Power factor, UPS, computer

[Uncle Peter]

I was wondering why this 2nd hand UPS I bought says 1500VA, 980W. Are they assuming you don't have decent PFC on your computers? When did they start putting decent PFC in computers?

I tested (with one of those plug in energy meters) a computer running a good graphics card and CPU flat out with a cheap power supply (CIT) with "passive power factor correction" (along with a stereo playing loud and two old LCD screens), as this is where the UPS will be used. The meter indicated a PF of 0.71. I then tried a computer with a Corsair power supply with "active power factor correction", running a dual chip graphics card flat out, and got a PF of 0.98.

[Uncle Peter]

This answer from Yahoo answers seems correct:

"This is of the specification written by UPS manufacturers is widely misunderstood by users. The actual load that can be supported can be up to 1500 VA OR 980W. These figures MUST NOT be exceeded. Power factor is irrelevant. Thus with a PF of 0.98 the load can be 980W/1000VA (both figures are within the limits above, limited by W) With a PF of 0.5 the load can be 1500VA/750W (limited by VA) Battery standby time is determined by the W (watts) of the load. Thus if it will give 5 mins at 980W you can approximate that it will give 10 mins at 490W."

Presumably the invertor has a wattage limit and the transformer has a VA limit.

[Andrew Gabriel]

Depends on the inverter/output design of the UPS.

Also depends on the nature of the low power factor (i.e. just because the load is taken only at the voltage peaks as with switched mode PSU, or because the load draws more than it needs in part of the cycle and gives some back to the supply in another part of the cycle, such as a reactive load).

[Uncle Peter]

So I guess the only answer is to make sure you exceed neither (although they do have cutouts). Mind you it looks like unless you have a very low power factor (below 0.67), you can't ever exceed the VA rating, so just checking the wattage suffices.

What's the lowest power factor you can have? If you connected a pure capacitor across an AC source, is that a PF of 0?

Or a more sensible question - what's the lowest power factor any real device that you'd link to an AC source has?

[harryagain]

All transformers are rated in VA (volts x amps) And other electronic voltage conversion devices.

Watts = VA x power factor

With resistive loads the power factor is unity (1) With other (inductive) loads the power factor is less than unity.

0.71 is a "poor"power factor that would be not really accepptable for any signifcant load.

[Uncle Peter]

Agreed, although computers aren't really a significant load, and even if they were, the expense of correcting it power factor may not be worth the effort. I find it odd that they bother in the likes of Corsair desktop PC power supplies.

And if you think 0.71 is not acceptable, why are UPSs designed to allow for it?

[harryagain]

Either a pure capacitor or a pure inductor would give a power factor of 0 No power (watts) would be consumed in the above cases. They are energy storage devices, they take in energy and then put it back resulting in a phase shift between voltage and current. Inductors cause the current to lag voltage. Capacitors cause current to lead voltage.

(This is a purely AC thing BTW)

Virtually pure capacitors exist in practice but not pure inductors.

[Capitol]

IIRC there is a standard requiring PFC for any item consuming over 100W.

[Uncle Peter]

What's the lowest PF you get with usable equipment?

[Uncle Peter]

According to Wikipedia it can go to -1?

[Uncle Peter]

Clearly not, or CIT would have got into trouble.

[Andrew Gabriel]

Most UPS's only care about the VA load. I suspect that's true of yours too.

If you have a large reactive load with low power factor, you'll need a UPS which can take back reactive energy, and that means a flywheel motor/generator converter, where the flywheel can store the energy given back to the supply, so it isn't lost.

Yes.

0.5 is probably the lowest I've come across for sizable loads, but some very low power devices with capacitor droppers can be lower still.

[Andrew Gabriel]

It's in one of the EU documents. It varies by type of equipment. For electronic lighting ballasts, the limit for uncorrected PF is 25W. (This is probably why many CFL manufacturers stop their ranges below 25W.)

[Uncle Peter]

Do you know where this document is, and why they made the rule? Also is it law?

[Farmer Giles]

How very 'CIVIL' of them. :-)

[Andrew Gabriel]

Andy Wade is the person here most likely to know which EN it's in.

It was introduced because increasing use of uncorrected switched mode PSU's which draw current only at the mains voltage peaks in more and more equipment was having a noticable distorting effect on the mains sine wave. This causes problems for power generation and distribution, and unlike reactive power factor correction, it cannot be corrected by capacitor banks.

I used to work at Sun Microsystem's Bagshot site where we had a large computer room of all the older computer equipment used for customer support purposes. The power draw was 2MVA, and the distortion of the mains waveform I measured on a scope was very noticable:

[Stephen]

This may be an urban myth, but I was *informed* that kilowatt hour meters can under record the true energy consumption significantly if the total power factor of the devices was less than 0.85.

According to my source, its to do with the voltage being a sine wave, the current being a sine wave an dthe angle difference between them.

Power is the product of the sine waves of current and voltage. so essentially P = V Sin Theta * I Sin Theta at any point in time for a PF of 1.0.

Apparently with a power factor of 0.85 or less, although the voltage and current waveforms are still the same, the fact that they have a much wider angle difference so P = V SIn Theta * I Sin (Theta - angle difference) and apparently the electricity meter starts recording imaginary power rather than real power.

My source also informed me that allegedly that by simply adding a suitably rated capacitor to your consumer unit, the PF could be changed by 90 degrees which them made the meter under read.

My source also informed me it is for this reason that automatic power factor correction equipment is often added at factories that run a lot of induction furnaces or motors to make sure the meters record the true energy consumption correctly.

[Stephen]

and my source also indicated that Power factor correction is also done to reduce instabilities in the national power grid and to reduce surges, brownouts, spikes and interference as well.

Apparently circuit breakers can trip if the power factor becomes too low.

[Uncle Peter]

That doesn't look too bad to me. I suppose eventually something that requires the peak to be high will fail?

It presumably is not a hard and fast law or I wouldn't be able to buy a non reactive PFC power supply. Almost all cheap PSUs (as in not big brands like Corsair, etc) do not have reactive PFC.

[Uncle Peter]

I prefer to understand why they do it, than just remembering parrot fashion.