• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

computer power supplies are major power quality concern

J

Jerboy

Banned
I just read here about computer power supplies. In a flame of competition, the distributor and manufactures pay little attention to its quality, let alone its influence on power quality. Their concern is what lures consumers, such as total maximum wattage. Computer power supplies starts drawing current suddenly near the peak of each cycle. When you have enough computers, this can cause the top of waveform to be shaved off and become flattened. On commercial buildings, computers are connected to 120V (phase-to-neutral) connection of 208V/120V(three phase wye system). The power pollution created by computers cause so much disturbance that it overloads the neutral in some cases. For that reason, it is necessary to install oversized neutral conductor on low voltage circuits. Electronic fluorescent ballasts can be just as bad if no correction. It has not been anywhere nearly as bad as computers as lighting industry has much stricter harmonic content restriction. Ballasts maybe have 20% at most harmonic content. Computers can have as high as 150%.

PS: Have you noticed your lights are not effected whatsoever regardless of what you do at the plug at your work or school?
Their wiring is different. At our homes, the lights are often on same branch circuit as nearby plugs. In commercial setups, they have 480V/277V thee-phase wye line. Lighting is powered from 277V. The outlets comes from 208V/120V wye which is transformed from 480V.
 
That's why - at least here in Europe - the "PFC" requirement was introduced. Since Jan 1, 2002, no power supply unit may be sold that doesn't behave properly on the mains side.

regards, Peter
 
That's why - at least here in Europe - the "PFC" requirement was introduced. Since Jan 1, 2002, no power supply unit may be sold that doesn't behave properly on the mains side.

regards, Peter
 
It's too bad that here in America we haven't enacted the same Active Power Factor Correction requirement.
It seems to cost about $10 or so extra to add Active PFC to a new computer power supply. I assume this is not a popular idea in the USA since it makes a power supply cost more, and nobody is willing to pay for it if they don't have to. $10 extra per PC is a big deal to OEM's who push out thousands and thousands of computers.

For now, most people in the USA (and even us computer nuts in the anandtech forums) don't know there's such a thing available.
However, if they did know, they would realize that over the life of the PC, the Active PFC power supply will more than pay you back in energy savings. Other benefits include reduced harmonics/cleaner power, reduced heat output, and greater power capacity. All these benefits occur because the power supply operates more efficiently.

Typical cheapo power supply power factor is as low as 60-70% (even lower in some extreme cases), while a power supply with Active PFC typically raises the power factor to >95% It's a total win win situation in that you kill harmonics AND you save 25% off the amount of electricity that you pay to keep your computer running. Think of the money and energy that could be saved in companies that run thousands of computers.

I specifically look for Active Power Factor Correction whenever I buy a new power supply for one of my computers. I wouldn't buy one without it.
 
I suggest we get a big post going.. and grow it and maybe even turn it into a website.. get petitions signed and get it to congress.. Wouldn't hurt to get the idea going.. After all this would improve the power quality.. and the energy consumption in the long run. (As stated above)
 
I doubt that far east PSU makers make substantially different ones for each part of the world ...
so I guess you guys in the US get PFC already anyway 🙂 Besides, I didn't notice any price change
from non-PFC PSUs to the new ones.

regards, Peter
 


<< I suggest we get a big post going.. and grow it and maybe even turn it into a website.. get petitions signed and get it to congress.. Wouldn't hurt to get the idea going.. After all this would improve the power quality.. and the energy consumption in the long run. (As stated above) >>


noble idea, but online petitions never do anything :/

so do PC power and cooling PSU's have PFC? any way to find out which PSU's do/do not have it?
 


<<

<< I suggest we get a big post going.. and grow it and maybe even turn it into a website.. get petitions signed and get it to congress.. Wouldn't hurt to get the idea going.. After all this would improve the power quality.. and the energy consumption in the long run. (As stated above) >>


noble idea, but online petitions never do anything :/

so do PC power and cooling PSU's have PFC? any way to find out which PSU's do/do not have it? >>




There shouldn't be any direct relationship between the actual energy consumption and power factor at the consumer's end. Actually PFC power supply might use half a watt or so more, because power factor correcting device use a miniscule amount of power.

Power factor is just one of power quality concern(low power factor doesn't use conductors efficiently). Other concerns includes crest factor(causes wave form flattening) and THD aka total harmonic distortion(causes problem in commercial buildings using on-site three phase transformer, big problem when you got 1,000 computers in a building). For all normal calculations, power factor is defined cosine of degree of difference between the current and the voltage phase. With two channel oscilloscope you can observe it. If power starts at zero degree, current starts at 30 degrees, your pf is cosine 30 or 86.7%.

Why it is bad to have low power factor?

You're amperage will be higher than necessary for the given watt, thus wasting the potential of the transmission line.

A 200W load with power factor of 0.5 on a 120V: 200/120/0.5= 3.3A, 396VA. You'll need to use wiring capable of 3.3A. If the power factor was unity(1.0 or 100%) you only need wiring capable of 1.55A.



Electronic fluorescent ballasts are built similar to computer power supplies with much higher control over power quality.

The one I got laying around here has 98% power factor, less than 20% THD. For the comparision, typical computer has a power factor of 50-60%, THD 100-150%.



<< so do PC power and cooling PSU's have PFC? any way to find out which PSU's do/do not have it? >>



A PSU with power factor corrector will more than likely have "PFC" in part of its model #.
 
Jerboy:

With all due respect, I beg to differ with the follwing statement of yours:

"There shouldn't be any direct relationship between the actual energy consumption and power factor at the consumer's end.........."

I am curious as to where you got this information from, or how you got this idea.

Here is what I know to be fact:

Anytime you raise the Power Factor of a system or device, you increase effciency and save electricity. This is just straight and simple electrical theory. Yes, I agree with you that the Active PFC system does use a little bit of energy....but not nearly enough to offset the increase in efficiency and power savings that Active PFC brings about. Below, I have selected a couple choice links for your inspection from both a computer equipment manufacturer, and a computer power supply manufacturer. They both reinforce my point that Active PFC power supplies save you money by reducing energy consumption.


http://www.fortra.net/news/press_action.cfm?press_id=22 states that their new power supply for their disk array system contains Active PFC. It saves up to 25.7% on electricity compared to the power supply of the older model.

http://www.seasonic.com/support/answer_pfc.htm# gives a description of the benefits of Active PFC in their new power supplies. Among these benefits are "Reduced electric utility bills..." with typical payback less than six months to a year.


I totally agree with your description of Power Factor in that it is cosine theta of the phase angle and that it is the ratio of true power(Watts) to apparent power(VA). I also concurr that bad PF results in wasted energy and the need for oversized wiring. You also mentioned Crest Factor (which is the ratio of Peak voltage to RMS [aka effective] voltage), and Total Harmonic Distortion (which is the percent of the sine wave effected by a steady state distortion).

What I would like to point out is that all of these 3 power abnormalities are related in that they are all caused by the rapid switching of the computers power supply. When you improve one symptom (such as Power Factor), you are very likely to improve the other 2 (Harmonics, Crest Factor,....etc)

The bottom line is that Active PFC saves you money in the long run, and helps to keep your power cleaner. 😀

 
GoldMember:

The enermax AX series is the only one that I know of that contains Active PFC. If you bought one, you most likely would know it, since it would have said Active PFC on the Box.

If you still aren't sure, one key thing you could look at is whether you have a 120/220V switch on the back of your power supply. If you do have one, then you don't have Active PFC. Without getting into theory, the very design of Active PFC makes it able to accept 120-220V (or any voltage in-between) without any voltage switch necessary. 😀
 
In domestic and light commercial installations (in the UK) you are charged only for true electrical energy consumed - there is no penalty for low power factor. A PFC supply won't save any money on utility bills, on such an installation.

PSUs with corrected PF do not have a different efficiency to those supplies without PFC - so there is no significant saving in energy consumed.

There is a cost to the electricity supplier of a low PF - they need to supply cables, transformers, etc. capable of supplying the apparent load, yet they are only able to charge for the true load. The exceptions are very large customers (e.g. heavy industry) who are charged penalties if their average PF is below a set limit - also because heavy machinery tends to have a very low PF due to the motors.
 
Some heavy industries are payed by the power company to run a big synchronous motor with no load on it just to correct power factor. In past years all the inductive fluorescent light ballasts used to be the problem. Stand next to a big dry type power transformer and listen to how the sounds it makes change during the day as various types of loads that dirty up the power are switched on and off. An oil filled transformer dampens the noise a lot but it still can sing and growl loud enough to hear it.
 


<< With all due respect, I beg to differ with the follwing statement of yours:
I am curious as to where you got this information from, or how you got this idea. >>



Ask any electrical engineers. Power factor has great deal of influence in energy lost in I^2R losses in power company's distribution system, but if you're talking purely power factor, it does not have anything to do with the energy consumed by the device.




<<
Anytime you raise the Power Factor of a system or device, you increase effciency and save electricity. >>



Yes you increase the efficiency and save electricity as a whole. If every factories was installed with power factor corrector, the sum of wattage used by factories will increase due to loss within power factor corrector. The energy consumption as a whole will be reduced however due to increased efficiency within distribution network. Power factor correctors are large and very expensive. These businesses will NOT install them without a financial gratification or legal obligation. Power companies are so eager to get people to increase power factor, because it saves them a bundle in transmission facility and energy saved within. The truth is that large industries install power factor correctors to avoid power factor penalty. Large industrial facilities are assessed a "power factor penalty charge" by the utility if their building power factor is below certain limit. On the other hand if it's above certain value they might be given a credit. Utility sometimes pay for part of the installtion cost of power factor correctors. You see how eager power company is about power factor? It's not the consumers whose going nuts about power factor.

Commercial and industrial users often have customer owned transformers. Commercial users gets their service at 4160V and customer owned transformer has to drop it to 480Y/277V for ligthing and HVAC, the using Delta-Wye trasformer, 208Y/120V is also made for plug in loads and misc load using 208V. Good power factor means these transformers can be smaller=cheap to make=less space=cost saving






<< This is just straight and simple electrical theory. Yes, I agree with you that the Active PFC system does use a little bit of energy....but not nearly enough to offset the increase in efficiency and power savings that Active PFC brings about. Below, I have selected a couple choice links for your inspection from both a computer equipment manufacturer, and a computer power supply manufacturer. They both reinforce my point that Active PFC power supplies save you money by reducing energy consumption. >>



If they have to explain what "power factor" is in the article, I am afraid it is not a technical literature intended for electrical engineers. Any respectable eletrical engineers will know what power factor is without being explained in layman's words.




<<
I totally agree with your description of Power Factor in that it is cosine theta of the phase angle and that it is the ratio of true power(Watts) to apparent power(VA). I also concurr that bad PF results in wasted energy and the need for oversized wiring. You also mentioned Crest Factor (which is the ratio of Peak voltage to RMS [aka effective] voltage), and Total Harmonic Distortion (which is the percent of the sine wave effected by a steady state distortion). >>



Bad power factor resulted in returned power that you're fully credited for if you're a residential and light commercial user.

If you have a load with current 80° leading (capacitor bank and resistor in series) from voltage that draws 15A from 120V AC source, its input is 1.8KVA.
cos 80 is 0.1736, so you have a power factor of 17.36%. Capaictor returns all the power back to line, thus 100% reactive.

In this particluar load, the heater is .3125kW.
Subtract that from 1.8 and you have 1.488KVA reactive.

When you connect this load to a power analyzer you'll see:

Volt: 120V
Amp: 15A
Power Factor: 0.17
Power: 312.5W
KVA: 1.8KVA
KVA reactive: 1.488KVA
Crest Factor: 1.41(theoretical maximum)

And if you use this for an hour, you'll only pay for 0.3125kWh if you're a residential customer.





<<
What I would like to point out is that all of these 3 power abnormalities are related in that they are all caused by the rapid switching of the computers power supply. When you improve one symptom (such as Power Factor), you are very likely to improve the other 2 (Harmonics, Crest Factor,....etc) >>



The extremely high crest factor is common to all setups using full-bridge and capacitor smoothers. When you connect this setup to 120Vrms AC, you will get 171V DC, because capacitors will be charged to pk voltage amplitude. Connect a heater to it(to load it). Look at the waveform across the capacitor with an oscillscope and check the low and high peak of the voltage. It will look like a slight wave on a water surface. The highest is the peak from 120Vrms, the lowest is when AC volgage is at zero crossing. The bigger the capacitor, smaller the difference assuming incoming frequency is kept steady.

Let's say lowest was 150V and highest was 171V. Seeing that capacitor never seens below 150V, it WILL NOT start replenishing charge until you reach 150V or more. As the sine wave progresses and reaches 150V it goes whoosh! and grabs power until wave peak is reached, thus resulting in extremely high. It is ok to assume that capacitor is offline from line, except when the instantaneous voltage is in 150V to 171V range.

I therefore, believe that high crest factor is a function of high frequency switching. From all three, crest factor is the most challenging problem. From what I read on articles you linked, it seems that PFC not only corrects power factor, but tries to make its load draw current as close as possible in sinusoidal pattern, which effectively lowers crest factor since capacitor can replenish charge through out the sinewave.



<< The bottom line is that Active PFC saves you money in the long run, and helps to keep your power cleaner. 😀 >>



Power factor is not factored into utility bills for residential and light commercial users. Heck, those dial style meters can not measure power factor.

Industrial customers have more expensive digital meter capable of measuring power factor and they're much more accurate.
 
Mark R has already answered but you won't save money in a domestic situation by correcting power factor cos meters read kilowatts not kva ,i often take out power correction devices due to nuisance tripping of earth leakage circuit breakers and blown capacitors etc,big companies with a lot of motors fit capacitor banks to correct the power factor as they are charged for kva
 
Here is a link to a site showing a ( typical, I think ) circuit solution for the peak current related power factor problem:
PFC circuit link
The link shows that a high freq inductor, power MOSFET or IGBT and input controller chip are the main items added to a power supply to make it a PFC supply.

BTW, in Europe and elsewhere with ~240 volt mains, the typical inrush current will be double that of 120 volt operation. I suspect the PF problem may also be a bit more serious, too.

Some comment on the thread.

It seems to me cosine theta, 3 phase delta and wye connections and their voltages are not related to this problem. Neither is smoking neutral wires.

Cosine theta and the "old" definition of power factor is related to bulk reactive loads, usually inductive, such as motors, etc., not peak charging harmonic PF. So, hanging a big cap on the line won't help out THIS problem at all. It's different.

The power grid is all 3 phase. But, essentially all home and office power is single phase. ( large building systems excepted ) So, delta - wye whatever is not related. Although going from delta to wye or vice versa will cancel or reduce harmonics upstream, most of the lost has already occurred in the neighborhood or office single phase distribution.

In all the wiring standards I know of, the neutral is sized the same as the legs. All the harmonic current that flows in the neutral has to flow from one or the other legs. Any harmonic current flowing in BOTH legs will cancel and not flow in the neutral at all. So, the neutral can not have more harmonic current than either leg. Therefore no smokin' neutrals.

Also, the problem is not better or worse because the supply is a switch mode design. If the power switching elements were replaced by resistors or light bulbs to load the input, the effect remains. The culprit is the diodes charging large caps with high current in a short time every half cycle. The supply then draws line current in short, high current pulses. Thus the high harmonic current and higher losses upstream.
 


<< Here is a link to a site showing a ( typical, I think ) circuit solution for the peak current related power factor problem:
PFC circuit link
The link shows that a high freq inductor, power MOSFET or IGBT and input controller chip are the main items added to a power supply to make it a PFC supply.

BTW, in Europe and elsewhere with ~240 volt mains, the typical inrush current will be double that of 120 volt operation. I suspect the PF problem may also be a bit more serious, too.

Some comment on the thread.

It seems to me cosine theta, 3 phase delta and wye connections and their voltages are not related to this problem. Neither is smoking neutral wires.

Cosine theta and the "old" definition of power factor is related to bulk reactive loads, usually inductive, such as motors, etc., not peak charging harmonic PF. So, hanging a big cap on the line won't help out THIS problem at all. It's different.

The power grid is all 3 phase. But, essentially all home and office power is single phase. ( large building systems excepted ) So, delta - wye whatever is not related. Although going from delta to wye or vice versa will cancel or reduce harmonics upstream, most of the lost has already occurred in the neighborhood or office single phase distribution.

In all the wiring standards I know of, the neutral is sized the same as the legs. All the harmonic current that flows in the neutral has to flow from one or the other legs. Any harmonic current flowing in BOTH legs will cancel and not flow in the neutral at all. So, the neutral can not have more harmonic current than either leg. Therefore no smokin' neutrals.

Also, the problem is not better or worse because the supply is a switch mode design. If the power switching elements were replaced by resistors or light bulbs to load the input, the effect remains. The culprit is the diodes charging large caps with high current in a short time every half cycle. The supply then draws line current in short, high current pulses. Thus the high harmonic current and higher losses upstream. >>



It should help considerably if you were to power computers off of three-phase and rectify with three-phase bridge(which consist of six diodes) since resulting DC never crosses zero and has much lower ripple than single phase, which reaches zero. Does that work?
 
I have a couple NMB 460 watt PSU for Athlon MP Thunder K7 motherboards. These have no voltage switch on the back, and the label on the PSU says AC INPUT 100-240V ~8.0A 47/63HZ. So I would assume these have this active PFC?
 
jerboy -


<< It should help considerably if you were to power computers off of three-phase and rectify with three-phase bridge ( which consist of six diodes) since resulting DC never crosses zero and has much lower ripple than single phase, which reaches zero. Does that work? >>


It sure would help. Six smaller current pulses instead two. If a 3 phase transformer is required as part of a big system, it can be wound for any number of output phases by combinations of the three. So, since diodes are cheap, a poly phase arrangement, say, 24 diodes and 12 phases is possible. The harmonic current upstream will all but vanish and the output, unfiltered, will be virtually DC. 3 phase power is a beautiful thing for high powered stuff.

mastertech01 -

It sure sounds like that supply is an active PFC type. They sound like good ones. With the pre-reg of active PFC, the rest of the circuitry can be pushed harder. 460 watts is getting up there. I noticed my new printer has a universal voltage input, also. I suspect this will soon be more common than not. I googled a link from Sparkle and they have PFC'ed supplies, too.

Sparkle? I wonder if Sparkle also uses the brand names Smolder, Crackle or Blaze. Sorry. And, no, I haven't been into the butyl cellosolve - lately.

 


<<
It sure would help. Six smaller current pulses instead two. If a 3 phase transformer is required as part of a big system, it can be wound for any number of output phases by combinations of the three. So, since diodes are cheap, a poly phase arrangement, say, 24 diodes and 12 phases is possible. The harmonic current upstream will all but vanish and the output, unfiltered, will be virtually DC. 3 phase power is a beautiful thing for high powered stuff. >>



Do you think computer power supply industry would ever consider building a different model unit for commercial/institutional use?


The lighting industry makes luminaires in small business/consumer model and institutional/commercial model. The former runs on standard 120V household current, the latter requires 277V, which is only available in buildings with 480V three phase service. These buildings receives service at five digits voltage three phase power.

In office buildings and such:

The first transformer gives 3ph 4160V for their massive HVAC equipment

Second transformer connected either directly to high voltage service or downstream of 4160V transformer gives them 480Y/277V. 480V is used for elevators, other motorized machineries, refrigerated show case and such. 277V is almost exclusive for lighting.

Third transformer gives them 208Y/120V. 208V isn't used much. This transformer is primarily there to serve 120V loads and computer peripherals represent large portion of 120V load.

Since they have three-phase 208V available, it's such a waste not utilizing three phase power to run computers.
 


<< It sure would help. Six smaller current pulses instead two >>



Not really - if you had 3 computers, one on each phase - then the total pulse size on each phase would be the same with this arrangement as with the conventional one.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top