April 8, 2009

Unintended consequences - CFL's and power quality

There are several standards that you use to measure electrical power. The Volt and the Ampere are ones that most people have heard of. Multiply the two together and you get Watts (10 Volts * 10 Amps = 100 Watts). Multiply Watts by Hours and you get a Watt Hour (or Joule if you want to get Metric -- one Joule is one Watt Second). One unit of measurement that is not in common use is Power Factor. Household service meters and most small business service meters do not measure Power Factor so maintaining a good one is not an issue. As the Voltage swings back and forth (60 cycles per second), a purely resistive load will have the maximum current flow happening at the maximum voltage. Simple enough. When you have an reactive load, the maximum current can lag behind the maximum voltage creating huge loads on the line when the line is at a low point of that cycle. This can distort the waveform, cause overheating and put a big load on the system. That being said, let's segue into the increased use of Compact Fluorescent Lights (CFL's) -- used wisely, these are a win/win deal especially the later units. They now have good color, come up to full brilliance in a few seconds and last for a long time if they are used correctly (only turned on and off infrequently). We do not use CFLs in our bathroom or cupboards but we do use them for interior lighting at the store and in our home -- it gets dark, the lights come on and stay on for a few hours. A lot of other people like them too as the price has been coming down (with the help of government subsidies (re: our tax dollars)). Problem is, most of these CFL's have a crappy Power Factor. From Electronic Design News:
Utilities suffer from CFLs’ poor power factor
Every CFL light contains a small ac-dc power supply with reactive components in it that will affect the CFL’s power factor (PF) – that is, the load presented to the ac line. The closer the PF is to 1, the better. A load with low power factor (<.85) draws more current and is less efficient than a load with a high power factor for the same amount of useful power. The higher currents required by the lower PF devices mean increased energy lost in the grid due to such things as I2R losses. These power losses don’t show up directly on our electricity bill, but the utilities sure see the effects.

I put one of my home CFL bulbs on my Kill-O-Watt power meter recently and measured its power factor: It was .57. This is lousy. Although each CFL is only 13W, there are millions of them out there. Why no PF regulation, as there is of higher-power, but less ubiquitous devices?

I emailed Peter Banwell of the EnergyStar program and asked if EnergyStar was considering making minimum PF a requirement for Energy Star compliance. He replied, “We looked at this in detail several years ago and decided against it, though there are a couple of utilities that still support the idea. We may take this up in the future, as the market share grows, but right now it is still in the noise in terms of impacts.”

Coincidentally, after our email exchange I ran into Mike Grather of Luminaire Testing Laboratory. He recently ran a series of life-cycle and performance tests on a batch of 100 CFLs with various power ratings averaging approximately 20W each. They assumed a PF for the lights of at least .75 and sized the power supply at 3KVA. However, when they powered up the bank of CFLs, the 3KVA supply was inadequate. Grather checked the power factor for the CFLs and found they ranged from .45 to .50. Their “real” load was about twice that implied by their wattage.

CFLs are still an efficient form of household lighting, but their poor PF number is leaving money on the table. However, it’s clear that at about $2 each there’s not a lot of room for adding power factor correction circuitry. On the other hand, utilities are already going to great lengths to encourage consumers to switch to CFLs, including subsidizing the price of CFLs. I doubt that consumers would be interested in paying more for a feature that actually benefits the utility directly, not them. Perhaps utilities will start to subsidize high-power-factor CFLs, rather than the mediocre ones we can buy now.
The upshot is that you are replacing a 60 Watt incandescent bulb with a 13 Watt CFL. You are paying the utility for 13 Watt Hours of energy each time this bulb is illuminated for one hour. The utility has to generate 26 Watts though. They are loosing 50% of their residential revenue for each bulb installed. The savings of 26 from 60 is still substantial but not as wonderful as it seems on the outset... Posted by DaveH at April 8, 2009 9:44 PM | TrackBack

BACKGROUND: I'm a mechanical guy, not too swift on electricial/electronic stuff, but I'm in the process of converting some of our home lighting from CFL to white LED bulbs. Overhead lighting circuit in our house is near capacity, so I want to reduce the current draw while adding a few new recessed can-lights, hence the necessity to look at LED bulbs for all of the new can-lights as well as replacing a few existing incandescent and CFL bulbs to reduce current in the circuit to make room for the new can-lights.

I, too, have a model P4400 Kill-A-Watt meter (a.k.a. "KAW") plugged directly into the wall socket (120 VAC) with a simple desk lamp plugged into the KAW meter, so I can measure the AC volts, watts, amps & power factor for various LED bulbs I've purchased. My testing of a half-dozen or more CFLs of different sizes from different manufacturers shows that the CFL power factors in the range of 0.50-0.65.

I used the KAW meter + desk lamp to measure less-expensive screw-in (Edison base) type of white LED bulbs (for example the Lights Of America 5W LED floodlamp available at Costco for $13), plus some nicer LED bulbs from InnovativeLight.com, and the KAW meter shows these LED bulbs as having PF's ranging from 0.31-0.65 - but wait - there's more!

I also plugged in the very expensive, high-end Cree LR6, which Cree rates at 0.100A and PF>0.95, and the KAW shows the Cree LR6 as pulling 0.15A and PF=0.62-0.65.

Similar result for the high-end Progress Lighting P8026-28 LED, which mfgr rates at 0.092A and PF>0.95, the KAW shows it at 0.15A and PF=0.53.

Inserting a Fluke 25 DMM in amps mode into the circuit shows the Cree LR6 pulling 0.08A and the Progress Lighting P8026-28 pulling 0.07A. (Note: I used the Fluke's "amps mode", not milliamps mode, because somewhere along the line I blew the meter's mA fuse, so I could only test using the 10A Fluke amperage circuit)

As a "known" load test, I measured a 60W Sylvania soft white incandescent bulb, and both the KAW meter and the Fluke 25 DMM are in complete agreement when I measure current & voltage through the incandescent, and the Kill-A-Watt also shows the incandescent bulb's PF=0.97, which is expected.

However, because of the highly conflicting results for amperage and power factor in comparing the KAW meter data to the Fluke data to the two high-end LED bulb mfgr's stated ratings, I am good and confused, I don't know which meter's data to believe, and I still don't know the true amperage that each LED bulb is pulling.

I definitely suspect the Kill-A-Watt P4400 meter is not up to the task of measuring accurately the amperage and PF of either CFL or LED bulbs.

Any suggestions?


- "noneyet"

Posted by: noneyet at June 2, 2009 3:18 AM

I hate to differ with you on the subject of power factor and the watthour meters used by the utility companies, but they do indeed respond to power factor. I've spend a good bit of time testing these meters and one of the tests performed is to put 100% rated load into the meter and then introduce an 0.5 power factor. The disk (or its electronic equivalent) speed drops by half, because at at 0.5 power factor despite the amps x volts (volt-amps) being 100%, the actual watts (what we pay for) drops (amps x volts x power factor).

That technical discussion aside, the utility company suffers from poor power factor, because even though they sell you fifty watts, they send you enough current for a hundred watts, and their generator, transformers and lines are rated for current and voltage, not watts.

It's a rather (to most folks) esoteric subject, but another problem comes in, too, in that solid state devices like the power supplies for CFL's generate harmonics, high frequencies that may cause problems with other equipment.

your friendly power guy

Posted by: mostly cajun at April 9, 2009 12:40 PM

Hi M.M.M.

You are right -- I very much oversimplified as the general readership is not familiar with Power Factor and its implications.

As for the metering -- the old-skool PF meters probably fake it and assume that it's a straight reactance causing the shift. When you add active electronics -- switching supplies and all that, it's a whole new ball game.

Sort of like trying to measure RMS back about 20 years ago -- you could get a lot of meters that faked it but to get a "True RMS" meter meant spending a lot of money.

As an offshoot, a lot of the over-unity energy people specifically use the early fake-RMS meters as they give deceptively high readings on non-sine waveforms.

As for arcing, out in the DaveCave, I have some analog synthesizer stuff as well as audio recording. I do have fluorescent shop lights when I need the room lit up but I use small incandescent task lights when I'm working. Much lower noise.

I have been to the Radio and Electronics museum -- a wonderful place. If I lived closer (and had the time), I would volunteer.

Posted by: DaveH at April 9, 2009 12:16 AM

You wrote:
"When you have an reactive load, the maximum current can lag behind the maximum voltage creating huge loads on the line when the line is at a low point of that cycle. This can distort the waveform, cause overheating and put a big load on the system."

Another way to say this is that a poor power factor represents a lot of circulating current in the line which is not dissipating any power (doing any work) in the load.

What I remember from designing switching power supplies is that the load isn't a pure capacitance, it's a bridge rectifier driving a capacitor input filter. The result of this is that the current into the supply is a high amplitude pulse at the peak voltage. In this case the poor power factor is because the RMS value of the spiky current wave is so mach higher than the actual power being delivered, with all the heating effects of the high RMS current. I suspect the problem is the same for a CFL, the electronic ballast is a switching power supply. If I get a few minutes free tomorrow I'm going to look at the voltage and current waves going into a CFL and see what's going on.

A lot of conventional power factor meters don't work very well for this situation, btw. I used to use an old Weston electromechanical power factor meter for measuring switching supplies until I found out that the results were mostly useless. It takes a genuine point-by-point convolution of the voltage x the current, compared to the RMS VxI to fully account for this type of power factor.

Another problem with CFLs they don't mention is radio interference. Some of them make quite a bit, which presents as a harsh buzzing sound all accross the AM and shortwave bands. Not fun if one is a ham or SWL. No surprise really, what's going on in a fluorescent lamp is basically a very long electrical arc, which generates RF noise like crazy. Back in the good old days they used spark-gaps as transmitters for just that reason. (Have you been to the radio museum in Bellingham? Lots of sparks, and other nifty stuff)

High power white LEDs will be the answer as soon as the cost is down, they don't have any of these problems.

Posted by: Man Mountain Molehill at April 8, 2009 11:13 PM