Re: "Mysterious" system crashes

On Nov 17, 4:56 am, "John Wallace" <johnwalla...@xxxxxxxxxxxxxxxx>
wrote:
"Doug Phillips" <dphil...@xxxxxxxxxxxx> wrote in message

news:d5cbca11-5541-4f02-ae51-531ce9ee0169@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx



On Nov 16, 5:01 pm, "Richard B. Gilbert" <rgilber...@xxxxxxxxxxx>
wrote:
Doug Phillips wrote:
On Nov 16, 3:54 pm, s...@xxxxxxxxxxxx (Steven M. Schweda) wrote:

From: Doug Phillips <dphil...@xxxxxxxxxxxx>

VA and the Amp rating on your tags are peak draw. Watts are average
draw. Figure Watts at around 65% of VA. [...]

If you believe this, you'd be wise to avoid offering advice (or
taking your own) on electrical topics. I'd offer a more detailed
explanation, but books already exist, and it is, in a sense, complex,
and requires a bit of imagination.

65% is a "rule of thumb." The calculation to arrive at actual numbers
is a bit more involved, but it isn't really "complex".

Ohm's law and the power formula involves simple math --- at least it
did when I took electronics and worked in that trade before getting
into software. I even remember some of the mnemonic rhymes.

Ohm's law is not strictly applicable to alternating current except in a
purely resistive circuit! With AC both capacitive and inductive
reactance must be taken into account in addition to pure resistance.

So E=IR and P=EI=I^2R for a light bulb but not necessarily for an
electric motor or any other reactive load. Some electric motors can be
made to appear as a capacitive load and the power company may give you
big discounts for running one. I believe it helps balance out the
inductive loads created by other customers. For further details, consult
an electrical engineer.

Ohm's law for AC isn't too different; E = IZ where E & I are the
oscillating voltage (usually V rather than E) and current, and Z is
the complex impedance for the oscillation frequency. Yes, AC is more
complicated than DC, and does get more complicated from there, but
doesn't most everything?

Watts = VA, (twinkle twinkle little star, power equals I squared R ---
dang, now I'll have that running through my brain. At least I've
gotten the resistor color code jingle out of my mind.) When VA and
Watts are both shown on an appliance tag, and they're different
numbers, VA is used to indicate peak, and Watts is used to indicate
the average draw; the amount of power consumption you're most likely
to see on your electric meter if the thing runs all of the time.

On to better things.

Doug, either you don't fully know what you're talking about here or you're
expressing it really awkwardly.


Maybe a little of both? I guess I've forgotten most of what I once
knew, and I always seem to get jumped on by the "experts" here when I
try to discuss complicated things in simple terms that non-experts can
understand.

I figure the "experts" already understand the complicated thing and
don't need an explanation. If an "expert" wants to jump in and provide
some technical translation, then that's fine with me. If an "expert"
corrects or clarities something a non-expert says, I'm fine with that,
too. A bald "that's wrong" reply is not helpful to anyone.

I didn't mention power factor, but the active power number (W=VA*pf)
is handy to use in guesstimating the change you might see on your
electric bill even though that isn't what the number actually means.
That's why I didn't mention power factor. You can do the real math to
get a more exact estimate if you want. A typical PC power supply (not
a 1:1 power supply or a light bulb) will have a pf range between (~).
55 and .75, or about .65 average. Beyond that, power factor means
little to the home-computer UPS buyer other than when he's trying to
understand how the power company calculates his electric bill, and I
didn't intend to go there, and that's a completely different problem
anyway.

I certainly never intended to move into an EE or EEE level discussion,
so I tried to not use engineering terms and I tried to avoid
addressing complex concepts. You don't need to be an engineer or worry
about capacitance and inductance and power factors to buy a for-home-
use UPS if you use the numbers available to you and know a few ball-
parking shot cuts. Otherwise, hire an engineer, take a course or read
a book if you can't live with ball-parking.

If you're an engineer, then by all means share your expertise. Try to
use words that non-engineers will understand, though, and please move
into the real world and out of the text books when you make a
recommendation, thank you. Save the techno-babble for paying customers
and cocktail parties.

Since I have bought and used more SOHO size UPS's than I can count and
I've had very few problems with the units I've bought I thought I'd
share my experience. If you want to call me ignorant and dismiss my
experience, then feel free to do so.

If we were discussing a data center, I wouldn't have said anything
except maybe "hire an engineer." But, this is a home system we're
talking about and Brad indicated that a $600 UPS was over his budget,
and I've been there and done that.

Brad has given us the important numbers for his two 120V appliances:
5.5A (660VA) and 7.0A (840VA) for 12.5A and 1500VA total. So he can
use one 20A circuit but he'd better be careful about what else he
plugs into that circuit. With this type of equipment I'd ball-park
that my electric bill would see a less than 1 kW per Hr increase if I
run the equipment continuously. It will likely not be that high, but
it could be, and there's no quick way to know without measuring it
under real-world conditions. If we were talking about a 1500W electric
space heater or light bulb, then I'd calculate things differently.

Using the APC web-site UPS size calculator (which actually offers
Alpha CPU choices!!) for just the PWS shows Back-UPS' ranging in price
from $60 to $150 (higher if you move up to the Smart-UPS). Plugging in
a 840VA requirement, we find units priced from $100-$150 on up.
Looking further, there are even a few 1500VA-plus sized units for a
bit more than the two smaller units.

So, we see that he can buy one or two UPS' that will handle the load
and offer some protection for less than $600. That's buying new. If he
can find a working UPS or two that just need new batteries, that
should cost even less and would be better than having nothing.

I bought two new UPS's to use on a multiple device 1500VA rated load
for around $300. I could have bought smaller units for less money but
I wanted a few more minutes of up time. Like I said; it's for home,
not a data center. If the home system has data center importance, then
apply the more exact science and be prepared to spend more money. You
should budget according to the importance and value of what you're
protecting. If you can't afford to protect the system at all and you
can't afford to lose it, then you're gambling with fate and the cards
are stacked against you.

Even a surge protector (one better than just a power-strip with a
circuit breaker, of course) is better than nothing, and a UPS is
better than a surge protector, and so on in steps up to data-center
level protection.

I consider a UPS and/or power conditioning to be an important part of
*every* computer installation. You might not have a problem running
without a UPS or conditioning and protection, but you're less likely
to have a problem if you use a UPS and if you do have a problem it's
less likely to be power related. If you stay with a name-brand, you'll
find most manufacturers even offer an Equipment Protection Policy
(read the fine print, though.)


Either way, it'd be in your own interests
(and others) to stick to stuff where your expertise can be helpful.


Understood. Thanks for looking out for my interests. Doesn't
experience count or should all the people who aren't card-carrying
experts just shut up?

As Steven already said, this may not be the best place for an in depth
explanation of a "complex" subject. Those who want to know more about the
relationship between Volts, Amps, Watts, VA, and VAr can go learn about
"power factor" (which I'm surprised hasn't been mentioned already, but there
we are). People who will know about power factor are people with a clue
about AC electrical power engineering (and trigonometry), and also people
who really understand UPSes (which may or may not include UPS salespeople).


I guess one does need an engineering degree to buy a home UPS. Darn.

Brad: I looked into UPSes once.


Once? How long ago was that?

The cheaper ones just pass through the
external mains when it's within the voltage limits, and switch to battery
when they decide the external mains voltage (or, maybe, frequency) is out of
tolerance (relevant keywords used to be "online", "offline" and "line
interactive" but these words are not often seen in the world of SoHo UPSes).
Whether one of these simplistic UPSes will help this particular problem is
moot, if the problem is related to "dirty" mains, as the UPS may not care
whether the mains is "clean" or not, so long as its own measured voltage
looks OK.


I'm pretty sure that all of my newer UPS' say they are line
interactive (with a voltage regulating circuit) to handle transients
and brownouts. They switch to battery during a black-out and output a
stepped (pseudo) sine-wave. The spec I'm looking at indicates zero
switching time, but I've never scoped it. Some old computers might
have a problem with a non-sinoidal wave shape but the newer UPS' show
much lower distortion than the older units. If you haven't looked in a
while, you might want to look again.


If I was on a limited budget I'd probably get a SoHo UPS for the "PC" but
maybe not for the disk stuff (but I openly admit I haven't fully thought
through the implications of VMS being live while disks are unpowered).


I'd protect both. I'd like to get the equipment onto a clean,
dedicated circuit if I could, but that often isn't practical. I'd
certainly want to be aware of and concerned about other devices
plugged in to that circuit.

If you are seriously looking into UPSes and spending serious money you may
also want to check how your chosen UPS behaves when external power is
restored *after* the UPS shuts down due to flat battery. Many of the cheap
ones *don't* automatically restart, you have to manually do something (which
isn't real helpful in a potentially-unattended environment). They don't tell
you this before you buy (or in the manual) though.


The newer units I have all sound an alarm if power gets too flaky
(technical term;-) but will self-reset when the power stabilizes. If
the battery flat-lines and the power comes back on then the UPS will
resume its work and the battery will begin recharging. Most units come
with Power-Chute software for PeeCee's. VAXman's VMS product was
mentioned in another post. If the system has shut-down, then what it
does when the power comes back is dependent on other factors.

The older (or really cheap) UPS's didn't (don't) always come back
automatically. I have an old Tripp-Lite 450 that wanted attention all
of the time (the new Tripp-Lite's are worth looking at, though.) Its
battery just died again this year and I don't know if I'll replace it
or not. It was finally just handling my cell-phone chargers and
cordless phone & answering machine, anyway, and I gave that job to a
little old APC unit that has a fairly new battery.

Why would the power/crash problem start to appear after your suspect heating
appliances have been in service some time? I'll guess that contacts have
started to arc seriously, and weren't arcing much before. But it's only a
guess. If the problem is related to arcing I'm really not sure that any
sensibly-priced UPS will help. Fix the appliance, as the dirty power may
also affect other nearby electronic kit (TV, audio, etc), sometimes
permanently.


Exposing electronic equipment to power glitches, frequent on/off -
heating/cooling cycles or a bad environment (like too high or low
humidity and/or temperature) can create many insidious problems.
Components can degrade from optimal to marginal and until they fail,
they can present intermittent symptoms that are hard to diagnose. A
power supply can just naturally lose efficiency over time, too.

You're right, obviously, that a UPS (even a non-sensibly priced one)
won't fix any damage that's already done.

I also looked into "mysterious crashes" earlier this week. Not on VMS, but
on a built-like-a-tank antique Compaq Deskpro running Win2K (so no chance of
any help from the OS). It hadn't stayed up for more than a few hours for the
last few months, and often when dying it scribbled on the boot sector.
Basically I'd abandoned it, but I need it again temporarily. I did the "PM"
thing already described by others (get rid of bulk dust, reseat memory,
cards, connectors, and cables), plus removed unused PCI cards, and changed a
badly creased IDE cable. So far it's been up longer than it has for ages
(but we know what Mr Murphy wants next don't we).


We've all met Mr. Murphy, I'm sure. He tells me I've probably said
something in this post that you disagree with so you (and Steven)
still think I'm an idiot --- but I can live with that.
.