Driving CFLs Requires Unique Voltages, Sequencing, and Attention to EMI, PFC, and Faults

The compact fluorescent bulb (CFL) is an energy-efficient alternative to the venerable incandescent bulb for area illumination: it converts approximately 25 percent of the input electrical energy to visible light output, compared to 5 percent for the traditional bulb (an improvement of a factor of five). However, unlike the incandescent bulb which can operate from AC or DC at a wide range of voltage levels (depending on design and construction specifics), or the LED which requires a regulated DC current source of 20 to 50 mA, the CFL needs a more complex startup and operating-voltage sequence.

In addition, there are other challenges for the CFL's sophisticated drive circuitry. It must meet regulations for EMI emissions, provide power factor correction (PFC), and deal with failure modes in the CFL bulb itself, all of which make driving the CFL a complicated issue. Finally, if there is a need to dim the CFL, which is often the case, the drive circuitry must implement dimming in a cost-effective and pleasing way, comparable to the low-cost Triac-based dimming of incandescent bulbs. All of the requisite CFL circuitry, called the ballast, must be packed into a housing at the base of the bulb to form a standard Edison-style screw base lamp. [Terminology note: a bulb is the glass enclosure; a lamp is the bulb plus circuitry and base unit, but the terms are often used interchangeably.]

CFL voltages and timing

In a CFL, electrons from a heated filament flow through the bulb and collide with mercury atoms in the tube. As a result of these collisions, photos are released, but they are in the UV wavelength range and thus not useful for illumination. To convert UV to visible light, the inside of the fluorescent tube is coated with a phosphor. As the emitted UV photons hit this phosphor, the coating glows and gives off the visible light of the bulb.

Sounds simple enough, but unlike the incandescent and LED bulbs which operate from a steady-state power rail, the CFL voltage is a sequence of three voltage phases (Figure 1). In the pre-ignition phase, the lamp needs a current to pre-heat its filaments. Next, the bulb needs a high-voltage DC level for ignition, followed by a high-frequency AC signal during the actual running (illumination) phase.

Figure 1: Unlike the steady-state voltage requirements needed by the incandescent bulb, or the constant-current drive of the LED, the CFL has three distinct voltage phases between off and normal illumination: pre-ignition, ignition, and operating.

All of these voltages and currents must be derived from a single AC-line voltage (120 or 220 VAC nominal). To do this, the line voltage is first rectified (full wave), and the rectified peak-voltage charges a capacitor which results in a stable DC voltage (Figure 2). This DC on the capacitor is then converted into a square-wave AC signal at higher frequency, via a half-bridge switching circuit. Finally, the high-frequency AC drives a resonant tank circuit which, after filtering, provides the sine-wave voltage and current for the lamp. This resonant tank circuit is key to operation, as it looks like a series LC-tank circuit with high Q during pre-ignition. However, once ignition occurs and the lamp is running, it looks like a series L/parallel-RC tank circuit.

Figure 2: Converting the AC line to the specific voltages and sequencing needed by the CFL requires several power-management functional blocks (Courtesy of International Rectifier).

The ballast of the control circuit manages, compensates, and corrects for operating issues. In a typical design, the controller sweeps through the frequency of the half-bridge rectifier, from a defined maximum value down towards the resonant frequency of the ballast output stage (high Q) at turn-on. This pre-heats the lamp filaments as the frequency decreases, with a resultant increase in lamp voltage and current draw. Once the lamp voltage goes above its ignition-voltage threshold and ignites, the controller circuit regulates the lamp current to maintain the desired power and thus illumination level.

Control issues

The control circuitry in the CFL ballast must be designed to accommodate failure modes, such as open filament, AC-line dips (brownout), and lamp "failure to strike" (non-strike), plus these other issues:
  • EMI emissions: The switching circuitry in the lamp is a source of undesired RF emissions, which can affect nearby wireless links and even propagate backwards through the AC power line. To minimize these and ensure the design meets regulatory demands, the controller must manage the edge-transitions that cause this EMI.
  • PFC: Regulations require that AC-line-operated loads have a power factor close to one (unity), with the specific power-factor value set by the wattage of the load. In general, PF of 0.9 or higher is desirable; associated with high power factor is low THD (<30 percent is a reasonable goal for this application). Although the CFL bulb itself is a filament load and thus resistive with PF = 1, the circuitry between the AC line and the filament is capacitive, thus changing the type of impedance of the load. Therefore, the ballast must implement one of several PFC techniques; “valley fill” passive PFC is the most common approach, due to its low cost and effectiveness at the typical power levels of the CFL.
  • Crest factor: This is the ratio between peak current and average current of the load. Even if the filament is intact, the lamp may fail to “strike” or ignite (often a temperature or transitory situation). In this case, the lamp voltage will rise, excessive currents will flow, and internal inductors can saturate. This can damage the MOSFETs in the half-bridge circuit of the ballast and result in a useless lamp. To avoid this problem, a properly designed ballast will monitor the crest factor, and turn off the internal circuits around the MOSFETs if the peak/average current and current flow are too high.
  • Lamp dimming: It is straightforward to dim an incandescent bulb using a Triac: the line AC is turned on/off at different points in the AC-waveform phase, which reduces the average power delivered to the bulb while maintaining the peak voltage, to ensure proper incandescence. This approach is not compatible with the drive needs of the CFL, and using a standard dimmer with a CFL results in no output or poor dimming operation in terms of range, consistency, and linearity.
To control dimming, a more complex feedback-driven circuit is used. Although there are several ways to implement dimming, one approach uses a measurement of the AC lamp current combined with a DC-reference voltage corresponding to the desired dimming level. The controller then regulates the low point or valley of the combined AC+DC signal by adjusting the operating frequency of the half-bridge circuit, which in turn causes the amplitude of the lamp current to change. In the peak-valley design, as the DC reference increases, the valley floor also increases, and the feedback circuit will cause the frequency to decrease; this increases the gain of the resonant tank circuit, which increases the lamp current. In short: CFL dimming is not a simple process.

ICs make drive, dimming more effective

Although it is possible build a CFL ballast which meets the various operational and fault mandates from discrete components, the performance will be marginal, control consistency will be poor, PFC will be inadequate, and protection against failures will be minimal, while cost and physical size will be high. Fortunately, IC vendors see the CFL ballast as a major opportunity and have developed devices with internal sophistication and complexity which address the issues effectively; some also provide dimming capability.

For example, the UBA20261 from NXP Semiconductor, with two internal 350 V, 1 Ω, 5 A MOSFETs (and similar UBA20262, with two internal 600 V, 3 Ω, 2.7 A MOSFETs) provides most of the circuitry needed for complete CFL drive and power-control function in a 20-pin IC, as shown by the complexity of its internal block diagram (Figure 3). The IC supports CFL functions including four-level step dimming adjustment using a standard on/off mains switch, adjustable memory retention time for step dimming, and an adjustable minimum dimming level.

Figure 3: ICs incorporate many of the needed functional blocks and features which makes the CFL more efficient, compact, and low cost, as shown in this internal block diagram of the 20-lead UBA20261 device from NXP Semiconductor.

Its internal state diagram is proof of the multiple steps needed to manage a CFL. The logic in the IC implements the state diagram providing the functions needed to manage the CFL in its various phases and fault modes, including protection against overtemperature, capacitive mode failure, overpower mode, and overcurrent in both boost and burn states (Figure 4).

Figure 4: The UBA20261 also implements a state diagram for the timing and sequencing of functions of the CFL.

Of course, a complete CFL lamp assembly still needs passive external components (Figure 5); here, the UBA20261 vendor provides equations which let designers calculate optimum values for their specific bulb and application; the vendor also provides a detailed BOM for a typical CFL bulb.

Figure 5: A complete CFL lamp assembly needs both drive and management functions, as well as passive components chosen and properly sized to yield the desired operating performance, as well as match the inherent characteristics of the particular CFL bulb chosen.

Finally, a demo board allows users to evaluate and fine-tune performance, while also showing that the entire working core of the demo board complete assembly could fit into the base of a CFL lamp with an Edison-type base (Figure 6), and screws into a standard socket.

Figure 6: The entire CFL drive circuit based on the UBA20261, both the IC and associated passives, fits onto a circular PC board at the center of this demo/evaluation board, and so could be housed in a standard Edison-base lamp.


Compact fluorescent lamps (CFLs) are rapidly becoming an attractive replacement for energy-wasting incandescent bulbs due to their high efficiency, compact form-factor, and low cost. Unlike incandescent lamps, which can operate directly from an AC or DC source of various voltages, or LED-based lamps, which require a low-voltage DC current source, the CFL needs a power supply (“ballast”) which provides both an initiation (strike) voltage as well as the proper operating voltage. This article has examined how to supply the unique voltage-drive requirements of the CFL as well as associated concerns of electromagnetic interference, power factor correction (PFC), and the challenges of dimming a CFL.

For more information on the parts discussed here, use the links provided to access product pages on the Hotenda website.

Additional Reading
  1. TechZone article “Compact Fluorescent Tribulations”