Blog Post

LED and Power: Quality Matters

October 18, 2012

LED replacement lamps look super-efficient on payback charts and utility bills, but they may be sucking more power than you realize.

LED replacement lamps like this one from Cree  have a high power factor; those intended for residential use often don't. Photo Credit: Cree, Inc.

GreenSpec and EBN have reviewed a number of LED replacement lamps over the years and have reported on improvements in efficacy (light output in lumens per watt of electricity consumption), color, and light quality; lower costs; and its increasing acceptance.

In the EBN article LEDs: The Future is Here, we explored briefly how LEDs interact with the power supply, but we were surprised when an email from Stefan Bernath, Alberta infrastructure energy coordinator, came in describing how LEDs were affecting his building’s power supply.

His email got me wondering if power quality and LEDs are going to be a bigger problem in the future.

Efficacy and power quality

Some 60-watt equivalent LED replacement lamps now have efficacies of over 90 lumens per watt (lpw) compared with only about 14 lpw for a standard 60-watt incandescent. When an incandescent bulb uses power from the utility, it may not be very efficacious, but its power factor (PF)—basically the amount of power coming from the utility used by the lamp—is 100% (1.0 on a scale of 0.0 to 1.0).

LED luminaires with separate drivers have power factors greater than 0.9, which is excellent, but an LED replacement lamp only can have a PF as low as 0.5 (Energy Star requires a PF of 0.7; those lamps are listed in GreenSpec).


BuildingGreen relies on our premium members, not on advertisers. Help make our work possible.

See membership options »

What is PF? Nancy Clanton, P.E., president of Clanton and Associates, used the following analogy: “If I pour beer into a glass and get some beer and a whole bunch of foam, the beer is the watts, which is usable power, but the foam is not usable.”

In other words, power factor represents the percentage of drinkable beer you have in your glass.

A customer using a 10-watt LED with a PF of 0.5 only pays for 10 watts, but the utility would have to generate twice that power in volt-amps to run that lamp; we end up thinking that we are saving more energy than we are—and the utility is paying for our foam.

Though the 20 volt-amps generated by the utility to run that LED is still much better than the 60 required to power the incandescent, as LED replacement lamps gain more market share, those 10 foam-like volt-amps are going to add up. If the true goal is to maximize the efficiency of utility power, use less fuel, and create fewer emissions, then the LED industry has some work to do.

Other LED inefficiencies

LEDs can have another effect on power in a building.  “LEDs are diodes and are trying to take a sine wave and turn it into a DC signal,” said Clanton, “and whenever you do that, you get garbage and junk on the line.”

This garbage is known as total harmonic distortion (THD). THD could potentially shorten equipment lifespan, increase power losses on the transmission line, heat up transformers, and affect the performance of the LEDs and other electronics. THD was a problem for early fluorescent fixtures and computer “cube farms,” and Clanton is worried we are repeating history.

According to Bernath, his team replaced about 500 40-watt and some 100-watt incandescent lamps with 8- and 12-watt LED replacements, respectively, at the Alberta Legislature Building (partially due to high replacement costs for incandescents in the high-ceilinged rooms). When electricians went back and measured power consumption, the team was surprised.

Though the energy savings were big (he predicts a payback of 0.64 years), they discovered that the THD of the LED replacement lamps was over 66% for the most efficient lamp on the market. “It [THD] was a big concern because it increased the current on the neutral wire,” he said.

Bernath is not too concerned because LEDs make up such a small percentage of the building’s overall load, and the gauge of the neutral wire can handle it, but wonders what the impact will be on the transformers and other equipment.

Residential/commercial split

The lamps used in the Alberta Legislature Building were A19-style LED replacement lamps, which are meant to replace standard screw-mount incandescent lamps, primarily in residential applications.

Gary Trott, vice president of product management at Cree said, “For our lamps, our spec is 20% THD with a power factor of 0.9.” This is a commercial standard, and makes Cree’s LED replacement lamps eligible for utility rebates. As usual, achieving this performance comes at a price.

“You can save a lot of money by having lower power quality,” said Trott. The drive to lower the price of LED replacement lamps and the lack of tough standards mean that fewer manufacturers are going to care about PF—and, especially, THD—which could have unforeseen consequences in terms of the nation’s power quality.

A standard on the horizon, finally

Trott is part of a group in California developing the California Quality Lamp Specification, which would improve on Energy Star standards by requiring a PF ≥ 0.9, along with a color rendering index (CRI) of 85. Though the standard does not address THD directly, Trott says that THD and PF are closely related, and as you improve the quality of the lamps to achieve a PF above 0.9, THD typically improves and becomes less of an issue.

When the California standard is developed and manufacturers can show that their lamps meet it, we’ll raise the bar for products listed in GreenSpec. In the meantime, I’m curious if any of you have had experience with LEDs and power quality. If so, we’d love to get your input.

Add new comment

To post a comment, you need to register for a BuildingGreen Basic membership (free) or login to your existing profile.


October 21, 2012 - 7:22 am

At the risk of teaching my Grandmother to suck eggs…

Dig out the instruments you used to use for drafting in pencil.

The ruler (just the straight edge) represents a power factor of 1.

In the 45 degree set-square, the length of the hypotenuse represents the current that the supply cable sees for the effective current represented by the length of a side: 1.414 times as big at a power factor of 0.707.

Correspondingly, the length of the hypotenuse of a 30/60-degree set-square represents the current that the supply cable sees for the effective current represented by the length of the short side: twice as big at a power factor of 0.5.

It is these increased currents at low power factors which are a problem for generators and distribution networks, and why utilities sometimes offer rebates for the installation of power factor correction equipment.

The main reason for introduction of fluorescent lighting in factories was that the effect of its "leading" power factor canceled much of that of the "lagging" power factor of the motors used to drive machinery, producing a good overall power factor (sorry, but explaining "leading" and "lagging" would take too long).

October 19, 2012 - 3:44 pm

I like the beer analogy for Power Factor, but the effects on the various bits and pieces between the lamp and the power station are still not too well explained here - or, I believe, well understood by many folk in the business.

As for THD, back in 1971 when I was working on CabinTaxi systems, speed control of AC electric motors was typically by means of phase-cutting thyristors. On behalf of my electrical engineer colleague I had to argue for DC feed to the network and inverter VFDs for the individual cabs to prevent the huge distortion which thyristor drives would have created in an AC feed system if many cabs were traveling at say 50% of maximun speed.

I therefore agree that we could well see DC grids for such purposes. High-power DC links are often used in Europe to connect countries with non-synchronized AC grids; and also for underground links, because they don't have the capacitive losses of AC lines.

October 19, 2012 - 3:19 pm

I spoke with Stefan this morning who noted that the payback was 0.64 and not 6.4 years. Big difference. Sorry about the mispllaced decimal!

October 18, 2012 - 7:23 pm

All this talk about DC power makes me wonder. Does this bode well for the PV industry? Would inverters become unnecessary? And, could a PV system power LED lamps without need for inverters?

October 18, 2012 - 5:02 pm

Tyson, A 10W LED lamp with a PF of .5 consumers 20 VA. (not watts).  I'm sure way back when the electric meters that only measured Watts and ignored PF were cheaper to build.  Most homes have very similar types of loads.  So they assume a typical PF for residential and charge accordingly.  You can also think of the "foam" as drag on a fast moving vehicle.  The more drag, the more gas you need to go the same distance, but the same work has been done.  Everything becomes waste heat eventually.  I know one generator manufacturer likes doing demonstrations where they would run their whole plant on their generators.  Well, most of the plant.  They only paid for "beer" on their electric bill.  Their generators made the "beer" and they would still have the Utility supply all of the "foam" at no cost during the demo.

Paula, thanks for the link.

Brent, nitpicking.  "When an incandescent bulb uses power from the utility, it may not be very efficacious, but its power factor (PF)—basically the amount of power coming from the utility used by the lamp—is 100%".   This sentence sounds weird.  PF is not the "amount" of power from the Utility.  It difficult to explain, maybe efficency of demand.  How do you explain losses from drag?  The incandescent bulb does not create light efficently but it does draw power efficently with a PF of 1.0.  60W A-lamp uses 60VA.  While the 10W L-Prize lamp uses 14VA.  The energy savings is actually 76% rather than the 83% lower power bill you'll get.

October 18, 2012 - 1:54 pm

Bill, if you haven't already, you might be interested to check out our coverage of DC microgrids. Folks who are at the forefront of that effort believe that the grid will gradually switch back to DC because we have a lot of new DC at both the generation end (renewables) and the consumption end (electronics)—and there are no longer greater inefficiencies in DC distribution.

October 18, 2012 - 1:33 pm

So you're saying that a 10W LED lamp with a PF of .5 (50%) really consumes 20W but the consumer only pays for 10W? If the lamp consumes 20W, the utility does have to generate additional volt-amps (watts), but I don't see how the utility is absorbing that cost (the foam part of  the beer analogy.)  I would assume that the 10W of "foam" lost to inefficiency are actually just producing wasted heat in the LED driver (transformer), but the customer is still paying for the full 20W that the LED lamp/driver is consuming.  Can you provide any clarification?  This seems logical to me, but I'm no engineer.

October 18, 2012 - 1:07 pm

Thanks Bill, that gravel road/THD analogy is great! And please, we encourage nitpicking, especially when paired with quality feedback.

Energy Star does not have a standard for THD (beyond the most low-bar <200% THD that these lamps require), so there is no reason/incentive for companies to supply that information. Apparently, THD just isn't on most LED manufacturers' radars right now, especialy when it comes to  residential A19 LED replacements.

With regards to the Philips lamp, the 66% THD caught me offguard as well. This number came from real-world tests done in Alberta, but I heard from a lighting engineer researching LED/THD that the actual THD of that lamp is 88%!  I have not been able to verify this, however, and can't explain the 66%/88% discrepancy.


October 18, 2012 - 10:46 am

Decent coverage of a complex issue.  I don't want to nit pick the details but I had one question.  You said one product had over 66% THD and gave a link to the Philips product you've written about previously.  I'm not understanding how this product can have such poor THD and have an Energy Star listing.  None of Philips literatures discloses their THD for this lamp.  I'm also looking forward to the new standard and increased product transparency.  Really bad THD causes a lot of heat build up in transformers.  Excessive heat shortens life spans in electrical equipment. 

I like the beer analogy for PF.  Now imagine that THD is like driving on a gravel road.  The higher the THD gets the more ruts and washboard effect on the road.  It doesn't kill your car outright, but it does shorten the life on specific part of the car.

Variable Frequency Drives are another huge source of THD in buildings.  With so many loads in buildings going to DC I wonder if we'll go back to the AC vs DC wars of Tesla and Edison.