular cabling, Newegg has a sale (8 hours
left) at http://tinyurl.com/y7vfwnrx for $70 with a $20 rebate card
(prepaid card you can use anywhere). Can't say how good is Corsair's
tech support since I've never had to use them for anything labelled
Corsair.

Have you used a wattmeter (plugs into an outlet, plug the PC in the
wattmeter) to see how much load your current setup consumes? The *new*
and claimed efficiency rating was 83%, so divide the wattmeter's rating
by 75% (it's old now) to see how much load the PSU is currently putting
out. Then remember that you might want to put in a better and more
power hungry video card, or a second one, more HDDs, more memory, etc.
You sure 750W will do for the next 10 years?

Alas my voltmeter is not fancy enough to measure wattage. It IS a
Fluke though.

Regarding orientation of the PSU, just flip it over to match its
mounting position. Typically the bottom-mounted PSU cases have a square
opening in the backpanel. However, many top-mounted PSU cases have a
larger cutout for the fan position and a smaller cutout where is the
switch(es) and the cord enters. Sometimes that reduced cutout can come
very close to the on-off switch. If flipping over the PSU so it fits in
a top-mounted case and if the reduced cutout is in the way of the
switch, you could use a nibbler to chew away to make the opening larger.
Just be sure the nibbling tool can handle the gauge of case sheeting.
The typical ones handle 18 gauge sheets. Some are rated only for soft
steel and aluminum while others just say "steel". Cheap ones are $20,
good ones around $50. Your old Gigabyte case looks like it probably
won't need any nibbling.

Yes my cutout is one large hole. You guys have made me sensitive to
mounting holes though and I believe my current supply is mounted
'upside down' (lettering) 2 screws on right corner from rear. and the
new supply, mounted fan down will be in that configuration also.


But before you go to a new PSU ...

Did you yet replace the CMOS coin cell battery (discussed in the other
subthread) on the mobo to make sure that wasn't the cause of your flaky
boot problem? If the battery dates back to 2008, it is WAY too old.

CMOS battery replaced 8-12-17. I didn't like the way it snapped in but
I THINK that it is not the problem. My understanding is the battery is
powering the BIOS and that does always run thru its C1 to FF routine
ok. It's later that the trouble occurs. Of course maybe I'm over
simplifying.

I'd kind of forgotten how much trouble it was to manage all those
cables when I put it together. I"m definitely going to have to have it
on a work table to swap in another ps. Shopping for a sturdy card
table too.

John B. Smith wrote:

Alas my voltmeter is not fancy enough to measure wattage. It IS a
Fluke though.

Was talking about a plug-in wattmeter, like http://tinyurl.com/y8rzwth4.
That one has just the standard NEMA-9 USA plug. There are other units
that have US, UK, and AU plugs, like http://tinyurl.com/y9bmz56w
(bewa that one ships from China and customs processing could result
in taking 45+ days to deliver; however, some Chinese sellers have
country-specific warehouses to provide immediate stock).

I have an ancient Ammeter Probe that can measure the current into an
outlet but it's a pain to wire up and insert inline. That's only good
for something like a max of 2 amperes. For high amps, I have to use an
inline adapter that uses the magnetic field of AC through separated
wires to induce a reading into the probe. A wattmeter is easier to use,
especially since you'll be comparing watts consumed to the wattage
rating of the PSU rather than calculating it from current and voltage
peak and RMS conversion.

CMOS battery replaced 8-12-17. I didn't like the way it snapped in but
I THINK that it is not the problem. My understanding is the battery is
powering the BIOS and that does always run thru its C1 to FF routine
ok. It's later that the trouble occurs. Of course maybe I'm over
simplifying.

When putting in a new CMOS battery, you want to wipe it clean of any
body oils, like fingerprints. That can caused oxidation (yep, even on
the polished steel used for the coin cell batteries) that increases
resistance of the connection. I wash my hands, rinse well to eliminate
any soap, use isopropyl alcohol to clean the coin cell battery, and
handle only by its sides when inserting (since the contacts are on the
flat bottom and flat top). If possible, I also rotate the battery a few
times to scrape the contacts.

The BIOS is burned into EEPROM(s). PROMs can be burned only once.
EPROMs can be erased (by removing a cover to expose to UV light) and
written anew. EEPROMS use a pin on the chip for using elevated voltage
to erase and then write anew. Doesn't matter if there is power or not
to an [E[E]]PROM: it retains its code in either case. The battery is
required for the CMOS table (in the RTC aka Real Time Clock chip or
circuitry). When the PC is powered down, voltage has to be maintained
on the CMOS logic to ensure it retains its state. If power is lost,
state is lost. If the CMOS table is cleared (using a couple header pins
on the mobo), electrically erased (a BIOS-initiated option), or
corrupted (its hash of settings doesn't match up later), or when you
load a preset of BIOS settings, the CMOS table is erased and settings
are copied from the [semi-]permanent BIOS. The settings in the CMOS
table are used. If corrupt or cleared, the settings from the BIOS must
get copied into the CMOS table again.

When there is live power to the PSU from the outlet through the cord,
the ATX PSU supplies +5V standby power used for the soft-power logic on
the motherboard. The logic requires power. When you turn off your
computer, the PSU is still supplying the +5VSB power. That will also
power the CMOS to have it retain its settings. Only when you pull the
cord from the PSU or turn off the outlet to which it is plugged with
there be no +5VSB downed to 3V to the CMOS logic. That's when you need
the battery to keep the CMOS settings retained. If the outlet has power
to which the PSU is plugged, the battery isn't used. If the outlet goes
dead or you pull the plug, the battery gets used.

Back in the days of AT PSUs, the Power switch was a mechanical switch
that disconnected the PSU from line power. Unless the outlet was always
live (guaranteed only with a huge UPS), and since users often power off
their PCs rather than leave them run 24x7, the CMOS battery was more
used. I had PCs with bad batteries that couldn't be replaced (unless I
did some soldering) with AT PSUs that would lose the RTC clock value
when powered off for more than a few minutes. Cure was to never power
off those PCs. ATX PSUs fixed that problem by having the PSU supply
+5VSB even when "off" as long as the outlet was live. Most users rarely
unplug their ATX PSU'ed PCs and the CMOS batteries only got drained
(other than self-drain) when there was a power outage. The assumption
is no one is such an idiot that they plug their PC into a wall outlet
controlled by a wall switch that also turns off the room lights but that
could happen which means the CMOS battery has to power the CMOS.

I'd kind of forgotten how much trouble it was to manage all those
cables when I put it together. I"m definitely going to have to have it
on a work table to swap in another ps. Shopping for a sturdy card
table too.

Make sure it is a hard surfaced folding work table. I have one with a
soft top (actually padded) and it is a pain to use as a work surface.
Card tables are sometimes soft[er], too. I keep watching for sales on
folding hardtop work tables but haven't gotten one yet. My buddy uses
them all the time when we work on building another garage, changing his
house, or some other construction project.

The problem is they have resin tops which is non-conductive. You'll
want an anti-static mat (one for the PC, another for putting parts) with
wires that go to the metal legs which should run the ground of a 3-wire
extension cord to a properly grounded outlet. Well, that's the rhetoric
of the anti-static paranoid and the proper warning.

ESD is a problem only when there is a *difference* in voltaic potential.
If the voltaic potential is the same between two object, them touching
results in no transfer of electrons (the columb charge is the same for
both). Zaps occur before of a difference in potential. If you're at
100K volts and so is the other object, there's no differential. As long
as the computer remains plugged into a properly grounded outlet, just
touch the chassis first to equalize yourself to it. You'll probably
already do that while taking off the side panel. Just don't go walking
around scraping your feet on carpet to then first touch something
inside. Touch the chassis first. You could even use an anti-static
wrist strap attached to the chassis. To eliminate the +5VSB inside the
case (on the mobo), you could also get a substitute power cord for the
PSU (Goodwill often has loads of them for super cheap). Slice the
outside insulation (doesn't matter if you hit the wires inside), snip
the hot and neutral wires and cut away sections of the hot and neutral
wires (at different lengths and then pull on their insulation to have it
extend farther than the wire inside to keep the hot and neutral from
shorting), tape the wire ends, and tape up the cord. Then you'll only
have the ground wire connecting the PSU to the grounded outlet, and
there will be no +5VSB inside.

On 12/26/2017 12:19 PM, VanguardLH wrote:
John B. Smith wrote:

Alas my voltmeter is not fancy enough to measure wattage. It IS a
Fluke though.

Was talking about a plug-in wattmeter, like http://tinyurl.com/y8rzwth4.
That one has just the standard NEMA-9 USA plug. There are other units
that have US, UK, and AU plugs, like http://tinyurl.com/y9bmz56w
(bewa that one ships from China and customs processing could result
in taking 45+ days to deliver; however, some Chinese sellers have
country-specific warehouses to provide immediate stock).

I have an ancient Ammeter Probe that can measure the current into an
outlet but it's a pain to wire up and insert inline. That's only good
for something like a max of 2 amperes. For high amps, I have to use an
inline adapter that uses the magnetic field of AC through separated
wires to induce a reading into the probe. A wattmeter is easier to use,
especially since you'll be comparing watts consumed to the wattage
rating of the PSU rather than calculating it from current and voltage
peak and RMS conversion.

CMOS battery replaced 8-12-17. I didn't like the way it snapped in but
I THINK that it is not the problem. My understanding is the battery is
powering the BIOS and that does always run thru its C1 to FF routine
ok. It's later that the trouble occurs. Of course maybe I'm over
simplifying.

When putting in a new CMOS battery, you want to wipe it clean of any
body oils, like fingerprints. That can caused oxidation (yep, even on
the polished steel used for the coin cell batteries) that increases
resistance of the connection. I wash my hands, rinse well to eliminate
any soap, use isopropyl alcohol to clean the coin cell battery, and
handle only by its sides when inserting (since the contacts are on the
flat bottom and flat top). If possible, I also rotate the battery a few
times to scrape the contacts.

The BIOS is burned into EEPROM(s). PROMs can be burned only once.
EPROMs can be erased (by removing a cover to expose to UV light) and
written anew. EEPROMS use a pin on the chip for using elevated voltage
to erase and then write anew. Doesn't matter if there is power or not
to an [E[E]]PROM: it retains its code in either case. The battery is
required for the CMOS table (in the RTC aka Real Time Clock chip or
circuitry). When the PC is powered down, voltage has to be maintained
on the CMOS logic to ensure it retains its state. If power is lost,
state is lost. If the CMOS table is cleared (using a couple header pins
on the mobo), electrically erased (a BIOS-initiated option), or
corrupted (its hash of settings doesn't match up later), or when you
load a preset of BIOS settings, the CMOS table is erased and settings
are copied from the [semi-]permanent BIOS. The settings in the CMOS
table are used. If corrupt or cleared, the settings from the BIOS must
get copied into the CMOS table again.

When there is live power to the PSU from the outlet through the cord,
the ATX PSU supplies +5V standby power used for the soft-power logic on
the motherboard. The logic requires power. When you turn off your
computer, the PSU is still supplying the +5VSB power. That will also
power the CMOS to have it retain its settings. Only when you pull the
cord from the PSU or turn off the outlet to which it is plugged with
there be no +5VSB downed to 3V to the CMOS logic. That's when you need
the battery to keep the CMOS settings retained. If the outlet has power
to which the PSU is plugged, the battery isn't used. If the outlet goes
dead or you pull the plug, the battery gets used.

Back in the days of AT PSUs, the Power switch was a mechanical switch
that disconnected the PSU from line power. Unless the outlet was always
live (guaranteed only with a huge UPS), and since users often power off
their PCs rather than leave them run 24x7, the CMOS battery was more
used. I had PCs with bad batteries that couldn't be replaced (unless I
did some soldering) with AT PSUs that would lose the RTC clock value
when powered off for more than a few minutes. Cure was to never power
off those PCs. ATX PSUs fixed that problem by having the PSU supply
+5VSB even when "off" as long as the outlet was live. Most users rarely
unplug their ATX PSU'ed PCs and the CMOS batteries only got drained
(other than self-drain) when there was a power outage. The assumption
is no one is such an idiot that they plug their PC into a wall outlet
controlled by a wall switch that also turns off the room lights but that
could happen which means the CMOS battery has to power the CMOS.

I'd kind of forgotten how much trouble it was to manage all those
cables when I put it together. I"m definitely going to have to have it
on a work table to swap in another ps. Shopping for a sturdy card
table too.

Make sure it is a hard surfaced folding work table. I have one with a
soft top (actually padded) and it is a pain to use as a work surface.
Card tables are sometimes soft[er], too. I keep watching for sales on
folding hardtop work tables but haven't gotten one yet. My buddy uses
them all the time when we work on building another garage, changing his
house, or some other construction project.

The problem is they have resin tops which is non-conductive. You'll
want an anti-static mat (one for the PC, another for putting parts) with
wires that go to the metal legs which should run the ground of a 3-wire
extension cord to a properly grounded outlet. Well, that's the rhetoric
of the anti-static paranoid and the proper warning.

ESD is a problem only when there is a *difference* in voltaic potential.
If the voltaic potential is the same between two object, them touching
results in no transfer of electrons (the columb charge is the same for
both). Zaps occur before of a difference in potential. If you're at
100K volts and so is the other object, there's no differential. As long
as the computer remains plugged into a properly grounded outlet, just
touch the chassis first to equalize yourself to it. You'll probably
already do that while taking off the side panel. Just don't go walking
around scraping your feet on carpet to then first touch something
inside. Touch the chassis first. You could even use an anti-static
wrist strap attached to the chassis. To eliminate the +5VSB inside the
case (on the mobo), you could also get a substitute power cord for the
PSU (Goodwill often has loads of them for super cheap). Slice the
outside insulation (doesn't matter if you hit the wires inside), snip
the hot and neutral wires and cut away sections of the hot and neutral
wires (at different lengths and then pull on their insulation to have it
extend farther than the wire inside to keep the hot and neutral from
shorting), tape the wire ends, and tape up the cord. Then you'll only
have the ground wire connecting the PSU to the grounded outlet, and
there will be no +5VSB inside.



A Kill-A-Watt meter Is really a must have when dealing with power
supplies(and many other power questions.

Rene

John B. Smith wrote:


Alas my voltmeter is not fancy enough to measure wattage. It IS a
Fluke though.

I recommend one of these. It measures watts (the "bill-able" quantity),
as well as measuring power factor on complex waveforms. It works
as well as a digital power meter on the outside of your house.
These devices use sigma-delta converters running at high speed,
to digitize the volts and amps readings for further processing.

http://www.p3international.com/products/p4400.html

The only weakness it has, is the shunt. To measure
current, it uses a shunt, and you really shouldn't
leave loads like electric kettles connected to the
thing, as it makes the shunt pretty hot. This is similar
to what happens to multimeters, where you insist on
measuring close to the 10 ampere limit. Some Kill-a-watts
have been ruined by a steady diet of near-the-limit
loads.

For "limitless" current measurement, I use a clamp-on
DC ammeter for that (the DC ones support AC and DC).
For example, it told me how many amps from 220V my
central air was using. And when my car wouldn't start
one day, it told me the peak current draw from the
car battery, was 150 amps DC (starter problem). And it
never gets even warm, because it measures magnetic field
rather than current flow. And reads out the value in amps.
It uses a Hall probe (the "Right Hand Rule" baked into
silicon). The lowest digit on mine is valued at 10mA,
so it's not all that sensitive. But it can measure up
to 400 amps. The car hasn't gotten there yet :-) The
40A DC range is good for work inside a PC, as a lot
of the colored wires in the main power cable can
be measured on that range. You put all the red wires
inside the jaws, to make a "total" current flow
measurement, and the magnetic fields around the
wires add together.

If you worked with mini-ITX builds, the combination of
those two meters, makes it easier to select a PICO
converter for an ITX project. As otherwise, there might
not be enough info to select a PICO at the same time as
you order a motherboard. You run the motherboard off
a conventional ATX supply first, to figure out what
size PICO fits best.

Paul

On Tue, 26 Dec 2017 18:04:10 -0500, Paul
wrote:

John B. Smith wrote:


Alas my voltmeter is not fancy enough to measure wattage. It IS a
Fluke though.

I recommend one of these. It measures watts (the "bill-able" quantity),
as well as measuring power factor on complex waveforms. It works
as well as a digital power meter on the outside of your house.
These devices use sigma-delta converters running at high speed,
to digitize the volts and amps readings for further processing.

http://www.p3international.com/products/p4400.html

I suppose I GOT to get one of those if they're that cheap. I don't
picture myself using it much though.

The only weakness it has, is the shunt. To measure
current, it uses a shunt, and you really shouldn't
leave loads like electric kettles connected to the
thing, as it makes the shunt pretty hot. This is similar
to what happens to multimeters, where you insist on
measuring close to the 10 ampere limit. Some Kill-a-watts
have been ruined by a steady diet of near-the-limit
loads.

For "limitless" current measurement, I use a clamp-on
DC ammeter for that (the DC ones support AC and DC).


Think I priced one of those once and got scared off. Again, I wouldn't
use it enough to pay for it.


For example, it told me how many amps from 220V my
central air was using. And when my car wouldn't start
one day, it told me the peak current draw from the
car battery, was 150 amps DC (starter problem). And it
never gets even warm, because it measures magnetic field
rather than current flow. And reads out the value in amps.
It uses a Hall probe (the "Right Hand Rule" baked into
silicon). The lowest digit on mine is valued at 10mA,
so it's not all that sensitive. But it can measure up
to 400 amps. The car hasn't gotten there yet :-) The
40A DC range is good for work inside a PC, as a lot
of the colored wires in the main power cable can
be measured on that range. You put all the red wires
inside the jaws, to make a "total" current flow
measurement, and the magnetic fields around the
wires add together.

If you worked with mini-ITX builds, the combination of
those two meters, makes it easier to select a PICO
converter for an ITX project. As otherwise, there might
not be enough info to select a PICO at the same time as
you order a motherboard. You run the motherboard off
a conventional ATX supply first, to figure out what
size PICO fits best.
I had to look up PICO. And a Windows-running computer that small is
kinda fascinating. And for my purposes useless I think.

Our local cable company recently digitized and encrypted our signal,
bless their conniving little hearts. No way can I get my one remaining
programmable vcr to work off it. And the tv at work no longer works.
(thassa pun). However with one of those micro-baby computers I suppose
I could hook the tv up to it thru an HDMI port and enjoy a movie at
lunch? Without headphone and sub-titles I'd rather read a book.



Paul

On Tue, 26 Dec 2017 12:19:29 -0600, VanguardLH wrote:

John B. Smith wrote:

Alas my voltmeter is not fancy enough to measure wattage. It IS a
Fluke though.

Was talking about a plug-in wattmeter, like http://tinyurl.com/y8rzwth4.
That one has just the standard NEMA-9 USA plug. There are other units
that have US, UK, and AU plugs, like http://tinyurl.com/y9bmz56w
(bewa that one ships from China and customs processing could result
in taking 45+ days to deliver; however, some Chinese sellers have
country-specific warehouses to provide immediate stock).

I have an ancient Ammeter Probe that can measure the current into an
outlet but it's a pain to wire up and insert inline. That's only good
for something like a max of 2 amperes. For high amps, I have to use an
inline adapter that uses the magnetic field of AC through separated
wires to induce a reading into the probe. A wattmeter is easier to use,
especially since you'll be comparing watts consumed to the wattage
rating of the PSU rather than calculating it from current and voltage
peak and RMS conversion.

CMOS battery replaced 8-12-17. I didn't like the way it snapped in but
I THINK that it is not the problem. My understanding is the battery is
powering the BIOS and that does always run thru its C1 to FF routine
ok. It's later that the trouble occurs. Of course maybe I'm over
simplifying.

When putting in a new CMOS battery, you want to wipe it clean of any
body oils, like fingerprints. That can caused oxidation (yep, even on
the polished steel used for the coin cell batteries) that increases
resistance of the connection. I wash my hands, rinse well to eliminate
any soap, use isopropyl alcohol to clean the coin cell battery, and
handle only by its sides when inserting (since the contacts are on the
flat bottom and flat top). If possible, I also rotate the battery a few
times to scrape the contacts.

The BIOS is burned into EEPROM(s). PROMs can be burned only once.
EPROMs can be erased (by removing a cover to expose to UV light) and
written anew. EEPROMS use a pin on the chip for using elevated voltage
to erase and then write anew. Doesn't matter if there is power or not
to an [E[E]]PROM: it retains its code in either case. The battery is
required for the CMOS table (in the RTC aka Real Time Clock chip or
circuitry). When the PC is powered down, voltage has to be maintained
on the CMOS logic to ensure it retains its state. If power is lost,
state is lost. If the CMOS table is cleared (using a couple header pins
on the mobo), electrically erased (a BIOS-initiated option), or
corrupted (its hash of settings doesn't match up later), or when you
load a preset of BIOS settings, the CMOS table is erased and settings
are copied from the [semi-]permanent BIOS. The settings in the CMOS
table are used. If corrupt or cleared, the settings from the BIOS must
get copied into the CMOS table again.

When there is live power to the PSU from the outlet through the cord,
the ATX PSU supplies +5V standby power used for the soft-power logic on
the motherboard. The logic requires power. When you turn off your
computer, the PSU is still supplying the +5VSB power. That will also
power the CMOS to have it retain its settings. Only when you pull the
cord from the PSU or turn off the outlet to which it is plugged with
there be no +5VSB downed to 3V to the CMOS logic.

Does't one other condition apply? If you switch the on/off switch on
the PSU to off, wouldn't you lose that +5VSB? The reason I don't power
down and try it myself is that switch is the main reason I'm thinking
of replacing my PSU. I think I explained in my first post that the
last time I switched the PSU off it was very cranky about getting
switched back on. So I suspected that flaky switch of causing a
momentary power drop during the boot. (though it never seems to bother
once Windows is up and running, so maybe that's not a great diagnosis)
I've been unplugging ac from the PSU if I wanted to mess around
inside. Thereby removing my ground cable to the pc at the same time.
Guess I better replace that telephone wire I used to have grounding
the chassis to the outlet plug screw that got torn off on my move. Or
butcher up another AC cable to provide a ground as you explained.

That's when you need
the battery to keep the CMOS settings retained. If the outlet has power
to which the PSU is plugged, the battery isn't used. If the outlet goes
dead or you pull the plug, the battery gets used.

Back in the days of AT PSUs, the Power switch was a mechanical switch
that disconnected the PSU from line power. Unless the outlet was always
live (guaranteed only with a huge UPS), and since users often power off
their PCs rather than leave them run 24x7, the CMOS battery was more
used. I had PCs with bad batteries that couldn't be replaced (unless I
did some soldering) with AT PSUs that would lose the RTC clock value
when powered off for more than a few minutes. Cure was to never power
off those PCs. ATX PSUs fixed that problem by having the PSU supply
+5VSB even when "off" as long as the outlet was live. Most users rarely
unplug their ATX PSU'ed PCs and the CMOS batteries only got drained
(other than self-drain) when there was a power outage. The assumption
is no one is such an idiot that they plug their PC into a wall outlet
controlled by a wall switch that also turns off the room lights but that
could happen which means the CMOS battery has to power the CMOS.

I'd kind of forgotten how much trouble it was to manage all those
cables when I put it together. I"m definitely going to have to have it
on a work table to swap in another ps. Shopping for a sturdy card
table too.

Make sure it is a hard surfaced folding work table. I have one with a
soft top (actually padded) and it is a pain to use as a work surface.
Card tables are sometimes soft[er], too. I keep watching for sales on
folding hardtop work tables but haven't gotten one yet. My buddy uses
them all the time when we work on building another garage, changing his
house, or some other construction project.

The problem is they have resin tops which is non-conductive. You'll
want an anti-static mat (one for the PC, another for putting parts) with
wires that go to the metal legs which should run the ground of a 3-wire
extension cord to a properly grounded outlet. Well, that's the rhetoric
of the anti-static paranoid and the proper warning.

ESD is a problem only when there is a *difference* in voltaic potential.
If the voltaic potential is the same between two object, them touching
results in no transfer of electrons (the columb charge is the same for
both). Zaps occur before of a difference in potential. If you're at
100K volts and so is the other object, there's no differential. As long
as the computer remains plugged into a properly grounded outlet, just
touch the chassis first to equalize yourself to it. You'll probably
already do that while taking off the side panel. Just don't go walking
around scraping your feet on carpet to then first touch something
inside. Touch the chassis first. You could even use an anti-static
wrist strap attached to the chassis. To eliminate the +5VSB inside the
case (on the mobo), you could also get a substitute power cord for the
PSU (Goodwill often has loads of them for super cheap). Slice the
outside insulation (doesn't matter if you hit the wires inside), snip
the hot and neutral wires and cut away sections of the hot and neutral
wires (at different lengths and then pull on their insulation to have it
extend farther than the wire inside to keep the hot and neutral from
shorting), tape the wire ends, and tape up the cord. Then you'll only
have the ground wire connecting the PSU to the grounded outlet, and
there will be no +5VSB inside.

John B. Smith wrote:

I had to look up PICO. And a Windows-running computer that small is
kinda fascinating. And for my purposes useless I think.

Our local cable company recently digitized and encrypted our signal,
bless their conniving little hearts. No way can I get my one remaining
programmable vcr to work off it. And the tv at work no longer works.
(thassa pun). However with one of those micro-baby computers I suppose
I could hook the tv up to it thru an HDMI port and enjoy a movie at
lunch? Without headphone and sub-titles I'd rather read a book.

You can get a TV tuner with CableCard hole.

https://www.newegg.com/Product/Produ...82E16815345006

That connects via Ethernet to your LAN. It's just the tuner
part. But, via the CableCard, it decrypts at least the regular digital
channels. The CableCard itself, the cable company may charge $5 a month
for the card. I don't know the details of that.

https://www.silicondust.com/product/hdhomerun-prime/

"Please note: This product HDHR3-CC will only work in the US."

Why exactly ? CableCard restriction ? I think a few other
countries use the same QAM, so that's not it. Maybe
they're referring to their "DVR service" ?

There is a "Country Selector" at the top of this page, and
indeed, according to the chart, we can't use this in Canada
with a CableCard. Weird. I guess our electrons go counter-clockwise
up here or something. You would think the usage of a CableCard
would not require an Internet connection (i.e. CableCard plugged
into TV set), and that would make geolocation a bit difficult.
It implies the cable company signal itself is encoded to enforce
this. I wonder what scheme Canada uses to encrypt digital cable ?

https://www.silicondust.com/hdhomerun/

I didn't realize the technology was so fragmented. So I guess it
takes an STB and an IR Blaster ? And who knows what else to capture
the signal ? Will the signal have Macrovision or Do Not Copy ?
What a zoo. An endless barrel of trouble.

http://www.digitalhome.ca/forum/37-r...cablecard.html

Paul