Compact fluorescent bulbs (CFL) are rapidly supplanting incandescent bulbs for lighting, with LED bulbs not far behind. The energy savings are substantial, and some jurisdictions are even mandating sales bans of incandescent bulbs altogether. As designers of lighting and control equipment, we must be aware of the changes in these loads, and how they affect our designs.
A recent design of an industrial control included a relay to switch up to a kilowatt of lighting, subject to software timers and various other conditions. We tested the setup by programming the relay to cycle a bank of incandescent bulbs at various rates from 2 to 20 seconds off, allowing the filaments to cool and provide various inrush currents. The tests ran until the relay held closed, indicating contact wear and failure from arcing. Several brands of relays were tested, with several contact plating options. After perhaps 200,000 total cycles of incandescent testing, we selected the best relay for the job and launched production.
Never once did we consider that CFL bulbs would give us problems.
The first support call came from a customer with thirteen 23-watt CFL bulbs installed in a workspace, as replacements for 60 W incandescent bulbs. He indicated that the lights held in the ON position after only a few actuations of our control relay.
We could not believe that, so we replaced the incandescents in our test rig with one CFL bulb. It was true! A relay that would run 100,000 cycles, switching asynchronously on numerous incandescent bulbs ran 10 to 100 cycles with a single CFL. Astonishing.
So the race ensued to determine the failure cause. Relays stick closed due to high switching currents, but how can a CFL bulb that draws 23 watts suck that much current, enough to weld the contacts on a 10 amp relay, one rated for incandescent service? We set up a fixture with a current monitor to see.
We found that some CFL bulbs draw huge inrush currents, peaking up to 17 amps, for a duration of 300 us or more!
We arranged a conference call with a large manufacturer of CFL bulbs, and spoke directly to the designer who confirmed this expected inrush value. A CFL bulb’s inverter has a capacitor/rectifier input circuit. The inrush is limited only by the resistance internal to the inverter (including the ESR of the filter capacitor/s), and the resistance of the mains circuit. The designer had no reservations about the design of these bulbs.
In our testing, we found that the distance from the facility’s breaker box is an important factor in controlling the amplitude of the inrush spike. Connecting our test rig to an outlet right next to the breaker box produced 17 amp spikes. On the other side of the building, the spikes were reduced to 5-8 amps, due to the resistance of the intervening copper wiring.
During our tests we also noticed that common CFL bulbs contain no power factor correction, but rather draw current as you would expect a capacitor/rectifier circuit to do, in short spikes at the peaks of the mains sine waveform.
The solution to our relay problem
The first thought for a solution was switching the relay near the zero crossing. This is nontrivial as relays have finite pull-in times on the order of milliseconds. The equipment that controls our lighting is microcontroller-based, but the code space is almost full, so the solution had to be simple. We were able to characterize the relay’s pull in time, take a sync hint from the AC line, and implement a near zero crossing switching scheme with a fixed offset related to the relay closure time. Just getting to within a millisecond of zero crossing reduced the inrush current by an order of magnitude or better, allowing the relay to switch multiple CFL bulbs reliably. Fortunately, we were able to build in CFL compatibility with a simple software update.
Thoughts on the design of CFL bulbs
Through this process, we were stunned at the poor design and cheap construction of CFL bulbs, and their complexity with respect to incandescents. There are several considerations that came to light even after we had modified our product’s software to work with CFL bulbs.
The first is CFL life under switching. To test our zero crossing relay solution, we set up a cycle test using numerous parallel connected CFL bulbs. The relays under test lasted well, but the CFL bulbs dropped like flies. Yes, it appears that CFL bulbs have a sensitivity to cycling on and off that shortens their life dramatically, to perhaps 500-1000 cycles. This phenomenon is known, but not widely publicized to consumers. Apparently the failure mechanism is in the tube, and not the electronic assembly (see the further reading references).
Another consideration is power factor. The power factor of an incandescent bulb is unity, as they are resistive loads. CFL (and LED) bulbs draw current, albeit less current, in short spikes. If there are requirements for power factor correction in other electronic equipment today, how long will it be after all incandescents are replaced by CFL or LED bulbs until there is a PFC requirement for them? Will that fit in the space of a standard CFL inverter? What would the cost increase be? (We did test one LED bulb, a Sylvania LED8A/DIH/F/830, which had a much more sinusoidal current waveform.)
CFL bulbs are also remarkably more complex than incandescents. We have replaced something very simple with a design 10x more complicated, containing many more types of materials from geographically diverse locations. There is a recycling requirement for CFL bulbs due to the mercury content. Does this recycling process also reclaim the materials in the inverters, or is that section of each bulb simply landfilled?
Did you know that CFL bulbs are not recommended to be installed upside down, or in enclosed fixtures? Will the typical consumer respect these limitations? Will fires result, or shortened bulb life? (Our customer had the offending CFLs installed upside down in ceiling fixtures in an industrial setting, without considering this limitation.)
What happens when consumers run CFL bulbs on timers and electronic controls? Will they understand why their controls are stuck ON, and what to do about it?
What happens when multiple CFLs in a fixture near the main breaker box are switched manually, and the switch contacts arc heavily? Does a fire hazard develop over time? Does the also-cheap light switch require replacement sooner than when switching incandescents?
And what of quality? Obviously, the cheap, consumer product level design of these bulbs leads to unreasonable failure in cycling applications. An incandescent bulb can be made inexpensively but reliably, due to the simplicity of the design. This is apparently not the case with CFLs.
Hopefully, CFL bulbs will be but a blip in the history of lighting, being replaced soon by LED bulbs, whose cost is decreasing, and they do not suffer from cycling failure (according to our limited tests).
This customer support exercise strengthened our product, but exposed to us the weakness of the CFL bulb. Let’s hope that the free market ultimately demands a product with better performance.
For Further Reading
Our experiences with CFL bulbs were but a small snapshot of their performance. The available designs and brands perform with wide variation. Detailed information appears in the following references.
- National Lighting Product Information Program, Specifier Reports, “Screwbase Compact Fluorescent Lamp Products,” Vol. 7, Num. 1, June 1999.
- Jump, Hirsch, Peters, Moran, “Welcome to the Dark Side: The Effect of Switching on CFL Measure Life,” 2008 ACEEE Summer Study on Energy Efficiency in Buildings.
- Wagner, “Inrush Current,” September 2011.