Active or Passive PFC?

[limee]

So is the general thought that for home use, a passive PFC power supply will save more $? (miniscule as it is)

In theory which would actually save more electricity? Not including power line feedback and such...

[larrymoencurly]

I thought that you'd save $$$ (or the equivalent in Euros) only if your utility charged you for reactive power (the kind that PFC tries to reduce) instead of just for real power. In the U.S., I believe that only commercial buildings, like factories, are charged for reactive power, and homes are not.

[lm]

Most environment friendly would be active pfc. Probably doesn't affect your electricity bill. The efficiency of the psu is more important than the power factor.

[davide_casarin]

my problem is to maximize blackout uptime. I have a passive pfc psu: what would happen if i had the same psu and pc but active pfc? would it last 60 % more ? it's all about the effects of a poor power factor to the energy balance... any ac guru out there?

[zuperdee]

my problem is to maximize blackout uptime. I have a passive pfc psu: what would happen if i had the same psu and pc but active pfc? would it last 60 % more ? it's all about the effects of a poor power factor to the energy balance... any ac guru out there?

As I understand it (and someone please correct me if I'm wrong), a higher power factor DOES mean lower effective AC current use. However, it makes no difference whatsoever to the actual CONVERSION EFFICIENCY of the PSU, which is the main thing that determines its heat output.

[MikeC]

larrymoencurl, lm and zuperdee are all correct. I think they've covered it all.

To emphasize: PF has nothing to do with AC/DC efficiency. The latter is what would affect davide_casarin's "blackout uptime" -- if you are talking about how long the PC would run off a UPS when Ac power is out? Not sure if that's what you mean.

[bomba]

My limited understanding is that non PFC switchers have very high short-duration current draw that taxes power generation equipment. Electrical utilities don't like switching supplies because they must oversize their generation equipment to meet the high peak power demands yet (billable) energy consumed is proportionately low. Although, I'm not 100%, I suspect that a PFC supply would give you more blackout UPS time.

[MikeC]

OK, I must have had a mental blackout there...

IF we are talking about UPS, then bomba is indeed correct. An Active PFC PSU will have something close to 1.0 PF. This means the WATTS is almost equal to Voltage x Amperes all the time. When you have a low PF, like 0.6, that means when your power draw is 60W, the V x A = 100. UPS units are rated in VA, not in watts (AFAIK) and a APFC PSU has lower VA than a non or passive PF PSU, so the "blackout uptime" would indeed be higher.

Thanks for the brain jog, bomba.

[zuperdee]

The question of power factor often seems to come up on these forums, so I just thought I'd try to see if I could explain it in plain English for everyone. Please be aware though that it is a VERY complex subject, and is probably best understood by those who are mathematically inclined. (Which I am not, sadly.) And please people, feel free to correct me if I am mistaken about any of this--I say this because I am merely a computer science person whose interest in electrical things is for hobbyist purposes--I am NOT an electrical engineer, so my depth of understanding is probably very limited.

At http://www.smpstech.com/tutorial/t01int.htm, you will find a great introduction to switching-mode power supplies. Their definition of power factor:

"The ratio of total active power to total apparent power in volt-amperes in an ac circuit, where voltage and current are rms values and include the effects of harmonics as well as the effects of phase displacement. If both voltage and current are sinusoidal, power factor is the cosine of the angle between them."

At http://www.bpa.gov/Corporate/KCC/defn/defnsmal/pq.htm, I found this definition of it:
"The ratio of power in watts to the product of volts times amperes. The power factor of an alternating-current transmission system is unity when the voltage and current are in phase. There is no power factor with direct-current power."

For those of you who aren't mathematically inclined: "unity" basically means the number 1. In other words, the power factor is 1.0 when the voltage are current are perfectly "in phase," or saying it another way, when the cosine of the angle between them is ZERO.

Now, as we know, with alternating current, the ideal current waveform is a perfect SINE WAVE, meaning that the crest factor (i.e., the ratio of peak instantaneous power to the average power) is 1.414. To understand what this means, you need to understand that what we normally call "120 volts AC" is technically 120 volts RMS (or root-mean-square). The PEAK voltage with 120 volts RMS AC, assuming the AC signal is perfectly sinusoidal, is roughly 170 volts.

The problem with things like computer PSUs is that without some form of Power Factor Correction, the current draw of the PSU may NOT necessarily be "in phase" with the voltage. At http://www.smpstech.com/tutorial/t01int.htm, they define power factor correction as:

"1) Technique of increasing the power factor so that the phase angle between the voltage and current approaches zero in an ac circuit. 2) Addition of capacitors to an inductive circuit to offset reactance."

Note that definition #2 is actually the definition of PASSIVE PFC. Active PFC is much more complex.

In other words, without some kind of power factor correction, the current draw of the PSU may NOT be "in phase" with the voltage it is being supplied. A perfect "in phase" current draw is normally achieved with simple linear resistive loads like tungsten light bulbs. With these kinds of loads, their current draw will vary directly with the voltage supplied to them. With this kind of a load, the current draw should also have a CREST FACTOR of 1.414.

But with PSUs, their current draw is NOT totally linear. Their resistance might be lower at some times and higher at others. With a situation like this, it means that your utility company must use a generator that is big enough to supply your PSU's PEAK demand, even though it isn't actually using that much power on AVERAGE.

So, to keep things simple for our purposes, just understand that THE LOWER the power factor, the LESS EFFICIENTLY it uses the GENERATING CAPACITY of your utlity. In other words, even though your PSU may require only say, 700 watts on AVERAGE (i.e., say over 2 complete AC cycles), but requires requires a 1500 watt generator to supply the PSU's PEAK demand moments, then this discrepency between the AVERAGE and PEAK load moments means that the usage of the utility companies' resources in not very EFFICIENT, since there are large moments of time when its PEAK output is not being USED.

[zuperdee]

Here's an analogy that might help a bit more:

Supposing you had a 1000-watt light bulb, and you rapidly switched it on and off for equal amounts of time over the course of an hour. Obviously, your utility would have to be capable of supplying 1000 watts, even though they'll never be able to charge you for it, since you had it on only half the time, and thus used only 500 watt-hours.

Now, if you had an electronic circuit that switched the lamp so fast that you didn't notice the lamp was switching on/off, you could pat yourself on the back, claiming that you'd just invented a way to lower your electricity usage by half, and you'd be right--you'd have a lamp that effectively gave off only 500 watts of light!!!!

But to your utility, they wouldn't be so happy, because they'd still have to be supplying 1000 watts during those moments when the light was on!!!!

Computer PSUs ARE switching devices, just like the switch light bulb analogy. They switch for a different reason, but that's another matter. The main point is that those PSUs with some kind of power factor correction will act more like simple resistors--that is, with PFC, they effectively try to make it look like they're drawing 500 watts all the time, rather than 1000 watts half the time.

[dukla2000]

Apart from the impact on your electricity generator coverred above, to me there is 1 reason I will never again buy a passive-PFC psu.

The normal/cheap/easy way to make a passive-PFC psu includes adding a (physically) large choke in the AC circuit. I have had 4 passive-PFC psus, 4 different manufacturers, all have a similar choke. And the thing looks like a transformer, it 'buzzes', because of its size it is bolted to the psu case, and the psu case vibrates = loud buzzing. Despite serious attempts on 2 of my psus to resolve the buzzing I had no success, except 1 of them kindly expired (possibly as a result of my exertions?) so I could bin it.

My recommendation is, if you are in the market for a PFC psu, at least for noise reasons get active-PFC.

[zuperdee]

The normal/cheap/easy way to make a passive-PFC psu includes adding a (physically) large choke in the AC circuit. I have had 4 passive-PFC psus, 4 different manufacturers, all have a similar choke. And the thing looks like a transformer, it 'buzzes', because of its size it is bolted to the psu case, and the psu case vibrates = loud buzzing. Despite serious attempts on 2 of my psus to resolve the buzzing I had no success, except 1 of them kindly expired (possibly as a result of my exertions?) so I could bin it.

LOL!!! Interesting--I've never heard of this buzzing problem before. The reason the choke "looks like a transformer" is probably because, for all practical purposes, it *IS* basically a single winding transformer.

I would have thought though that the better way to do Passive PFC would be to use a heavy-duty capacitor. Personally, I have NEVER seen a PSU with a heavy choke bolted to the PSU casing like that before--I'm not sure why.

[dukla2000]

Personally, I have NEVER seen a PSU with a heavy choke bolted to the PSU casing like that before--I'm not sure why.

Are you in America? My suspicion is because there are so few (relatively) active or passive-PFC psus there.

Try this pic for what I mean.
The newer chokes (like the one pictured) that have the thick plastic base are much better than the earlier ones that had metal mounting points, but still not (IMHO) good enough for silence.

Picture borrowed from NB-Fortron UltraSilent 350 Watt review at au-ja!

[zuperdee]

Are you in America? My suspicion is because there are so few (relatively) active or passive-PFC psus there.

Yep, I am in America.

The newer chokes (like the one pictured) that have the thick plastic base are much better than the earlier ones that had metal mounting points, but still not (IMHO) good enough for silence.

I can imagine. hehe.

Like I said, I would think that the better way to do passive PFC would be with a capacitor instead of a choke. (And most technical descriptions I've read describe the use of a capacitor for this purpose.) Like I say though, I'm not an electrical engineer, so I don't really know why they'd do it with a choke.

In ANY case, there's clearly no question in my mind that Active PFC is the way to go, if you can afford it. Personally, I seriously doubt that we'll see regulations requiring Active PFC here in America ANYTIME soon. However, I also think it might not be necessary, because I think natural market forces will probably be making us see more Active PFC PSUs here in the future, anyway.

[davide_casarin]

Thanks zuperdee for his explanation and links, after much mumbling and reading I agree with him about PF.
The phase displacement between current and voltage is only confusing in our switching psu world, it makes sense in traditional inductor-resistor-capacitor-genrator circuits where you can actually see a sinewave current displaced. Here we have a ugly switching game and the resulting current wave is square and irregular, probably does not even have a fixed frequency, it's daring to call it displaced sinewave.
About capacitors to smooth it, I have tried many times to do similar things, and the result is that capacitors are always way small, so only high frequency and/or low currents can be smoothed. I have heard that they use a capacitor in parallel between car battery and amplifiers to compensate short high spikes in current drain: surely we are talking about good currents, but those barrels are said to be 1 or 2 liters big, and expensive.
About humming chunks, my psu is passive PFC, does have a metal inductor and 2 large capacitors, but the noise it makes is audible only with your ear pressed against the psu case so i don't bother.
To sum it up, because of my bad PF i just have to provide a more-than-necessary powerful UPS and the time I last without AC depends only on efficiency and system drain ? Thats fine then.
About power companies having to deal with our switching spikes, when you feed 1000 PSU their spikes are randomly spread, then add some inductance and capacitance from their lines, transformers and other equipment, and they should not end up with a very different drain than if everybody had a big resistor instead of a pc.

[zuperdee]

To sum it up, because of my bad PF i just have to provide a more-than-necessary powerful UPS and the time I last without AC depends only on efficiency and system drain ? Thats fine then.

That's interesting--I never thought of this before!! In other words, a bad PF will mean you need a larger UPS than would otherwise be necessary, just to supply the peak current needed by your PSU.

This is a VERY interesting insight. I don't know why it never occurred to me before--my UPS used to be enough for my previous computer... Then, when the power went out with my new computer once, my UPS made a sound, complaining that it was being overloaded. I thought my UPS was going bad, but maybe it was because of my PSU's bad power factor!!! Thanks for the insight!!

[MikeC]

Re - PSU VA and UPS - I think I've mentioned this here before, but not made a big deal of it. Seasonic does make a big deal of it in their marketing materials. For large numbers of PCs (in offices) it makes a huge difference in both electricity consumption and UPS costs.

UPS units are rated in VA, and it is this number that's relevant. To obtain maximum VA for any PSU, calculate the AC power draw from the efficiency, then divide the AC power by the Power Factor.

Example:

A 300W PSU with 70% efficiency draws a max of 428W from the AC line. Conside the effect of PF on VA --

428W divided by .6 (typical non-PFC) = 713VA
428W divided by .8 (typical passive-PFC) = 535VA
428W divided by .95 (typical active-PFC) = 450VA

Pretty dramatic, isn't it? The cost of a 450VA UPS is obviously much lower that that of a >700VA UPS.

How about an 80% efficient 300W PSU? It draws 375W in AC --

375W divided by .6 (typical non-PFC) = 625VA
375W divided by .8 (typical passive-PFC) = 467VA
375W divided by .95 (typical active-PFC) = 394VA

The results speak for themselves.

NOTE: Keep in mind that in typical usage, most systems draw only 100~200W in AC.

[davide_casarin]

Re - PSU VA and UPS
For large numbers of PCs (in offices) it makes a huge difference in both electricity consumption and UPS costs.

Difference in UPS cost, ok, but if it also makes difference in electricity consumption we are back at the starting point.

We disturb and test the compensating abilities of ups inverter or power company alternator, but that's not a power drain.
I see it like a guy on a bike trying to go at an exact speed (or a sinewave speed, or any feasible curve we decide is right) and me pulling back with a rope with sudden irregular pulls: being equal the work done, it's more difficult to compensate my distrb than for example a costant brake:
the latter only requires a costant +x % of biking force and after the starting period the result is perfect;
the previous one requires fast and intelligent compensation or an elastic, and still does not come out perfect. The intelligent motor would have to know in advance what kind of disturb it's going to stand, elaborate it and syncronize the correction, and it is not feasible in normal life.
Luckily the big power geneators barely feel the tiny disurb of my psu and the voltage swing does look 99% as it should, and if during a blackout my psu inverter's sinewave looks dirt with some 20+khz noise, that's an emergency situation and as long as pc stays on it's ok.
Interesting enough, the dual conversion ~~> ---- >~~ true online UPSs of big companies are said to deliver a sinewave cleaner than the outer power network: isn't it easier for their PSUs to heavily disturb their inverter thus having always an ugly ac power, rather than no UPS trying ruin the good job of the power plant alternators ?
So it looks to be right the opposite. But being the dual conversion UPS less efficient, noisy and expensive, and still bought, I likely miss some puzzle pieces.

[zuperdee]

The intelligent motor would have to know in advance what kind of disturb it's going to stand, elaborate it and syncronize the correction, and it is not feasible in normal life.

The analogy doesn't quite hold for PSUs with Active PFC. After all, with a PSU invertor, you know pretty much what its current draw pattern is like, so it is possible to compensate for it. And with power factors >0.99, it seems to me Active PFC does a pretty good job.

Luckily the big power geneators barely feel the tiny disurb of my psu and the voltage swing does look 99% as it should, and if during a blackout my psu inverter's sinewave looks dirt with some 20+khz noise, that's an emergency situation and as long as pc stays on it's ok.

That leaves out of account that if MILLIONS of people use such devices, it does tend to add up, and at that point, it poses a SIGNIFICANT threat to the overall capacity of the electric system. That's why PFC is now required in the EU. Here in the U.S., they've opted instead to try to make the grid more robust with respect to irregular loads. (How they do it without requiring greater peak capacity, I know not.)

[Nowhere_man]

I find this very interesting. I don't have a kill-a-watt meter but I do have a Maytag volt-watt meter, model no. 38519.

Quoting the manual

"The Maytag No. 38519 volt-wattmeter is made up of two seperate indicating meters combined into one compact instrument. On the left is an iron-vane-movement volt meter with an accuracy of +- 3% of AC voltage measurements. This meter covers two voltages ranges, 0-150 volts and 0-300 volts selected by a toggle switch located on the instrument panel below the meter face. The voltage ranges are color coded red.

On the right is a dynamometer type wattmeter which reads true power consumption of a circuit. There is automatic conpensation for power factor errors in AC wattages. The wattmeter features three ranges selected by a three position toggle switch below the meter face. The ranges are 0-500 watts, 0-1000 watts, and 0-2000 watts."

Obviously this is older technology than a kill-a-watt meter. I've had it laying around for a year now un-used. I rekon I'll hook it up and see what happens.

[davide_casarin]

I was referring to a generator/inverter being driven by a computer/man to compensate the remote psu noise. common pfc is at the other side.
Could it be swapped back on the line to the generator in a lab circuit ? Probably, if it's at the first stage of power conditioning.
If yes, Can we add up the noise of a city and correct it at the plant achieving a virtual 100% users apfc ?
Probably, but
1) the benefits for the net would be 0: all these squares giving a hard time to the whole net and reflecting on and off;
2) the size, cost, dissipation of this superfilter would be comparable to the plant in a way similar to how it compares in a psu;
so say it being it feasible and useful, it still makes sense to not do it and tell ppl to use millions of small ones.

About APFC, afaik that's just magic. And they say the circuit design is complex so I am excused.

Yes millions of psu should add up and at some random unlucky istants ask a strangely high power and require a larger generator, but still, no energy is lost, unless a generator consumes more fuel also during the low load istants so that the average fuel/watt is higher for an irregular load than for a plain load of the same average value, wich I don't know.

Yes I heard this EU rule too, but isn't it just PPFC mandatory? Passive isn't so appealing: 0.6 corrected to 0.65, higher cost, weight, volume, possible noise, and single input voltage.

[zuperdee]

Yes millions of psu should add up and at some random unlucky istants ask a strangely high power and require a larger generator, but still, no energy is lost, unless a generator consumes more fuel also during the low load istants so that the average fuel/watt is higher for an irregular load than for a plain load of the same average value, wich I don't know.

Well, it isn't quite that simple... But here's another analogy that might help you visualize it: supposing you had a 1000-watt light bulb. Now, supposing you were to switch it fast enough that it was on half the time and off the other half the time. Let's suppose we switch it on 500 times and off again 500 more times in a second. Doing this might effectively make your light bulb give off as much light as a 500-watt light bulb, and in fact, you would technically be drawing only 500 watt-hours in an hour. However, your utility will still have to invest in running a 1000-watt generator to supply the peaks that your light bulb switching circuit is drawing. In other words, they are having to invest in extra generation capacity that can't effectively be used, because it is supplying your light bulb with 1000 watts half the time.

Yes I heard this EU rule too, but isn't it just PPFC mandatory? Passive isn't so appealing: 0.6 corrected to 0.65, higher cost, weight, volume, possible noise, and single input voltage.

Actually, as I understand it, the EU standard doesn't even technically say ANY kind of PFC is required. As I understand it, all it effectively says is that the device must have a certain minimum PFC level. So yes, you COULD get away with the bare minimum of compliance using Passive PFC. However, I'd be willing to bet that Active PFC will become preferred, if enough people like Dukla2000 notice the extra coil hum from the choke, and if enough manufacturers start noticing the other benefits, like being able to produce a single, unified circuit for both 115 and 230-volt markets.

[Nowhere_man]

MikeC wrote

NOTE: Keep in mind that in typical usage, most systems draw only 100~200W in AC.

That is the case with my system right now. My antique meter works

It might seem a bit off topic but this discussion got me curious.

Thanks for the info guys.

[davide_casarin]

Nowhere_man, why don't you play around with yours and write here the VA and W readings ?

I don't understand what is not that simple.
Your analogy is clear, but the sizing of generator is one problem, that seems clarified, and Energy waste is another problem, wich at my present knowledge, at least on theory, does not exist.
In practice, more knowledge about % highload - lowload/average load, stabilizing systems used in power plants, efficiency of a large generator with low/instable load etc. are to be known.
That is to say, if they care to stabilize, and if they use dissipative or non disspative systems to do it.
I tend to think that some energy is wasted.
If you have to use a truck to tow a small but spiky backforce, you do waste energy compared to a small car with same average backforce, because many power losses scale with motor/vehicle size regardless of load (for example a truck tows for 1 mile at 50 mph a 0 load and uses 0.5 liters while a car does the same 0work with 0.2 liters).

[davide_casarin]

I just read this in APC website, specificating their biggest UPS:
Fully-rated power kVA equals kW:
Reduces cost by eliminating the need for an oversized UPS for Power Factor Corrected (PFC) loads.
Imput power factor correction:
Eliminates the need for an external phase-compensation unit and minimizes installation costs by enabling the use of smaller generators and cabling.

This proves that PFC can as well be accomplished at the generator side (this UPS is a in-line model, so it actually generates all the output power).
Then why don't they just do PFC at the plant or in many boxes around the net instead of making laws about having decent PF loads ? To save money ?