In the bad old days, most of the field applications engineers in power management had served several successful years as power supply designers prior to going into field applications. They had an intrinsic knowledge of how stuff works, from whiteboard down to the pcb layout, EMI signatures, transformer design and component attributes. Times have changed. Due to lots of restructuring, downturns and basic evolution at IC companies, what used to be clear concise information from those that knew it is now, by and large, grey and cloudy hearsay. As an active power supply designer and also a Field Applications Engineer, I’d like to make some attempt at clarification. To accomplish this, I’d like to go through the reasons for using active PFC, the claims made in favor of active PFC, and lastly the topologies and design careabouts.
I once heard one of the newer FAE’s tell a room full of designers “PFC of course stands for power FREQUENCY correction” the words that followed were more shameful. In his defense, he was a software engineer prior to this engagement. For better or worse, if this are the new sales tactics, we really need some old school knowledge!
Let’s start out with the basics. Active power factor correction is simply a means to bring the distortion of the line current down to a low value with active power electronics circuitry. In switchmode power supplies, current and voltage tend to be in phase; however, the harmonics in the line current waveform are excessive, as shown in Figures 1 and 2, left side. (Note: due to the high impedance of my isolation transformer, the clipped top of the voltage in Figure 1 caused by the narrow conduction angle of the diodes is exacerbated.)
A good engineer would ask the question: Why do we need PFC? To answer this question requires delving into some of the fundamental concepts of power distribution and transmission.
If we investigate power distribution in the U.S., there are two different categories. The first is residential, and the second is commercial/industrial. There may be subcategories, too. I’d imagine that a steel mill with 128 kV linesets coming into the premises is on a different scale than the 480 V drop at the local Kinko’s. We won’t worry about that here.
In the commercial and industrial arena, a surcharge is assessed when PF drops below 0.85. PF is monitored in these applications. It’s easy to see why PFC would be advantageous for appliances in this market—it’s direct. Low PF = higher bill. Correct that and save money.
In most residential applications, we see a slightly different story. The energy consumed by most households today is measured by a rotating disc watt-hour meter as shown in Figure 3. This type of power meter measures kW*H. This meter is set up like a shaded pole induction motor with the mains voltage driving the potential winding.
In this winding the voltage is fairly constant and the inductance is very high. The resultant excitation current in this winding lags the mains voltage by 90 degrees due to the high inductance. The current legs then drive shading windings. The phase difference between the potential and current windings is then 90 degrees for a resistive load, thereby setting up a maximal resultant torque on the disc and spinning the disc at a speed proportional to the power flowing through the meter. There are some damping and counter-torque mechanisms on the disc that we won’t discuss here. The rotating disc then drives gears and a display that is calibrated to kW*H.
The flux in the current winding is a measurement of MMF or magnetomotive force. Now for the odd part, the stuff that Mr. (or perhaps it was Dr.) Power Frequency Correction didn’t tell us: What do we pay for? If we assume that the potential winding voltage is fairly constant, the MMF caused by the line current drawn then determines the speed of rotation of the disc. This mechanism measures real power. As justification, consider a purely reactive load. All current is at the fundamental, and 90 degrees out of phase. There is no displacement torque on the disk in this case.
MMF is a time-averaged function. For an AC half cycle, any combination of harmonics in the line current that results in the same time-averaged value will make the disk in the meter spin at the same rate (assuming the fundamental is at the same phase angle with respect to the potential winding). There are practical limits to this. For example, if we were to draw heavy current in the 601st harmonic, the steel laminations would have a lot of eddy current loss, as would the copper. The rotating disk would never see that energy. The outcome of this time-averaged function is odd for residential households in the greater U.S. at present. Power factor doesn’t matter, at least in terms of the billable kW*H that we pay for. The question still stands unanswered: Why do we care about power factor at the residential level?
If we research further, we can talk to the engineers at the power company. (Note: They are hard to get a hold of initially. Navigating the toll-free dial-in with the goal of speaking to an engineer is tricky). They will agree with the statements on how a rotating-disc watt-hour meter works and responds, as will the folks that make the meters (Fig. 4). Why then would we need PFC in our homes? Why does Mr. PFC tell us that “PFC is good! PFC is great! \\[rah, rah, woof woof\\]”?
To wax theoretical, we are cascading a non-ideal power conversion stage onto another downstream converter. While the downstream converter can likely be optimized due to having a pre-regulated input (optimal turns ratio, duty cycle, etc.), this is still a non-ideal stage added to the mix. The composite efficiency of the equipment likely won’t be higher. We pay for that! For a numerical example, if we have a 100 W output flyback converter at 80% efficiency without PFC, we consume 125 W to run this converter with a line current waveform that looks like the one in Figure 1 (left side). If we add a PFC stage to this converter that pre-regulates the B+ voltage as seen by the primary winding of the flyback, we can drop the secondary turns and run at a more optimal duty cycle instead of having excessive secondary turns to accommodate low line conditions. This may raise the efficiency of this flyback converter a couple of percentage points due to having lower peak currents through the primary switch and secondary rectifier for the same power output, as well as less flyback voltage reflected to the primary. Let’s say the efficiency of the flyback stage jumps up to 85%. Let’s further say that the efficiency of the PFC stage is 90% at this throughput power level at nominal line voltage. The efficiency of the whole converter is then 76.5%. For the same output power, we have to put in roughly 131 W compared to 125 W without PFC. How then is this advantageous?
This is where we have to dig in further with the kind engineers at the utility company. Where would those harmonics cause heating? Where would that excessive stress show up? They would cause heating in the distribution transformers. Those harmonics would cause excessive eddy current losses and proximity effect losses in the core laminations and the copper windings. Further, those transformers aren’t cheap or easily available (Fig. 5). Failures at those aggregate power levels can require thousands of man hours, not to mention the machinery. And they can leave lots of people in the dark.
It makes sense, then, to avoid overstressing those distribution transformers by suggesting power factor correction circuitry on critical appliances and loads. In speaking with various engineers at various power companies in the U.S., I have learned that some urban areas actually do monitor power factor at the residential level with newer digitizing power meters that no longer rely on a rotating disc. These meters accurately sample and digitize the entire voltage and current waveform and compute and log real power, reactive power, and power factor. These newer meters will proliferate in time and replace the rotating disc units entirely. In many other countries, PFC is a requirement above a certain power level as a barrier to entry in that market, for example, the European countries under EN61000-3-2 for equipment over 75 W power input.
While this certainly is a lot of info, when we hear “PFC saves you money” from Mr. PFC, we can better understand what it is that he is saying. Cascading a non-ideal PFC stage certainly doesn’t make the net kW*H of the end equipment any less, and thus the meter doesn’t spin any slower, and our bill doesn’t magically decrease. It does, however, limit the heating in the distribution transformers from the excessive harmonic currents that would have been drawn for that load. If that multi-million dollar, multi MVA distribution transformer were to fail from overheating caused by excessive harmonics, the consumers tied to that transformer would foot the bill, possibly in a subtle fashion as a rate hike, surcharge, or perhaps a freezer full of food gone bad from a 28-hour blackout. While this isn’t nearly as evident as the direct surcharge for poor power factor on the industrial and commercial side, it is equally or perhaps even more painful on the residential consumer’s pocket book.
We have reinforced some basic concepts so far: 1) most residential power meters in the U.S. measure current via MMF applied to the rotating disc, and this method doesn’t register any excessive charges for harmonics; 2) adding a non-ideal PFC stage in cascade with existing equipment does not improve the overall efficiency of the equipment; and 3) the main reason for the push for PFC in residential applications in the US is to minimize heating in the large-scale power grid machinery by minimizing harmonic currents.
As a side note, there are actually methods to classify the amount of harmonic currents in a given load that a large-scale transformer supports. This is called the K-factor of the transformer. A curious reader may want to look at: http://www.powertransformer.us/federalpacifick-factortransformers.html for more information on this. On the shoulders of sound engineering principles, traceable to the fundamental roots of the problem, we can say that there is in fact a need for PFC in both residential and commercial/industrial appliances. In the next section we will discuss how PFC works, various topologies and design careabouts. As for where the market is going, I’ve seen a lot of research money being allocated to monitoring and correcting residential loads. The rotating disc watt-hour meter may be on the extinction list in the next 20 years. The emerging smart grid systems may bring the rotating disc power meter to extinction even quicker.
IR Apec 2005 Paper “One Cycle Control IC Simplifies PFC Designs”
Unitrode U134, “UC3854 Controlled Power Factor Correction Circuit Design”, Philip C. Todd
Navpers “Basic Electricity 10086A” US Navy publication, 1960