APC question

I have an 18 Ah APC for my computer.

My computer,LCD monitor, and modem are plugged into it.

Battery indicator shows 29 mins of time left.

Does that sound reasonable?

I thought it would give much greater run time than with the original battery that it came with.

Thanks,
Andy

In message , Andy
writes:
I have an 18 Ah APC for my computer.

My computer,LCD monitor, and modem are plugged into it.

Battery indicator shows 29 mins of time left.

Does that sound reasonable?

Work it out! What does the "18 Ah" refer to: the capacity of a battery?
What voltage battery?

Assuming it's a 12V battery, 18 Ah @ 12V is 216 watt-hours; allowing for
conversion inefficiency and the fact that the unit's own protection
circuitry won't let the battery be run completely flat, less than 200
Watt-hours. What wattage do your computer, monitor, and modem consume?
Assuming by modem you mean your ADSL modem/router, I'd guess 12-15W for
that; monitor in the tens of watts - say 70-90; computer - well, since
you mention a monitor, I assume it's a "desktop" not a laptop, so
anything from say 150W to several times that. So 250 to 750 watts, being
supplied by something with 200 Wh capacity. So 29 minutes sounds highly
plausible. (I may be wildly out in my guesses for the consumption of the
three items, but you get the idea.)

I thought it would give much greater run time than with the original
battery that it came with.

What Ah was that? What run time did the system quote for that? And, is
your "17 Ah" battery fully charged?

Thanks,
Andy
--
J. P. Gilliver. UMRA: 1960/1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

I admire you British: when things get tough, you reach for humour. Not
firearms. - Sigourney (Susan) Weaver, RT 2017/11/4-10

Andy wrote:
I have an 18 Ah APC for my computer.

My computer,LCD monitor, and modem are plugged into it.

Battery indicator shows 29 mins of time left.

Does that sound reasonable?

I thought it would give much greater run time than with the original battery that it came with.

Thanks,
Andy

Mine has 2 lead acid gel batteries, 270 watts, 400 VA.
It is is good for for about 6 min.
The servers at work each have massive battery packs.
One server and small lcd is good for 10 min at most.

On Thursday, July 26, 2018 at 11:35:39 PM UTC-5, J. P. Gilliver (John) wrote:
In message , Andy
writes:
I have an 18 Ah APC for my computer.

My computer,LCD monitor, and modem are plugged into it.

Battery indicator shows 29 mins of time left.

Does that sound reasonable?

Work it out! What does the "18 Ah" refer to: the capacity of a battery?
What voltage battery?

Assuming it's a 12V battery, 18 Ah @ 12V is 216 watt-hours; allowing for
conversion inefficiency and the fact that the unit's own protection
circuitry won't let the battery be run completely flat, less than 200
Watt-hours. What wattage do your computer, monitor, and modem consume?
Assuming by modem you mean your ADSL modem/router, I'd guess 12-15W for
that; monitor in the tens of watts - say 70-90; computer - well, since
you mention a monitor, I assume it's a "desktop" not a laptop, so
anything from say 150W to several times that. So 250 to 750 watts, being
supplied by something with 200 Wh capacity. So 29 minutes sounds highly
plausible. (I may be wildly out in my guesses for the consumption of the
three items, but you get the idea.)

I thought it would give much greater run time than with the original
battery that it came with.

What Ah was that? What run time did the system quote for that? And, is
your "17 Ah" battery fully charged?

Thanks,
Andy
--
J. P. Gilliver. UMRA: 1960/1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

I admire you British: when things get tough, you reach for humour. Not
firearms. - Sigourney (Susan) Weaver, RT 2017/11/4-10

Battery is 12 V.

18 Ah is the capacity.

22 inch monitor

Of course the battery was fully charged. :-)

Andy

Andy wrote:
I have an 18 Ah APC for my computer.

My computer,LCD monitor, and modem are plugged into it.

Battery indicator shows 29 mins of time left.

Does that sound reasonable?

I thought it would give much greater run time than with the original battery that it came with.

Thanks,
Andy

Using my crystal ball on your equipment,
the calc goes like this. The only real known
in the equation, is that "1 hour has 60 minutes".

12V*18AH 60m
------------ * --- * eff = minutes of time
100W+30W+10W 1H

216WH 60m
----- * --- * eff = 93m * eff
140W 1H

If eff was 0.7, time is 65m. I have no idea what the
efficiency of the inverter on a UPS is. The 0.7 assumes
a conventional switcher with no tricks. It really
shouldn't be all that efficient, as it's switching
12V on the primary side, and that can't be all that good.
We cannot assume the UPS is "too inefficient", as that
would make the UPS heat up when the mains go off, and
I've never noticed a bad heating effect on mine when
the mains go off.

The battery AH rating goes down at high current
draw. Usually the product has a curve for derating
purposes.

The idle power of my best modern machine in the
house, a desktop, is 60W. My newest machine
is 100W to 110W at idle. I picked 100W for
my example calc. The oldest machine in the
house is 150W, and on that one, the power
doesn't vary at all under high CPU load.
The old machines are... just wasteful.
Low MIPS, high watts. If gaming, whatever
video card was in the old box, the power would
double on the card.

Your best bet is to find the curve for your product,
to take some of the unknowns out of the calc.
You can also use a Kill-A-Watt meter to determine
AC load of each appliance connected to the UPS.

https://www.homedepot.com/p/Kill-A-W...4400/202196386

Useful numbers from the Kill-A-Watt meter would be
"Watts" and "PF". And the documentation for the UPS
will have a different output limit for "Watts" versus
"VA" loading (the ratio of those is PF), which is making
a statement about the PF or Power Factor of the load(s).
Old ATX PSUs have a PF of 0.65 to 0.7 or so (phase relationship
of voltage to current on mains side). A modern ATX PSU
with active PFC (PF correction) runs 0.99 and has
the voltage and current are in-phase like on a
light bulb load. Your three electrical loads
all have switchers, so the PF should be
similar for them. Again, a rough wild-guess
pulled from thin air.

If you have:

1) A clamp-on DC ammeter (clamped on battery lead).
2) A multimeter to measure battery voltage.
3) A Kill-A-Watt to measure output load power.

you could get a value for eff yourself. Then
the only remaining part of the puzzle is
battery performance -- how the 18AH is
degraded to 9AH if you draw 40A out
of the battery during the event. A lighter
electrical load, gets closer to the battery
nominal AH rating during the discharge process.

Also, it isn't really all that good of an idea
to be running the battery pak flat for fun. That
reduces the service life of the battery. By pampering
mine, I got 10 years from it. Yes, the power
goes off here, but I try to do an orderly shutdown
in about 2 minutes if I can manage it. Only
a small percentage of the 18AH gets used when
you do it that way, and it doesn't take too much
time to restore the charge.

For car batteries, you only run them down 25% before you
stop using it. So if a car battery was 18AH, you can
only safely use 4.5AH (16 minutes worth) if the load
switches itself off at a safe point in time. There
are also deep discharge batteries for things
like golf carts or forklifts, which have thicker
plates, and more of their 18AH could be used
without causing immediate battery failure.
I don't know what that factor is for an SLA
battery in a UPS. I treated my UPS like it was
a car battery (calcium added to lead plates),
just to be on the safe side.

That will give you some idea how many variables
there are.

eff (switcher/inverter on output stage)
battery derating curve at load point (not efficient at high load)
percentage of battery used before UPS threshold trips out

actual load and PF of item(s) making up the load

Lots of variables.

Paul

On Friday, July 27, 2018 at 2:44:20 AM UTC-5, Paul wrote:
Andy wrote:
I have an 18 Ah APC for my computer.

My computer,LCD monitor, and modem are plugged into it.

Battery indicator shows 29 mins of time left.

Does that sound reasonable?

I thought it would give much greater run time than with the original battery that it came with.

Thanks,
Andy

Using my crystal ball on your equipment,
the calc goes like this. The only real known
in the equation, is that "1 hour has 60 minutes".

12V*18AH 60m
------------ * --- * eff = minutes of time
100W+30W+10W 1H

216WH 60m
----- * --- * eff = 93m * eff
140W 1H

If eff was 0.7, time is 65m. I have no idea what the
efficiency of the inverter on a UPS is. The 0.7 assumes
a conventional switcher with no tricks. It really
shouldn't be all that efficient, as it's switching
12V on the primary side, and that can't be all that good.
We cannot assume the UPS is "too inefficient", as that
would make the UPS heat up when the mains go off, and
I've never noticed a bad heating effect on mine when
the mains go off.

The battery AH rating goes down at high current
draw. Usually the product has a curve for derating
purposes.

The idle power of my best modern machine in the
house, a desktop, is 60W. My newest machine
is 100W to 110W at idle. I picked 100W for
my example calc. The oldest machine in the
house is 150W, and on that one, the power
doesn't vary at all under high CPU load.
The old machines are... just wasteful.
Low MIPS, high watts. If gaming, whatever
video card was in the old box, the power would
double on the card.

Your best bet is to find the curve for your product,
to take some of the unknowns out of the calc.
You can also use a Kill-A-Watt meter to determine
AC load of each appliance connected to the UPS.

https://www.homedepot.com/p/Kill-A-W...4400/202196386

Useful numbers from the Kill-A-Watt meter would be
"Watts" and "PF". And the documentation for the UPS
will have a different output limit for "Watts" versus
"VA" loading (the ratio of those is PF), which is making
a statement about the PF or Power Factor of the load(s).
Old ATX PSUs have a PF of 0.65 to 0.7 or so (phase relationship
of voltage to current on mains side). A modern ATX PSU
with active PFC (PF correction) runs 0.99 and has
the voltage and current are in-phase like on a
light bulb load. Your three electrical loads
all have switchers, so the PF should be
similar for them. Again, a rough wild-guess
pulled from thin air.

If you have:

1) A clamp-on DC ammeter (clamped on battery lead).
2) A multimeter to measure battery voltage.
3) A Kill-A-Watt to measure output load power.

you could get a value for eff yourself. Then
the only remaining part of the puzzle is
battery performance -- how the 18AH is
degraded to 9AH if you draw 40A out
of the battery during the event. A lighter
electrical load, gets closer to the battery
nominal AH rating during the discharge process.

Also, it isn't really all that good of an idea
to be running the battery pak flat for fun. That
reduces the service life of the battery. By pampering
mine, I got 10 years from it. Yes, the power
goes off here, but I try to do an orderly shutdown
in about 2 minutes if I can manage it. Only
a small percentage of the 18AH gets used when
you do it that way, and it doesn't take too much
time to restore the charge.

For car batteries, you only run them down 25% before you
stop using it. So if a car battery was 18AH, you can
only safely use 4.5AH (16 minutes worth) if the load
switches itself off at a safe point in time. There
are also deep discharge batteries for things
like golf carts or forklifts, which have thicker
plates, and more of their 18AH could be used
without causing immediate battery failure.
I don't know what that factor is for an SLA
battery in a UPS. I treated my UPS like it was
a car battery (calcium added to lead plates),
just to be on the safe side.

That will give you some idea how many variables
there are.

eff (switcher/inverter on output stage)
battery derating curve at load point (not efficient at high load)
percentage of battery used before UPS threshold trips out

actual load and PF of item(s) making up the load

Lots of variables.

Paul

Thanks.

I have a lot of info to go over.

It ran for 1 hr. and showed 17 mins left.

So I put it back on AC power to avoid data loss.

I am happy with that time. :-)

Andy

On Friday, July 27, 2018 at 2:44:20 AM UTC-5, Paul wrote:
Andy wrote:
I have an 18 Ah APC for my computer.

My computer,LCD monitor, and modem are plugged into it.

Battery indicator shows 29 mins of time left.

Does that sound reasonable?

I thought it would give much greater run time than with the original battery that it came with.

Thanks,
Andy

Using my crystal ball on your equipment,
the calc goes like this. The only real known
in the equation, is that "1 hour has 60 minutes".

12V*18AH 60m
------------ * --- * eff = minutes of time
100W+30W+10W 1H

216WH 60m
----- * --- * eff = 93m * eff
140W 1H

If eff was 0.7, time is 65m. I have no idea what the
efficiency of the inverter on a UPS is. The 0.7 assumes
a conventional switcher with no tricks. It really
shouldn't be all that efficient, as it's switching
12V on the primary side, and that can't be all that good.
We cannot assume the UPS is "too inefficient", as that
would make the UPS heat up when the mains go off, and
I've never noticed a bad heating effect on mine when
the mains go off.

The battery AH rating goes down at high current
draw. Usually the product has a curve for derating
purposes.

The idle power of my best modern machine in the
house, a desktop, is 60W. My newest machine
is 100W to 110W at idle. I picked 100W for
my example calc. The oldest machine in the
house is 150W, and on that one, the power
doesn't vary at all under high CPU load.
The old machines are... just wasteful.
Low MIPS, high watts. If gaming, whatever
video card was in the old box, the power would
double on the card.

Your best bet is to find the curve for your product,
to take some of the unknowns out of the calc.
You can also use a Kill-A-Watt meter to determine
AC load of each appliance connected to the UPS.

https://www.homedepot.com/p/Kill-A-W...4400/202196386

Useful numbers from the Kill-A-Watt meter would be
"Watts" and "PF". And the documentation for the UPS
will have a different output limit for "Watts" versus
"VA" loading (the ratio of those is PF), which is making
a statement about the PF or Power Factor of the load(s).
Old ATX PSUs have a PF of 0.65 to 0.7 or so (phase relationship
of voltage to current on mains side). A modern ATX PSU
with active PFC (PF correction) runs 0.99 and has
the voltage and current are in-phase like on a
light bulb load. Your three electrical loads
all have switchers, so the PF should be
similar for them. Again, a rough wild-guess
pulled from thin air.

If you have:

1) A clamp-on DC ammeter (clamped on battery lead).
2) A multimeter to measure battery voltage.
3) A Kill-A-Watt to measure output load power.

you could get a value for eff yourself. Then
the only remaining part of the puzzle is
battery performance -- how the 18AH is
degraded to 9AH if you draw 40A out
of the battery during the event. A lighter
electrical load, gets closer to the battery
nominal AH rating during the discharge process.

Also, it isn't really all that good of an idea
to be running the battery pak flat for fun. That
reduces the service life of the battery. By pampering
mine, I got 10 years from it. Yes, the power
goes off here, but I try to do an orderly shutdown
in about 2 minutes if I can manage it. Only
a small percentage of the 18AH gets used when
you do it that way, and it doesn't take too much
time to restore the charge.

For car batteries, you only run them down 25% before you
stop using it. So if a car battery was 18AH, you can
only safely use 4.5AH (16 minutes worth) if the load
switches itself off at a safe point in time. There
are also deep discharge batteries for things
like golf carts or forklifts, which have thicker
plates, and more of their 18AH could be used
without causing immediate battery failure.
I don't know what that factor is for an SLA
battery in a UPS. I treated my UPS like it was
a car battery (calcium added to lead plates),
just to be on the safe side.

That will give you some idea how many variables
there are.

eff (switcher/inverter on output stage)
battery derating curve at load point (not efficient at high load)
percentage of battery used before UPS threshold trips out

actual load and PF of item(s) making up the load

Lots of variables.

Paul

"Crystal ball" comments are unnecessary.

I provide what info I think is necessary.

If more is need, just ask away.

Andy

Andy wrote:
On Friday, July 27, 2018 at 2:44:20 AM UTC-5, Paul wrote:
Andy wrote:
I have an 18 Ah APC for my computer.

My computer,LCD monitor, and modem are plugged into it.

Battery indicator shows 29 mins of time left.

Does that sound reasonable?

I thought it would give much greater run time than with the original battery that it came with.

Thanks,
Andy
Using my crystal ball on your equipment,
the calc goes like this. The only real known
in the equation, is that "1 hour has 60 minutes".

12V*18AH 60m
------------ * --- * eff = minutes of time
100W+30W+10W 1H

216WH 60m
----- * --- * eff = 93m * eff
140W 1H

If eff was 0.7, time is 65m. I have no idea what the
efficiency of the inverter on a UPS is. The 0.7 assumes
a conventional switcher with no tricks. It really
shouldn't be all that efficient, as it's switching
12V on the primary side, and that can't be all that good.
We cannot assume the UPS is "too inefficient", as that
would make the UPS heat up when the mains go off, and
I've never noticed a bad heating effect on mine when
the mains go off.

The battery AH rating goes down at high current
draw. Usually the product has a curve for derating
purposes.

The idle power of my best modern machine in the
house, a desktop, is 60W. My newest machine
is 100W to 110W at idle. I picked 100W for
my example calc. The oldest machine in the
house is 150W, and on that one, the power
doesn't vary at all under high CPU load.
The old machines are... just wasteful.
Low MIPS, high watts. If gaming, whatever
video card was in the old box, the power would
double on the card.

Your best bet is to find the curve for your product,
to take some of the unknowns out of the calc.
You can also use a Kill-A-Watt meter to determine
AC load of each appliance connected to the UPS.

https://www.homedepot.com/p/Kill-A-W...4400/202196386

Useful numbers from the Kill-A-Watt meter would be
"Watts" and "PF". And the documentation for the UPS
will have a different output limit for "Watts" versus
"VA" loading (the ratio of those is PF), which is making
a statement about the PF or Power Factor of the load(s).
Old ATX PSUs have a PF of 0.65 to 0.7 or so (phase relationship
of voltage to current on mains side). A modern ATX PSU
with active PFC (PF correction) runs 0.99 and has
the voltage and current are in-phase like on a
light bulb load. Your three electrical loads
all have switchers, so the PF should be
similar for them. Again, a rough wild-guess
pulled from thin air.

If you have:

1) A clamp-on DC ammeter (clamped on battery lead).
2) A multimeter to measure battery voltage.
3) A Kill-A-Watt to measure output load power.

you could get a value for eff yourself. Then
the only remaining part of the puzzle is
battery performance -- how the 18AH is
degraded to 9AH if you draw 40A out
of the battery during the event. A lighter
electrical load, gets closer to the battery
nominal AH rating during the discharge process.

Also, it isn't really all that good of an idea
to be running the battery pak flat for fun. That
reduces the service life of the battery. By pampering
mine, I got 10 years from it. Yes, the power
goes off here, but I try to do an orderly shutdown
in about 2 minutes if I can manage it. Only
a small percentage of the 18AH gets used when
you do it that way, and it doesn't take too much
time to restore the charge.

For car batteries, you only run them down 25% before you
stop using it. So if a car battery was 18AH, you can
only safely use 4.5AH (16 minutes worth) if the load
switches itself off at a safe point in time. There
are also deep discharge batteries for things
like golf carts or forklifts, which have thicker
plates, and more of their 18AH could be used
without causing immediate battery failure.
I don't know what that factor is for an SLA
battery in a UPS. I treated my UPS like it was
a car battery (calcium added to lead plates),
just to be on the safe side.

That will give you some idea how many variables
there are.

eff (switcher/inverter on output stage)
battery derating curve at load point (not efficient at high load)
percentage of battery used before UPS threshold trips out

actual load and PF of item(s) making up the load

Lots of variables.

Paul

"Crystal ball" comments are unnecessary.

I provide what info I think is necessary.

If more is need, just ask away.

Andy

For some components of the math, the values need
to be *measured*, as calculating leaves a wide
error margin. I could easily be off by a factor
of two, even if I knew what CPU your computer used.

My new machine uses a 156W CPU, my older computer
used a 36W CPU. Those are measured, full power numbers,
not idle. Measured with the clamp-on ammeter. At idle,
both processors use around 13W (12V at 1.1 amps).
It's hard to say on some computers, where exactly
the rest of the load is coming from.

The Kill-A-Watt meter is cheap, and provides
an accurate overall picture for individual AC powered
devices. In fact, I would recommend a Kill-A-Watt
as a "pre-purchase" before you go shopping for
a UPS, to enable you to save money by buying
just the right size of UPS for the job.

The clamp-on ammeter answers questions such
is "why is this motherboard getting hot" or
"did Intel make a CPU that draws more than
the rated power". Those kinds of questions.
When the Kill-A-Watt gives a number you didn't
expect, a $200 to $300 Clamp-On DC ammeter
is able to give more details about individual
subsystems within the computer. A $100 AC ammeter
is not a substitute, because the "interesting"
things inside the computer are DC. There are
AC and AC/DC ammeters with jaws that clamp
around wires, and for computer work, I need
the DC.

The DC also comes in handy for your
car. I now know a defective starter motor
on the car draws 150 amps, and causes the
battery voltage to temporarily drop to 9V
when the starter motor stalls. The clamp-on
DC ammeter with "peak hold" told me the highest
current flow number seen. You cannot easily
do such a measurement with a shunt. And using
shunts all over the place while working, is inconvenient.

As far as the rest of the variables at play goes,
some are pretty expensive to figure out on your
own. Working out the battery efficiency
curve experimentally, would probably destroy
at least one battery doing all the charge/discharge
cycles. The manufacturer derating curve (available
per UPS model), is handy.

By using stepwise refinement, you can make
yourself a better number.

*******

I don't really have a lot of tools here for
electrical work. These are my main ones:

1) $100 multimeter (for volts and ohms, not current).
You don't need all the ranges mine has, and you can
get similar ones for $20. I can do Hfe on a transistor
with mine.

2) $300 clamp-on AC/DC ammeter 10mA to 400A dynamic range
Has been used for both AC and DC work. The 400A DC range
allowed me to work on the car.

3) $30 Kill-A-Watt P4400. Cheap and accurate.
Don't make a habit of running it near the upper
limit, for more than a few minutes. You don't
really need to check your electric kettle :-)
If you want to experiment with dumb loads,
test a 60W light bulb and see if it draws 60W.
I generally draw the line at 500W loads.

And that's about it. I don't own a scope. Or scope
probes. A good scope costs $35000. Nice probes
can cost $2000 (amplifier in probe tip). Cheap scopes
aren't really worth owning. We had a $35000 scope
on every lab bench at work. You can see my home budget
is rather limited by comparison.

Paul

In message , Paul
writes:
[]
And that's about it. I don't own a scope. Or scope
probes. A good scope costs $35000. Nice probes
can cost $2000 (amplifier in probe tip). Cheap scopes
aren't really worth owning. We had a $35000 scope
on every lab bench at work. You can see my home budget
is rather limited by comparison.

Paul

Sometimes good, if a little old-fashioned in technology, ones come up
when companies upgrade - especially if they have sell-to-staff
arrangements. I have one that was good in its day (Tek 7000 series, with
high-bandwidth [for then] plugins) through such a scheme; OK, it's bulky
because it's CRT, but ... depends what you want to do with it. In an
earlier generation, a colleague had one - Tek 545 series I _think_ -
which was a _genuine_ two-beam tube (not simulated by rapid
alternation): two timebases, so quite disconcerting to watch!
--
J. P. Gilliver. UMRA: 1960/1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

10.0 times 0.1 is hardly ever 1.0.

J. P. Gilliver (John) wrote:
In message , Paul
writes:
[]
And that's about it. I don't own a scope. Or scope
probes. A good scope costs $35000. Nice probes
can cost $2000 (amplifier in probe tip). Cheap scopes
aren't really worth owning. We had a $35000 scope
on every lab bench at work. You can see my home budget
is rather limited by comparison.

Paul

Sometimes good, if a little old-fashioned in technology, ones come up
when companies upgrade - especially if they have sell-to-staff
arrangements. I have one that was good in its day (Tek 7000 series, with
high-bandwidth [for then] plugins) through such a scheme; OK, it's bulky
because it's CRT, but ... depends what you want to do with it. In an
earlier generation, a colleague had one - Tek 545 series I _think_ -
which was a _genuine_ two-beam tube (not simulated by rapid
alternation): two timebases, so quite disconcerting to watch!

The 7000 series had some nice plug-in modules.

My favorite (and not too common), was the diff amp
that on the highest gain setting, could display
your heartbeat while you held onto a pair of wires.
It would do 1mV differential full scale or so.
It probably wouldn't have been good enough for
brain waves at 0.1mV. And I don't remember anyone
trying that either (there never seem to be the
right electrodes around when you need a pair).

For computer work, or digital logic design, you
need gear fast enough to display what's going on.
Our kit stopped at 40GHz statistical sampling
(20GHz Nyquist) back then, with rigid plumbing.
And then you have to find someone to help you
with sampling technique.

We had one electronic component that had a solution
for this. It was a chip with what I call "perfect IO",
which is matched to the transmission line so there
are no reflections (TTL logic, by comparison, isn't
even remotely close to being "perfect IO" - ordinary
logic signals are horrid).

At the time, that chip interface didn't go too fast,
maybe 2Gbit/sec or so as a bus lane of sorts. Well, one
of our engineers phoned up the chip maker and said
"what's the best way to verify the SI on this thing?".
And the person on the other end of the line said
"the chip has an internal sampling scope on that
signal". The chip could take samples of its own
signals, across the internal diff termination
resistor. There was a little utility for pulling
out the resulting waveform table, for display
on a computer screen. I thought that was a
great concept, but industry-wide, I don't think
that caught on. And there was no paper trail
for that feature either, so we didn't get to
find out how it was implemented inside. The main
chip spec made no mention of that, but their
engineers were probably using that feature for
verifying the perfection of their IO pad design
in a chip lab. The engineer who got that answer,
was kinda in a state of shock (that the answer would
be so easy). It required zero prep work in the lab.

*******

Back in those days, the fastest scope you could buy,
was one of these. The Nyquist is 70GHz or so.
The unit might use liquid nitrogen for cooling.

Hypres scope

https://ieeexplore.ieee.org/iel5/6/6373637/06448112.pdf

In the promo picture there, the second scope from
the left looks like "son of 7000", with the same
sort of module layout.

If it wasn't for the cryogenic requirement for that
scope, I bet they would have sold a lot more. It's not
that LN2 is particularly hard to come by, but finding
a body to keep filling the dewar would be hard. It's
like finding someone to water house plants at work.

Paul

On Friday, July 27, 2018 at 3:13:37 PM UTC-5, Paul wrote:
Andy wrote:
On Friday, July 27, 2018 at 2:44:20 AM UTC-5, Paul wrote:
Andy wrote:
I have an 18 Ah APC for my computer.

My computer,LCD monitor, and modem are plugged into it.

Battery indicator shows 29 mins of time left.

Does that sound reasonable?

I thought it would give much greater run time than with the original battery that it came with.

Thanks,
Andy
Using my crystal ball on your equipment,
the calc goes like this. The only real known
in the equation, is that "1 hour has 60 minutes".

12V*18AH 60m
------------ * --- * eff = minutes of time
100W+30W+10W 1H

216WH 60m
----- * --- * eff = 93m * eff
140W 1H

If eff was 0.7, time is 65m. I have no idea what the
efficiency of the inverter on a UPS is. The 0.7 assumes
a conventional switcher with no tricks. It really
shouldn't be all that efficient, as it's switching
12V on the primary side, and that can't be all that good.
We cannot assume the UPS is "too inefficient", as that
would make the UPS heat up when the mains go off, and
I've never noticed a bad heating effect on mine when
the mains go off.

The battery AH rating goes down at high current
draw. Usually the product has a curve for derating
purposes.

The idle power of my best modern machine in the
house, a desktop, is 60W. My newest machine
is 100W to 110W at idle. I picked 100W for
my example calc. The oldest machine in the
house is 150W, and on that one, the power
doesn't vary at all under high CPU load.
The old machines are... just wasteful.
Low MIPS, high watts. If gaming, whatever
video card was in the old box, the power would
double on the card.

Your best bet is to find the curve for your product,
to take some of the unknowns out of the calc.
You can also use a Kill-A-Watt meter to determine
AC load of each appliance connected to the UPS.

https://www.homedepot.com/p/Kill-A-W...4400/202196386

Useful numbers from the Kill-A-Watt meter would be
"Watts" and "PF". And the documentation for the UPS
will have a different output limit for "Watts" versus
"VA" loading (the ratio of those is PF), which is making
a statement about the PF or Power Factor of the load(s).
Old ATX PSUs have a PF of 0.65 to 0.7 or so (phase relationship
of voltage to current on mains side). A modern ATX PSU
with active PFC (PF correction) runs 0.99 and has
the voltage and current are in-phase like on a
light bulb load. Your three electrical loads
all have switchers, so the PF should be
similar for them. Again, a rough wild-guess
pulled from thin air.

If you have:

1) A clamp-on DC ammeter (clamped on battery lead).
2) A multimeter to measure battery voltage.
3) A Kill-A-Watt to measure output load power.

you could get a value for eff yourself. Then
the only remaining part of the puzzle is
battery performance -- how the 18AH is
degraded to 9AH if you draw 40A out
of the battery during the event. A lighter
electrical load, gets closer to the battery
nominal AH rating during the discharge process.

Also, it isn't really all that good of an idea
to be running the battery pak flat for fun. That
reduces the service life of the battery. By pampering
mine, I got 10 years from it. Yes, the power
goes off here, but I try to do an orderly shutdown
in about 2 minutes if I can manage it. Only
a small percentage of the 18AH gets used when
you do it that way, and it doesn't take too much
time to restore the charge.

For car batteries, you only run them down 25% before you
stop using it. So if a car battery was 18AH, you can
only safely use 4.5AH (16 minutes worth) if the load
switches itself off at a safe point in time. There
are also deep discharge batteries for things
like golf carts or forklifts, which have thicker
plates, and more of their 18AH could be used
without causing immediate battery failure.
I don't know what that factor is for an SLA
battery in a UPS. I treated my UPS like it was
a car battery (calcium added to lead plates),
just to be on the safe side.

That will give you some idea how many variables
there are.

eff (switcher/inverter on output stage)
battery derating curve at load point (not efficient at high load)
percentage of battery used before UPS threshold trips out

actual load and PF of item(s) making up the load

Lots of variables.

Paul

"Crystal ball" comments are unnecessary.

I provide what info I think is necessary.

If more is need, just ask away.

Andy

For some components of the math, the values need
to be *measured*, as calculating leaves a wide
error margin. I could easily be off by a factor
of two, even if I knew what CPU your computer used.

My new machine uses a 156W CPU, my older computer
used a 36W CPU. Those are measured, full power numbers,
not idle. Measured with the clamp-on ammeter. At idle,
both processors use around 13W (12V at 1.1 amps).
It's hard to say on some computers, where exactly
the rest of the load is coming from.

The Kill-A-Watt meter is cheap, and provides
an accurate overall picture for individual AC powered
devices. In fact, I would recommend a Kill-A-Watt
as a "pre-purchase" before you go shopping for
a UPS, to enable you to save money by buying
just the right size of UPS for the job.

The clamp-on ammeter answers questions such
is "why is this motherboard getting hot" or
"did Intel make a CPU that draws more than
the rated power". Those kinds of questions.
When the Kill-A-Watt gives a number you didn't
expect, a $200 to $300 Clamp-On DC ammeter
is able to give more details about individual
subsystems within the computer. A $100 AC ammeter
is not a substitute, because the "interesting"
things inside the computer are DC. There are
AC and AC/DC ammeters with jaws that clamp
around wires, and for computer work, I need
the DC.

The DC also comes in handy for your
car. I now know a defective starter motor
on the car draws 150 amps, and causes the
battery voltage to temporarily drop to 9V
when the starter motor stalls. The clamp-on
DC ammeter with "peak hold" told me the highest
current flow number seen. You cannot easily
do such a measurement with a shunt. And using
shunts all over the place while working, is inconvenient.

As far as the rest of the variables at play goes,
some are pretty expensive to figure out on your
own. Working out the battery efficiency
curve experimentally, would probably destroy
at least one battery doing all the charge/discharge
cycles. The manufacturer derating curve (available
per UPS model), is handy.

By using stepwise refinement, you can make
yourself a better number.

*******

I don't really have a lot of tools here for
electrical work. These are my main ones:

1) $100 multimeter (for volts and ohms, not current).
You don't need all the ranges mine has, and you can
get similar ones for $20. I can do Hfe on a transistor
with mine.

2) $300 clamp-on AC/DC ammeter 10mA to 400A dynamic range
Has been used for both AC and DC work. The 400A DC range
allowed me to work on the car.

3) $30 Kill-A-Watt P4400. Cheap and accurate.
Don't make a habit of running it near the upper
limit, for more than a few minutes. You don't
really need to check your electric kettle :-)
If you want to experiment with dumb loads,
test a 60W light bulb and see if it draws 60W.
I generally draw the line at 500W loads.

And that's about it. I don't own a scope. Or scope
probes. A good scope costs $35000. Nice probes
can cost $2000 (amplifier in probe tip). Cheap scopes
aren't really worth owning. We had a $35000 scope
on every lab bench at work. You can see my home budget
is rather limited by comparison.

Paul

I have order a kill o watt.

It seems to be a worthwhile tool to have.

I also have a RMS multimeter.

Andy

On Saturday, July 28, 2018 at 2:56:47 PM UTC-5, Andy wrote:
On Friday, July 27, 2018 at 3:13:37 PM UTC-5, Paul wrote:
Andy wrote:
On Friday, July 27, 2018 at 2:44:20 AM UTC-5, Paul wrote:
Andy wrote:
I have an 18 Ah APC for my computer.

My computer,LCD monitor, and modem are plugged into it.

Battery indicator shows 29 mins of time left.

Does that sound reasonable?

I thought it would give much greater run time than with the original battery that it came with.

Thanks,
Andy
Using my crystal ball on your equipment,
the calc goes like this. The only real known
in the equation, is that "1 hour has 60 minutes".

12V*18AH 60m
------------ * --- * eff = minutes of time
100W+30W+10W 1H

216WH 60m
----- * --- * eff = 93m * eff
140W 1H

If eff was 0.7, time is 65m. I have no idea what the
efficiency of the inverter on a UPS is. The 0.7 assumes
a conventional switcher with no tricks. It really
shouldn't be all that efficient, as it's switching
12V on the primary side, and that can't be all that good.
We cannot assume the UPS is "too inefficient", as that
would make the UPS heat up when the mains go off, and
I've never noticed a bad heating effect on mine when
the mains go off.

The battery AH rating goes down at high current
draw. Usually the product has a curve for derating
purposes.

The idle power of my best modern machine in the
house, a desktop, is 60W. My newest machine
is 100W to 110W at idle. I picked 100W for
my example calc. The oldest machine in the
house is 150W, and on that one, the power
doesn't vary at all under high CPU load.
The old machines are... just wasteful.
Low MIPS, high watts. If gaming, whatever
video card was in the old box, the power would
double on the card.

Your best bet is to find the curve for your product,
to take some of the unknowns out of the calc.
You can also use a Kill-A-Watt meter to determine
AC load of each appliance connected to the UPS.

https://www.homedepot.com/p/Kill-A-W...4400/202196386

Useful numbers from the Kill-A-Watt meter would be
"Watts" and "PF". And the documentation for the UPS
will have a different output limit for "Watts" versus
"VA" loading (the ratio of those is PF), which is making
a statement about the PF or Power Factor of the load(s).
Old ATX PSUs have a PF of 0.65 to 0.7 or so (phase relationship
of voltage to current on mains side). A modern ATX PSU
with active PFC (PF correction) runs 0.99 and has
the voltage and current are in-phase like on a
light bulb load. Your three electrical loads
all have switchers, so the PF should be
similar for them. Again, a rough wild-guess
pulled from thin air.

If you have:

1) A clamp-on DC ammeter (clamped on battery lead).
2) A multimeter to measure battery voltage.
3) A Kill-A-Watt to measure output load power.

you could get a value for eff yourself. Then
the only remaining part of the puzzle is
battery performance -- how the 18AH is
degraded to 9AH if you draw 40A out
of the battery during the event. A lighter
electrical load, gets closer to the battery
nominal AH rating during the discharge process.

Also, it isn't really all that good of an idea
to be running the battery pak flat for fun. That
reduces the service life of the battery. By pampering
mine, I got 10 years from it. Yes, the power
goes off here, but I try to do an orderly shutdown
in about 2 minutes if I can manage it. Only
a small percentage of the 18AH gets used when
you do it that way, and it doesn't take too much
time to restore the charge.

For car batteries, you only run them down 25% before you
stop using it. So if a car battery was 18AH, you can
only safely use 4.5AH (16 minutes worth) if the load
switches itself off at a safe point in time. There
are also deep discharge batteries for things
like golf carts or forklifts, which have thicker
plates, and more of their 18AH could be used
without causing immediate battery failure.
I don't know what that factor is for an SLA
battery in a UPS. I treated my UPS like it was
a car battery (calcium added to lead plates),
just to be on the safe side.

That will give you some idea how many variables
there are.

eff (switcher/inverter on output stage)
battery derating curve at load point (not efficient at high load)
percentage of battery used before UPS threshold trips out

actual load and PF of item(s) making up the load

Lots of variables.

Paul

"Crystal ball" comments are unnecessary.

I provide what info I think is necessary.

If more is need, just ask away.

Andy

For some components of the math, the values need
to be *measured*, as calculating leaves a wide
error margin. I could easily be off by a factor
of two, even if I knew what CPU your computer used.

My new machine uses a 156W CPU, my older computer
used a 36W CPU. Those are measured, full power numbers,
not idle. Measured with the clamp-on ammeter. At idle,
both processors use around 13W (12V at 1.1 amps).
It's hard to say on some computers, where exactly
the rest of the load is coming from.

The Kill-A-Watt meter is cheap, and provides
an accurate overall picture for individual AC powered
devices. In fact, I would recommend a Kill-A-Watt
as a "pre-purchase" before you go shopping for
a UPS, to enable you to save money by buying
just the right size of UPS for the job.

The clamp-on ammeter answers questions such
is "why is this motherboard getting hot" or
"did Intel make a CPU that draws more than
the rated power". Those kinds of questions.
When the Kill-A-Watt gives a number you didn't
expect, a $200 to $300 Clamp-On DC ammeter
is able to give more details about individual
subsystems within the computer. A $100 AC ammeter
is not a substitute, because the "interesting"
things inside the computer are DC. There are
AC and AC/DC ammeters with jaws that clamp
around wires, and for computer work, I need
the DC.

The DC also comes in handy for your
car. I now know a defective starter motor
on the car draws 150 amps, and causes the
battery voltage to temporarily drop to 9V
when the starter motor stalls. The clamp-on
DC ammeter with "peak hold" told me the highest
current flow number seen. You cannot easily
do such a measurement with a shunt. And using
shunts all over the place while working, is inconvenient.

As far as the rest of the variables at play goes,
some are pretty expensive to figure out on your
own. Working out the battery efficiency
curve experimentally, would probably destroy
at least one battery doing all the charge/discharge
cycles. The manufacturer derating curve (available
per UPS model), is handy.

By using stepwise refinement, you can make
yourself a better number.

*******

I don't really have a lot of tools here for
electrical work. These are my main ones:

1) $100 multimeter (for volts and ohms, not current).
You don't need all the ranges mine has, and you can
get similar ones for $20. I can do Hfe on a transistor
with mine.

2) $300 clamp-on AC/DC ammeter 10mA to 400A dynamic range
Has been used for both AC and DC work. The 400A DC range
allowed me to work on the car.

3) $30 Kill-A-Watt P4400. Cheap and accurate.
Don't make a habit of running it near the upper
limit, for more than a few minutes. You don't
really need to check your electric kettle :-)
If you want to experiment with dumb loads,
test a 60W light bulb and see if it draws 60W.
I generally draw the line at 500W loads.

And that's about it. I don't own a scope. Or scope
probes. A good scope costs $35000. Nice probes
can cost $2000 (amplifier in probe tip). Cheap scopes
aren't really worth owning. We had a $35000 scope
on every lab bench at work. You can see my home budget
is rather limited by comparison.

Paul

I have order a kill o watt.

It seems to be a worthwhile tool to have.

I also have a RMS multimeter.

Andy

Kill o watt shows

90 watts for computer during startup
about 60 watts after desktop shows

monitor uses 75 watts.

Andy wrote:
On Saturday, July 28, 2018 at 2:56:47 PM UTC-5, Andy wrote:
On Friday, July 27, 2018 at 3:13:37 PM UTC-5, Paul wrote:
Andy wrote:
On Friday, July 27, 2018 at 2:44:20 AM UTC-5, Paul wrote:
Andy wrote:
I have an 18 Ah APC for my computer.

My computer,LCD monitor, and modem are plugged into it.

Battery indicator shows 29 mins of time left.

Does that sound reasonable?

I thought it would give much greater run time than with the original battery that it came with.

Thanks,
Andy
Using my crystal ball on your equipment,
the calc goes like this. The only real known
in the equation, is that "1 hour has 60 minutes".

12V*18AH 60m
------------ * --- * eff = minutes of time
100W+30W+10W 1H

216WH 60m
----- * --- * eff = 93m * eff
140W 1H

If eff was 0.7, time is 65m. I have no idea what the
efficiency of the inverter on a UPS is. The 0.7 assumes
a conventional switcher with no tricks. It really
shouldn't be all that efficient, as it's switching
12V on the primary side, and that can't be all that good.
We cannot assume the UPS is "too inefficient", as that
would make the UPS heat up when the mains go off, and
I've never noticed a bad heating effect on mine when
the mains go off.

The battery AH rating goes down at high current
draw. Usually the product has a curve for derating
purposes.

The idle power of my best modern machine in the
house, a desktop, is 60W. My newest machine
is 100W to 110W at idle. I picked 100W for
my example calc. The oldest machine in the
house is 150W, and on that one, the power
doesn't vary at all under high CPU load.
The old machines are... just wasteful.
Low MIPS, high watts. If gaming, whatever
video card was in the old box, the power would
double on the card.

Your best bet is to find the curve for your product,
to take some of the unknowns out of the calc.
You can also use a Kill-A-Watt meter to determine
AC load of each appliance connected to the UPS.

https://www.homedepot.com/p/Kill-A-W...4400/202196386

Useful numbers from the Kill-A-Watt meter would be
"Watts" and "PF". And the documentation for the UPS
will have a different output limit for "Watts" versus
"VA" loading (the ratio of those is PF), which is making
a statement about the PF or Power Factor of the load(s).
Old ATX PSUs have a PF of 0.65 to 0.7 or so (phase relationship
of voltage to current on mains side). A modern ATX PSU
with active PFC (PF correction) runs 0.99 and has
the voltage and current are in-phase like on a
light bulb load. Your three electrical loads
all have switchers, so the PF should be
similar for them. Again, a rough wild-guess
pulled from thin air.

If you have:

1) A clamp-on DC ammeter (clamped on battery lead).
2) A multimeter to measure battery voltage.
3) A Kill-A-Watt to measure output load power.

you could get a value for eff yourself. Then
the only remaining part of the puzzle is
battery performance -- how the 18AH is
degraded to 9AH if you draw 40A out
of the battery during the event. A lighter
electrical load, gets closer to the battery
nominal AH rating during the discharge process.

Also, it isn't really all that good of an idea
to be running the battery pak flat for fun. That
reduces the service life of the battery. By pampering
mine, I got 10 years from it. Yes, the power
goes off here, but I try to do an orderly shutdown
in about 2 minutes if I can manage it. Only
a small percentage of the 18AH gets used when
you do it that way, and it doesn't take too much
time to restore the charge.

For car batteries, you only run them down 25% before you
stop using it. So if a car battery was 18AH, you can
only safely use 4.5AH (16 minutes worth) if the load
switches itself off at a safe point in time. There
are also deep discharge batteries for things
like golf carts or forklifts, which have thicker
plates, and more of their 18AH could be used
without causing immediate battery failure.
I don't know what that factor is for an SLA
battery in a UPS. I treated my UPS like it was
a car battery (calcium added to lead plates),
just to be on the safe side.

That will give you some idea how many variables
there are.

eff (switcher/inverter on output stage)
battery derating curve at load point (not efficient at high load)
percentage of battery used before UPS threshold trips out

actual load and PF of item(s) making up the load

Lots of variables.

Paul
"Crystal ball" comments are unnecessary.

I provide what info I think is necessary.

If more is need, just ask away.

Andy
For some components of the math, the values need
to be *measured*, as calculating leaves a wide
error margin. I could easily be off by a factor
of two, even if I knew what CPU your computer used.

My new machine uses a 156W CPU, my older computer
used a 36W CPU. Those are measured, full power numbers,
not idle. Measured with the clamp-on ammeter. At idle,
both processors use around 13W (12V at 1.1 amps).
It's hard to say on some computers, where exactly
the rest of the load is coming from.

The Kill-A-Watt meter is cheap, and provides
an accurate overall picture for individual AC powered
devices. In fact, I would recommend a Kill-A-Watt
as a "pre-purchase" before you go shopping for
a UPS, to enable you to save money by buying
just the right size of UPS for the job.

The clamp-on ammeter answers questions such
is "why is this motherboard getting hot" or
"did Intel make a CPU that draws more than
the rated power". Those kinds of questions.
When the Kill-A-Watt gives a number you didn't
expect, a $200 to $300 Clamp-On DC ammeter
is able to give more details about individual
subsystems within the computer. A $100 AC ammeter
is not a substitute, because the "interesting"
things inside the computer are DC. There are
AC and AC/DC ammeters with jaws that clamp
around wires, and for computer work, I need
the DC.

The DC also comes in handy for your
car. I now know a defective starter motor
on the car draws 150 amps, and causes the
battery voltage to temporarily drop to 9V
when the starter motor stalls. The clamp-on
DC ammeter with "peak hold" told me the highest
current flow number seen. You cannot easily
do such a measurement with a shunt. And using
shunts all over the place while working, is inconvenient.

As far as the rest of the variables at play goes,
some are pretty expensive to figure out on your
own. Working out the battery efficiency
curve experimentally, would probably destroy
at least one battery doing all the charge/discharge
cycles. The manufacturer derating curve (available
per UPS model), is handy.

By using stepwise refinement, you can make
yourself a better number.

*******

I don't really have a lot of tools here for
electrical work. These are my main ones:

1) $100 multimeter (for volts and ohms, not current).
You don't need all the ranges mine has, and you can
get similar ones for $20. I can do Hfe on a transistor
with mine.

2) $300 clamp-on AC/DC ammeter 10mA to 400A dynamic range
Has been used for both AC and DC work. The 400A DC range
allowed me to work on the car.

3) $30 Kill-A-Watt P4400. Cheap and accurate.
Don't make a habit of running it near the upper
limit, for more than a few minutes. You don't
really need to check your electric kettle :-)
If you want to experiment with dumb loads,
test a 60W light bulb and see if it draws 60W.
I generally draw the line at 500W loads.

And that's about it. I don't own a scope. Or scope
probes. A good scope costs $35000. Nice probes
can cost $2000 (amplifier in probe tip). Cheap scopes
aren't really worth owning. We had a $35000 scope
on every lab bench at work. You can see my home budget
is rather limited by comparison.

Paul
I have order a kill o watt.

It seems to be a worthwhile tool to have.

I also have a RMS multimeter.

Andy

Kill o watt shows

90 watts for computer during startup
about 60 watts after desktop shows

monitor uses 75 watts.

Which means your runtime estimate calculation
should have a lesser number now.

When picking a UPS, both the watt and VA numbers
can be used. As UPSes usually have a max watt number
and a max VA number on output. The holdup times
should be in a graph for the product, on the
product web page. The graph may be in the user
manual, but not always.

Paul

On Monday, July 30, 2018 at 4:13:36 PM UTC-5, Paul wrote:
Andy wrote:
On Saturday, July 28, 2018 at 2:56:47 PM UTC-5, Andy wrote:
On Friday, July 27, 2018 at 3:13:37 PM UTC-5, Paul wrote:
Andy wrote:
On Friday, July 27, 2018 at 2:44:20 AM UTC-5, Paul wrote:
Andy wrote:
I have an 18 Ah APC for my computer.

My computer,LCD monitor, and modem are plugged into it.

Battery indicator shows 29 mins of time left.

Does that sound reasonable?

I thought it would give much greater run time than with the original battery that it came with.

Thanks,
Andy
Using my crystal ball on your equipment,
the calc goes like this. The only real known
in the equation, is that "1 hour has 60 minutes".

12V*18AH 60m
------------ * --- * eff = minutes of time
100W+30W+10W 1H

216WH 60m
----- * --- * eff = 93m * eff
140W 1H

If eff was 0.7, time is 65m. I have no idea what the
efficiency of the inverter on a UPS is. The 0.7 assumes
a conventional switcher with no tricks. It really
shouldn't be all that efficient, as it's switching
12V on the primary side, and that can't be all that good.
We cannot assume the UPS is "too inefficient", as that
would make the UPS heat up when the mains go off, and
I've never noticed a bad heating effect on mine when
the mains go off.

The battery AH rating goes down at high current
draw. Usually the product has a curve for derating
purposes.

The idle power of my best modern machine in the
house, a desktop, is 60W. My newest machine
is 100W to 110W at idle. I picked 100W for
my example calc. The oldest machine in the
house is 150W, and on that one, the power
doesn't vary at all under high CPU load.
The old machines are... just wasteful.
Low MIPS, high watts. If gaming, whatever
video card was in the old box, the power would
double on the card.

Your best bet is to find the curve for your product,
to take some of the unknowns out of the calc.
You can also use a Kill-A-Watt meter to determine
AC load of each appliance connected to the UPS.

https://www.homedepot.com/p/Kill-A-W...4400/202196386

Useful numbers from the Kill-A-Watt meter would be
"Watts" and "PF". And the documentation for the UPS
will have a different output limit for "Watts" versus
"VA" loading (the ratio of those is PF), which is making
a statement about the PF or Power Factor of the load(s).
Old ATX PSUs have a PF of 0.65 to 0.7 or so (phase relationship
of voltage to current on mains side). A modern ATX PSU
with active PFC (PF correction) runs 0.99 and has
the voltage and current are in-phase like on a
light bulb load. Your three electrical loads
all have switchers, so the PF should be
similar for them. Again, a rough wild-guess
pulled from thin air.

If you have:

1) A clamp-on DC ammeter (clamped on battery lead).
2) A multimeter to measure battery voltage.
3) A Kill-A-Watt to measure output load power.

you could get a value for eff yourself. Then
the only remaining part of the puzzle is
battery performance -- how the 18AH is
degraded to 9AH if you draw 40A out
of the battery during the event. A lighter
electrical load, gets closer to the battery
nominal AH rating during the discharge process.

Also, it isn't really all that good of an idea
to be running the battery pak flat for fun. That
reduces the service life of the battery. By pampering
mine, I got 10 years from it. Yes, the power
goes off here, but I try to do an orderly shutdown
in about 2 minutes if I can manage it. Only
a small percentage of the 18AH gets used when
you do it that way, and it doesn't take too much
time to restore the charge.

For car batteries, you only run them down 25% before you
stop using it. So if a car battery was 18AH, you can
only safely use 4.5AH (16 minutes worth) if the load
switches itself off at a safe point in time. There
are also deep discharge batteries for things
like golf carts or forklifts, which have thicker
plates, and more of their 18AH could be used
without causing immediate battery failure.
I don't know what that factor is for an SLA
battery in a UPS. I treated my UPS like it was
a car battery (calcium added to lead plates),
just to be on the safe side.

That will give you some idea how many variables
there are.

eff (switcher/inverter on output stage)
battery derating curve at load point (not efficient at high load)
percentage of battery used before UPS threshold trips out

actual load and PF of item(s) making up the load

Lots of variables.

Paul
"Crystal ball" comments are unnecessary.

I provide what info I think is necessary.

If more is need, just ask away.

Andy
For some components of the math, the values need
to be *measured*, as calculating leaves a wide
error margin. I could easily be off by a factor
of two, even if I knew what CPU your computer used.

My new machine uses a 156W CPU, my older computer
used a 36W CPU. Those are measured, full power numbers,
not idle. Measured with the clamp-on ammeter. At idle,
both processors use around 13W (12V at 1.1 amps).
It's hard to say on some computers, where exactly
the rest of the load is coming from.

The Kill-A-Watt meter is cheap, and provides
an accurate overall picture for individual AC powered
devices. In fact, I would recommend a Kill-A-Watt
as a "pre-purchase" before you go shopping for
a UPS, to enable you to save money by buying
just the right size of UPS for the job.

The clamp-on ammeter answers questions such
is "why is this motherboard getting hot" or
"did Intel make a CPU that draws more than
the rated power". Those kinds of questions.
When the Kill-A-Watt gives a number you didn't
expect, a $200 to $300 Clamp-On DC ammeter
is able to give more details about individual
subsystems within the computer. A $100 AC ammeter
is not a substitute, because the "interesting"
things inside the computer are DC. There are
AC and AC/DC ammeters with jaws that clamp
around wires, and for computer work, I need
the DC.

The DC also comes in handy for your
car. I now know a defective starter motor
on the car draws 150 amps, and causes the
battery voltage to temporarily drop to 9V
when the starter motor stalls. The clamp-on
DC ammeter with "peak hold" told me the highest
current flow number seen. You cannot easily
do such a measurement with a shunt. And using
shunts all over the place while working, is inconvenient.

As far as the rest of the variables at play goes,
some are pretty expensive to figure out on your
own. Working out the battery efficiency
curve experimentally, would probably destroy
at least one battery doing all the charge/discharge
cycles. The manufacturer derating curve (available
per UPS model), is handy.

By using stepwise refinement, you can make
yourself a better number.

*******

I don't really have a lot of tools here for
electrical work. These are my main ones:

1) $100 multimeter (for volts and ohms, not current).
You don't need all the ranges mine has, and you can
get similar ones for $20. I can do Hfe on a transistor
with mine.

2) $300 clamp-on AC/DC ammeter 10mA to 400A dynamic range
Has been used for both AC and DC work. The 400A DC range
allowed me to work on the car.

3) $30 Kill-A-Watt P4400. Cheap and accurate.
Don't make a habit of running it near the upper
limit, for more than a few minutes. You don't
really need to check your electric kettle :-)
If you want to experiment with dumb loads,
test a 60W light bulb and see if it draws 60W.
I generally draw the line at 500W loads.

And that's about it. I don't own a scope. Or scope
probes. A good scope costs $35000. Nice probes
can cost $2000 (amplifier in probe tip). Cheap scopes
aren't really worth owning. We had a $35000 scope
on every lab bench at work. You can see my home budget
is rather limited by comparison.

Paul
I have order a kill o watt.

It seems to be a worthwhile tool to have.

I also have a RMS multimeter.

Andy

Kill o watt shows

90 watts for computer during startup
about 60 watts after desktop shows

monitor uses 75 watts.

Which means your runtime estimate calculation
should have a lesser number now.

When picking a UPS, both the watt and VA numbers
can be used. As UPSes usually have a max watt number
and a max VA number on output. The holdup times
should be in a graph for the product, on the
product web page. The graph may be in the user
manual, but not always.

Paul

O.K.

I ran it for an hour on the APS and shut it down when it showed I had around 15 minutes left.

But the remaining time value kept changing. I think it is only somewhat
accurate.

My settings when On UPS power,

When computer is inactive for 10 minutes, it is put to sleep.

When UPS power critically low, computer is shutdown.

When inactive for 10 mins, display is put to sleep.

Andy

In message , Andy
writes:

[144 lines snipped!!!]


O.K.

I ran it for an hour on the APS and shut it down when it showed I had
around 15 minutes left.

Where does this "show" - on a display on the UPS/APS, or on the computer
(via a serial or USB link to the UPS/APS)?

But the remaining time value kept changing. I think it is only somewhat
accurate.

Or, like time to finish a multi-file move or copy, it might change as
the load changes. Might be worth having Task Manager open and seeing if
there's any (inverse) correlation between CPU load and projected time
left.

My settings when On UPS power,

When computer is inactive for 10 minutes, it is put to sleep.

When UPS power critically low, computer is shutdown.

When inactive for 10 mins, display is put to sleep.

Andy
--
J. P. Gilliver. UMRA: 1960/1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

Does my Bradshaw look big in this?