News For LED: High-Frequency AC Direct Drive for LED Luminaires

Here is the germ of an idea for another way to fix the flicker associated with AC LEDs. Your feedback is welcome on whether or not this could be workable.

We have discussed here on several occasions that the Achilles heel of LED lighting may be the drivers, and in particular the capacitors used in the driver circuits. These concerns stem directly from the fact that the vast majority of LED luminaires drive the LEDs with direct current. This leads to thoughts of so-called driverless LED luminaires, and/or so-called AC LEDs, which we have been discussing a lot lately.

I reviewed some variants on driverless and AC LEDs in a blog last June. Shortly afterwards, resident expert Ron Lenk laid out some thoughts on optimizing LED design for direct AC drive. I particularly agree with one key point in Ron's article, in that if the drive circuit is mainly a simple rectifier then, even though there is no negative voltage in the waveform, such systems are still "AC" LED luminaires.

However, there is plenty of thoughtful discussion and argument on these topics, as shown by comments from member Adrian and counter-commentary from another resident expert Ed Rodriguez. Keeping the fires burning, editor-in-chief Keith Dawson updated us on AC drivers just a month ago.

Regardless of where you stand on the issues, everyone agrees that with direct-drive designs, flicker can be a real problem. This subject has been discussed from various angles by Bill Reisenauer and Ron Lenk, in two posts.

In another discussion, Ron Amok properly put me straight when I suggested a magic bullet solution of an AC phase-shifting device that could be used to create a waveform without 0 crossings. Ron's point, as well as some of the basis of Ron Lenk's posts, was that no matter how you slice it, you need energy storage. Most of the time, energy storage means you need capacitors. Once you have a rectifier and capacitors, you are well on your way to some variant of a "conventional" LED driver -- i.e., one that presents something approximating DC to the LEDs.

Food for thought

I have been thinking about this problem for some time, wondering if there were other solutions we haven't discussed. The purpose of my column this time is to toss out some food for thought. I may get pummeled, or not, but I'm interested in the reactions from the many experts here.

I have spent some time researching ways to frequency-convert AC power, because it seems to me that if a direct AC-drive LED system has unacceptable flicker when powered from 50 Hz or 60 Hz line current, then why not power the system using a higher frequency? A high enough frequency should make the flicker undetectable by even the 1 percent of the population who can "see" 60 Hz flicker (as Ron Lenk has described).

Much of the work in frequency conversion focuses on more efficient use of electric motors. This is motivated by the large fraction of electrical energy consumed by electric motors. According to the International Energy Agency, electric motor-driven equipment consumes 69 percent of energy used in the Industrial sector, or about 3,000 terawatt-hours per year. In the last few decades, upgrading single-speed motor systems with so-called Variable Frequency Drives (VFDs) has been pursued as one strategy to improve efficiency. As the name suggests, VFDs use frequency conversion (along with many other subtleties) to alter the power supplied to AC electric motors. However, in many cases the waveforms used in VFDs are probably not suitable to consider as drivers for AC LEDs.

Another common application for AC-to-AC frequency conversion is the ground electrical power units for aircraft. Many aircraft use a 400 Hz electrical system rather than the 60 Hz we use in the US for most purposes. The ground power units provide clean sinusoidal power at 400 Hz, which seems tailor-made for an LED application. Unfortunately, such units are large and complex.

A key component in many of these designs is Insulated-Gate Bipolar Transistors (IGBTs), which are used essentially as switching devices in the power supplies. Another drawback of many AC-to-AC power supplies is that they actually use rectification, filtering, and energy storage to convert the input to DC then use the IGBTs as part of a circuit to convert back to clean AC at the desired frequency and voltage. For example, International Rectifier provides a reference design of the DC-to-AC stage using IGBTs. It goes without saying that the bill of materials for such power supplies is quite large, and if one were going to convert to DC anyway, then a traditional LED driver would be the way to go.

The provocation

In a 2012 paper, authors T. Friedli and J. W. Kolar reviewed "Milestones in Matrix Converter Research." A key takeaway from this paper is that the basic ideas of matrix converters have origins in the 1950s. While matrix converters are also a way to do AC-to-AC frequency conversion, they do not convert the input to DC, and in fact do not require energy storage. Fuji Electric notes:

...the matrix converter arranges semiconductor switches into a matrix configuration and controls them to convert an input AC voltage direction into the desired AC voltage. Since the input AC voltage is not converted to a DC voltage, there is no need for an energy storage device such as an electrolytic capacitor.

Thus I arrive at my provocation. Although work in matrix converters is also mainly driven from control needs of (relatively high-power) electric motors, they represent a semiconductor device that could be used to direct drive AC LEDs (still using a rectifier) at high enough frequency to eliminate perceptible flicker. My impression is that more work would be needed to develop efficient solutions for lighting; the existing products would not provide a drop-in solution.

Nonetheless, do you think this could be a viable alternative, especially combined with LEDs tailored to tolerate voltage variation, as Ron Lenk suggested?

Related posts: