Pure Sine Wave UPSes for New Dell PCs

I had to give myself a basic education on alternating current, output
waveforms and active PFC power supplies just to be able to ask this
question.

It all started because I heard that new Dell PCs need a new kind of UPS
(uninterruptible power supply.)

It's a fact: To comply with regulations, Dell now includes something
called an active-PFC power supply with their PCs. I also understand that
users and power companies benefit from PFC, as does the environment.

The problem is, active PFC power supplies supposedly need current with a
pure sine wave output - and a Dell technical rep told me that all new
Dell computers need this - but virtually all UPSes for home use have a
stepped approximation of a sine wave.

Has anybody priced a UPS with pure sine wave output? Those buggers are
expensive!

On the other hand, according to APC: Starting in mid-2008, all APC
Back-UPS (home and small office) products were revised to better handle
load requirements for PFC devices...allowing the majority of Back-UPS
being manufactured now to handle most (not all) PFC loads within their
wattage range much better than older units. Their output is still
step-approximated, however, and APC still finds the occasional PFC
device that they have trouble with.

Here's where I throw up my hands. Do I really need to pay an
arm-and-a-leg for a pure sine UPS, or is most any properly sized UPS
'good enough'?

Daddy

 

Not good enough. For the XPS 435T/9000, at least, it absolutely won't
work with a stepped approximation. You really do need a true sine wave.

Tom Lake

 

In fact, I just finished a 'chat' with a tech from APC. They recommend a
pure sine wave UPS for the Studio XPS 9000, and I think my 8100 is
similar enough for the same recommendation.

Even APC's least expensive sine wave UPS (from their Smart UPS line) is
more than a third of what I paid for my PC in the first place! And
although I haven't looked at other makers, I can't imagine that their
sine wave UPSes are so much less expensive.

This sounds to me like a big consumer backlash that is waiting to
explode. Dell does not warn potential customers: "Attention - If you buy
this PC you will also need an expensive UPS to protect it." And it's not
only Dell's problem...any new PC with an active PFC power supply is
going to need an expensive UPS.

I can buy a new PC for the price of a properly-sized UPS for my system.
This whole thing is just too surreal. The UPS makers must be drooling.

Daddy

 

Thank you once again, William, for your comprehensive and informative
reply. Please see my reply to Tom Lake. Like you, I find it hard to
swallow that new PCs *need* this expensive type of UPS.

Daddy

 

Hi!

No. A *properly designed* power supply will work with a UPS that has a
stepped approximation to a sinewave output. I've done this often
enough with various power supplies that have active PFC circuits to
say with reasonable assurance that it will work.

I maintain that Dell is cheapskating on the quality of their power
supplies, and they may have a nasty backlash from doing so. (But then
again, I'm a pretty big believer in "do it right the first time" for
stuff like this.)

There's also the PS/2s...I have a herd of those and some are hooked up
to UPS units with stepped approximation to a sinewave output.
Sometimes the power supply harmonizes a little bit when on battery
power, but they work. The late model 400 (actually 399.1!) watt power
supply has an active PFC circuit and it seems to get along fine on
power from the average UPS. (Some people do report power ups and power
downs, but I have never been able to duplicate this.)

It's not cheap or particularly easy to design a UPS with true sine
wave output. What an AC generator does just by virtue of its
mechanical design a UPS must do electronically. The complexity of the
circuit is much greater--if you ever get a chance, compare the two
designs sometime.

I couldn't think of the term last night, but a lot of really old UPS
designs put out a simple squarewave and don't even bother trying to
make it look like a sinewave in any way.

There were some exceptions--I had a very nice (and old!) APC 1200VA
UPS with true sine wave output. A lot of BIG UPS units have true sine
wave outputs. Here comes your money saving tip--even today you can
find these and they are typically priced on a "if you can carry it you
can have it basis". That is because the batteries in them will be
totally dead. Good quality car batteries or deep cycle marine
batteries can usually serve as replacements, although you may have to
buy a battery box and wire up external cables.

That was exactly what I did with that APC unit. It took 48 volts worth
of input power to its inverter, so I wired up four big car batteries
in series. The thing would run a moderately sized 120 volt room air
conditioner or washing machine, although it had a hard time starting
either one.

The same thing works with many smaller UPS units as well, although
sometimes the charging circuit flips out and won't do its job. The
older APC UPS units sold from the mid to early 90s handle a battery
retrofit perfectly. I'm using several that have been converted to
operate from a lawn tractor battery. One of them is at the heart of a
backup lighting system--it is connected to several CFLs. These sit
idle until the power goes out--at which point the UPS goes to battery,
a contactor with a normally closed contact goes closed and the backup
lights come on. With a few 60W equivalent CFLs, the whole shebang runs
for at least two hours with ease.

Of course, if you do this you have to be wary of the fact that you're
dealing with high current circuitry that really won't understand how
you were just curious and wouldn't touch THAT again! Even one car
battery contains enough stored energy to weld, cauterize, start big
fires, etc. You must be sure that any wiring you do is safe and will
stand the load placed on it.

Oh, and charge sealed lead acid batteries in a decently ventilated
area only. Otherwise explosive hydrogen gas would build up.

William

 

Very interesting, William, as usual. But a lawn tractor battery is not
an option for me. Nor can I imagine how I would explain it to my wife.

Daddy

 

I have made my decision, based on further research.

The issue concerning a UPS and a power factor corrected power supply
arises when the power supply recovers from a temporary power shortage.

A PFC power supply can have a high "inrush current" -- the amount of
current it draws when first switched on. A highly efficient power supply
(with apparent power approx. equal to actual power) can draw as much as
100% of its full power for an extended period of time (approx. 50 ms.)

A UPS that delivers its output power in a sine waveform is more likely
to provide power for long enough vs. a UPS with a stepped approximation
of a sine wave.

Another factor in UPS selection is the transfer time. A PFC power supply
can request power from the UPS in as little as 4 milliseconds. If the
UPS has a transfer time longer than 4 ms, it may not be able to provide
the power supply with power in time. (This entire discussions does not
apply to an "online" UPS.)

It's probably impossible to come up with accurate measurements for the
actual inrush current and minimum transfer time of a given power supply.
So, practically speaking, the only way to know for sure whether your
computer is adequately protected is to 'pull the plug' from the wall
socket. That's what I plan to do tonight.

Daddy

 

I maintain that Dell is cheapskating on the quality of their power

Not necessarily. The PS in the 9000 is rated for 475 Watts and puts out a
true 475 Watts. I've seen some name brand supplies put out less
than their rating.

Tom L

 

Well, I conducted my little experiment...pressed the power button on my
UPS to turn it off, and...my computer and monitor immediately shut off.

That's not necessarily because my UPS doesn't produce a sine waveform.
For one thing, at 550 VA, my UPS is way under-powered. For another, this
sissy little UPS probably doesn't have a fast enough transfer time.

By the way, don't waste your time asking Dell technical support about
power supplies or UPSes. They don't have a clue.

Daddy

 

Hi!

That's not quite what I'm getting at. How well filtered is the input
for the supply? How well will it continue to operate under stressful
conditions (like a low line voltage)?

Meeting the published specifications for output is just one of many
criteria. And while most of Dell's power supplies are of good quality,
a comparison to something like a PC Power and Cooling unit would make
them look cheap and maybe even a little flimsy. Given that Bestec
supplies are showing up in some low end Dell boxen, I'd not be
surprised to learn that even the higher end machines are having
quality cutbacks here.

Sure. And there are always dishonest no-name makers to worry about as
well:

http://greyghost.mooo.com/psuthoughts/

That's the first of two that I've seen. After finding another one (in
another eMachines computer), I see that whoever made it did do some re-
engineering, mostly accomplished by adding more metal to the
heatsinks. I guess they hoped this would make the supply live a little
bit longer.

The second one wouldn't even produce enough power to let the machine
operate correctly, and I'd bet it wasn't pushing a load of more than
perhaps 80-100 watts across all of its outputs.

All problems disappeared upon replacement of the power supply with an
honestly specified AGI 350 watt unit. The owner claimed that no one
except Best Buy had ever serviced the machine. I don't have a high
opinion of their service operations, but I'd hope that they would at
least use decent quality, honestly specified replacement parts.

William

 

Hi!

Actually, it sounds like it's because you turned it off.

You should get a bigger one. An overloaded UPS won't always take it
gracefully. I've seen overloaded UPS units go off with a bang,
especially the cheaper ones.

The power supply filter caps should bypass that problem, as they
usually hold enough energy to see the power supply past little dips
and sags. They only manage to hold about a second or so's worth of
energy most of the time.

William

 

This gives me an opportunity to ask this question:

Since the 9000 has a PFC power supply - even though a Dell tech support
rep assured me that "no Dell computer have PFC supply" - does that mean
that its 475 watt power supply really delivers a maximum of 475 watts?

Daddy

 

Hi!

That's not what PFC does. PFC tries to compensate for the fact that
the components inside a power supply (particularly the capacitors and
inductors) may mess up the relationship of voltage to current in an AC
power waveform. In other words, your computer's power supply may be
"polluting" the power line. With resistive loads (like heaters and
such) the voltage and current are in perfect synchronization with one
another, giving these devices a power factor of "1".

Depending upon how it is loaded, the power factor of your PC's power
supply can change. A passive PFC circuit performs a fixed amount of
power factor correction and does not adapt to differing load
conditions.

Active PFC works by adjusting the level of power factor correction
based on how hard the power supply is working. The idea is to get
things as close as is possible to a power factor of 1 (the perfect
value).

Whether your power supply will meet its listed output is another
matter entirely. The ratings of each major component--the switching
transistors, rectifiers, transformers, capacitors and other stuff--
must be considered. There must also be sufficient cooling for these
parts, especially the rectifiers and switching transistors. Otherwise
they will go "boom".

See my other posting for a comparison of two other power supplies, one
of them dishonestly specified. You can easily see how the honestly
specified supply has much better build quality--it can more easily
dissipate heat and the major components are larger, serving as a rough
indicator that they are capable of more work.

William

 

Hi!

The general rule is to multiply the current rating printed on the
nameplate by three. That's your inrush current figure. As that rating
is frequently a worst case rating, for a fully loaded machine doing
everything at once, you'll have some headroom here.

If you could find a meter with a sufficiently fast response time with
a "peak hold" function, that could be used to figure it out at as
well.

A UPS is seldom if ever going to be powering your system from a true
cold start (where inrush current would be an issue). Every ATX power
supply I have ever seen keeps a small portion of its circuitry alive
when plugged in (and the switch at the back, if there is one, turned
on). The main filter capacitors are also usually kept charged in this
case. So you really only have to be concerned about the average amount
of power your system is using.

There are inexpensive plug in meters that can tell you all of this
quickly and easily. Most have multiple scales of measurement (watts,
amps, line voltage, volt amps) that are reasonably accurate. The P3
Kill-A-Watt is one such meter, and it is readily available. One of
these can be a very worthwhile investment. Then figure about 20-30%
headroom on top of that for your UPS capacity.

The way I'd do it is to plug everything into the meter, power up all
the equipment you want the UPS to protect, do something that runs your
CPU usage up to 100% for a few minutes

You should test your UPS periodically. Better units have a button to
press for testing (a feature curiously absent from the plug strip
types sold by APC, Tripp-Lite and Cyber Power) and even better units
conduct timed self tests on a repeating schedule (usually every two
weeks). Those that don't have a dedicated "push to test" button can
sometimes be tested by pressing and holding the power button for a few
seconds.

Although I have never had a problem doing so, pulling the plug on your
UPS to test it can result in the grounds for each piece of computing
equipment entering a "floating" state. This does assume that the
grounds are properly connected when the UPS is plugged in, as they
should be. (This means that if your UPS has its "site wiring fault"
light on, you need to find out why and have the problem fixed.)
However, it is a guaranteed test.

William

 

Hi!

(oops!)

...., turn on all peripheral devices and take a reading with your meter
set to report volt-amps (because that's how UPS units are rated).

Don't put your printer on the UPS. Although inkjet printers are
unlikely to cause trouble, laser printers will. The fuser demands an
enormous inrush current whenever it fires, and this will overwhelm the
inverter in all but the very largest UPS units, causing it to shut
down (best case) or fail in an exciting way.

William

 

I think you are deescribing the"Kill-A-Watt".
http://www.amazon.com/P3-International-P4400-Electricity-Monitor/dp/B00009MDBU

I know of a case where a UPS wouldn't supply juice if it wasn't
plugged into the utility mains. It was a "field day" exercise and the
intent was to power a WiFi repeater off the grid.

 

I have learned several things from this entire exercise:

1- Dell technical support reps don't know what a PFC power supply is,
and aren't sure whether any Dell computer has one.

2- Dell technical support reps don't really understand the published
specifications for the computers they are supporting.

3- None of the big UPS makers will come out and say: You can't use our
home and small office UPSes if your computer has a PFC power supply.

4- None of the formulas, online calculators or techniques to calculate
or estimate the correct size UPS for your computer arrive at the same
conclusion.

5- Nobody agrees on the correct method to calculate or estimate the
correct size UPS.

6- No OEM will tell you: If you are thinking about purchasing this
particular computer, consider this: It will require a UPS that will cost
you half as much as the computer itself for adequate power protection.

7- No UPS maker will say whether they'll honor their warranty if you use
a stepped approximation unit with a PFC power supply.

I apologize for ranting.

Daddy

 

Hi!

I am not sure why that comes as a surprise. To them, a power supply is
a silver box with wires. They might know it has a fan or two inside.
They also know that when power supplies in computers fail, you send
out a new replacement.

Few people do component level repair on power supplies. I've been
incredibly frustrated in my attempts. The only reason I tried is
because the supplies in question aren't made any more, and I've been
losing them at a rate faster than I can replenish my supply. (In case
you're wondering, the supplies are the 335 watt Delta SMP-332AB units
from IBM PS/2 Model 95 computers. All I know about the failure is that
the controller is being told to shut down, and that was what someone
much better equipped than myself and more experienced had to say about
it.)

And it's dangerous--switching power supplies can supply enormous
amounts of electrical current, the high switching frequency can really
bite if you happen to come in contact with something operating at that
frequency and portions of the circuitry are directly line connected.
That last one makes for an extremely dangerous environment, as the AC
power in your home can deliver more than enough current to seriously
injure or kill you.

That's because it certainly *should* work. If I ever come into one of
these Dell machines where the power supply doesn't "like" a stepped
sinewave input, you can bet that I'll be doing a teardown.

I'm deadly serious:
http://greyghost.mooo.com/hdtunetakeapart/
http://greyghost.mooo.com/rx4109takeapart/
although I usually don't go to the trouble of making web pages, or
even ones really intended for public consumption.

Anyone want to send me one of the PSUs for an exam? Private e-mail can
be found from the above pages or wct <atsign> walshcomptech <dot> com.
I'll return it to you fully reassembled, and it's unlikely to be
damaged in the process because I take extreme care when doing these
things. I will test it prior to returning it, and should it not work,
I'll front the cost for a replacement.

(Yes, that's only my word, but it's all you need as I take it quite
seriously.)

Couple of explantions: 1) there's more than one way to skin a cat, 2)
everyone has a different idea, 3) some are more truthful than others.
I'd put more faith in a calculator constructed by someone who *uses*
UPS units than I would someone who is *selling* them.

See above.

It'd cost them the sale, and maybe in the worst case, even land them
in a court of law to defend their actions. When a computer from the
competition can be proven to operate perfectly on stepped sinewave
output, why won't theirs?

Not saying that's correct, but you can be sure it's how the average
consumer is going to view it.

If the warranty to which you refer is the one that covers the UPS
itself against any failures or issues, they don't really have a choice
outside of snubbing the customer or coming up with an allegation that
you were abusing the unit. There are some pretty strict laws governing
what can and cannot be excluded or limited in a warranty, at least in
the US. It's unlikely that the UPS would be damaged even if the
computer failed to operate correctly from the power supplied by said
UPS.

As for any such "connected equipment warranty", these are practically
worthless. You should not buy power protection equipment based on what
any such policy might claim. I suppose that someone, somewhere has
received a payout from one of these policies, but when you think about
it, only a fool provides a multi-thousand dollar warranty for the few
tens or hundreds of dollars that represent the profits from the sale.
If you haven't, look at the fine print in one of those things
sometime.

No problem. Sometimes you just gotta speak your mind.

William

 

Daddy,

Great summary. I'll wait for your recommendations.

 

The specs for my 6-month old Precision 3400 included the sine wave UPS
recommendation. While the 375W PS is smaller than many (yet more than
sufficient for my needs), my BACK-UPS ES 750 has always worked
flawlessly with this machine. Just one user's experience -- YMMV, of
course!