Cirrus Logic AN349 User Manual

Lamp

Bridge + EMI Filter +

Boost PFC

Input

Figure 1. Example Ballast Second Stage and Lamp Connections

AN349

Application Note

MIGRATING FROM THE L6562 TO THE

CS1601 POWER FACTOR CORRECTION IC

1. Introduction

The CS1601 belongs to Cirrus Logic's family of industry-first digital Power Factor Correction (PFC) ICs offering best-in-class performance. The CS1601 has been optimized for lighting applications. This document compares the CS16 01 with the ST Microelectronics L6562. A commercially available ballast designed to light two T05HO lamp s is used as a benchmark. The test conditions and setup for the ballast are described in Section 2. The third section of the document details the advantages of using the CS1601 over the L6562. The fourth section discusses the differences in theory of operation. Section 5 compares the feature sets of L6562 and CS1601. Section 6 describes a step -by-step pro cedure to show how to migrate an L6562-ba sed PFC frontend pre-regulator to a CS1601-based solution.

2. Description of Test Setup

Fluorescent ballasts are a challenging application for PFC due to the presence of large load tra nsients and strict THD requirements in a highly cost-sensitive market. Given the wide input voltage ra nge of the application and the demanding form factor requirements for these applications, many design trade-offs have to be made to complete a design using the current analog Critical Conduction Mode (CRM) solutions.

2.1 Ballast Specifications & Requirements

- Input Voltage Range: 108V to 305V 50/60Hz

- Output Load: 2x T5HO lamps amounting to ~108W nominal

- Link capacitor: 2x 47F, 250V caps in series

- Low THD (<10%)

- High Power Factor (>0.9)

- No lamp flicker under any test condition

http://www.cirrus.com

MAR ‘11

AN349REV1

AN349

THD Vs Line @ 60Hz

80 100 120 140 160 180 200 220 240 260 280 300

Line Voltage (V)

CS1601 ST L6562

3. Advantages of CS1601 over L6562

The CS1601 uses a revolutionary digital algorithm that has significant advantages ov er the existing anal og CRM-based L6 562. The major advantages are described in the following sections.

3.1 Smaller Boost Inductor

The CS1601 has a digitally implemented variable-frequency, discontinu ous conduction mode (VF-DCM) -based algori thm that permits delivering the same power and the same peak current rati ngs with a significa ntly smaller inductance . This results in an inductor which is considerably smaller in physical dimensions. In the fluorescent ballast appl ication introduce d in Section 2, migration from a L6562-based solution to one using the CS1601 re sulted in the inductor being 45% smaller. The same inductor could be wound with 26% fewer turns or in a smaller core size which would result in a more compact design. The 2x EE19 core sets used in some L6562-based fluorescent ballasts could be replaced with a single EE25 inductor reducing the total cost and size of the final solution.

3.2 Lower Total Harmonic Distortion (THD)

The CS1601-based design has lower THD than the L6562-based solution. In the fluorescent ballast example, it can be seen that the THD across the line at full load was more than 2.5% lower than L6562 as shown in Figure 2.



% TH

Figure 2. Comparison of THD vs. AC Line Voltage

The L6562 cannot support large differential filter capacitors on the DC side of the brid ge rectifier. In CRM controllers, the maximum switching frequency at the trough of the AC line is theoretically infinity. The L6562 sets a limit on this maximum, increasing distortion of the AC line when a larger capacitor is placed across the rectified (DC) AC line. The L6562-based ballast necessitates expensive AC capacitors placed on the AC line side of the bridge rectifier, increasing the cost of the EMI filter. Since the THD in the CS1601-based solution is significantly lower, it allows the designer to reduce the EMI filter cost by moving a substantial portion of the differential filter capacitance to the DC side of the bridge rectifier. This eliminates the need for expensive AC capacitors and reduces overall bill of material (BOM) cost.

3.3 Light Load Performance

The flexibility of moving filters to either side of the rectifier also offers another significant advantage over the L6562 with respect to light-load PF and high-line-voltage THD. In traditional analog solutions, placing capacitors at the output of the bridge rectifiers increases THD, especially at light-load conditions and high line voltage. At the trough of the AC line, CRM controllers run at very high frequencies and switch intermittently to limit switching losses. Placing capacitors at the input side of the bridge reduces PF since the EMI capacitor swings to twice the input voltage. Power supply design ers have to make trade-offs between light load efficiency, power factor, and THD. The L6562 has errors and delays resulting from the multiplier, comparator, and gate drivers. These result in errors in peak current at light loads. In the trough these cause significant deterioration of THD performance. As a workaround to this problem, traditional CRM controllers have to greatly increase the inductance of the boost inductor to limit the slope of the current and limit the error in peak current that is caused. Since the CS1601 is a variable-frequency DC M control le r with different frequency profile, the limitations encountered when using the L6562 are not present in a system designed with the CS1601. The CS1601 offers optimal performance with minimal design constraints.

3.4 Near Unity Power Factor

Improved THD performance necessitates large AC capacitors for differential mode EMI filters. Because the voltage swing on capacitors on the AC line side of the bridge rectifier is twice that of the capacitors placed on the DC side of the bridge, power factor is reduced. Figure 3 compares the PF between the L6562-based ballast and the CS1601-based ballast. The CS1601 maintains

2 AN349REV1

AN349

PF Vs Line @ 60Hz

0.93

0.94

0.95

0.96

0.97

0.98

0.99

1.01

80 100 120 140 160 180 200 220 240 260 280 300

Inpu t Voltage (V)

CS1601 ST L6562

near unity power factor across the operating input voltage range. The CS1601-based solutions grea tly minimizes the need for making this trade-off. The CS1601 can provide very low distortion at near unity PF.



Figure 3. Comparison of Power Factor vs. AC Line Voltage

3.5 Lower Link Overshoot During Lamp Strike

The CS1601-based ballast demonstrates better performance during lamp strike and start-up. The scope capture of the lamp strike with the L6562 is shown in Figure 4. The scope capture of the lamp strike with the CS1601 is shown in Figure 5. In both the scope captures, the PFC output is presented in blue and the bridge rectifier output signal is presented in red. From Figure 4, it can be observed that with the L6562-based ballast, during lamp start-up, the link voltage rails up to 520V before settling down to the nominal value of 485V. In the CS1601-based ballast, the link voltage does not go above 492V, which is the voltage threshold for the over voltage protection (OVP). Since the output capacitors are rated for 500V (Two 47uF, 250V capacitors in series), the lifetime of these capacitors are improved by keeping the voltage below 500V under all conditions. The CS1601 requires a lower inductance for the same power level, resulting in less stored energy during startup. At the end of startup, the stored energy is transferred to the output bulk capacitor. Less stored energy results in a smaller V

Note: Neither solution produced any flicker during lamp strike.

overshoot.

link

AN349REV1 3

Figure 4. Ballast Startup with L6562

Figure 5. Ballast Startup with CS1601

COMP

VCC

GND

Ref2

Starter

Stop

Voltage

Regulator

Overvoltage

Detection

40k

Multiplier & THD

Optimizer

Disable

Driver

25V

RSQ

ZCD

INV

UVLO

Starter

2.1V

1.6V

Zero Current

Detector

15V

5pF

MULT

Internal

Supply 7V

2.5V

AN349

4. Theory of Operation

Figure 4 and Figure 5 are the internal block diagrams of the L6562 and the CS1601, respectively.

4 AN349REV1

Figure 6. L6562 Block Diagram

AN349

POR +

ST(th)

STP(th)

Voltage

Regulator

VDD

ZCD

ZCD(th)

Zero Cross

Detect

GND

IFB

IAC

LEB

15k

24k

15k

24k

ADC

ZCB

600

Threshold

CS Clamp

CS(clamp )

CS(t h)

STBY

600k

Swit ching Fre q uency of L6562 Acr o ss Lin e

0 45 90 135 180 225 270 315 360

Pha s e Angl e (degrees)

Swi tching Frequency (KHz)

Figure 7. CS1601 Block Diagram

The basic principle of operation of the L6562 can be summarized as follows:

- The L6562 is a CRM controller which sets a peak current reference which is a function of the line voltage multiplied with the output of the error amplifier.

- During the design phase, the maximum steady-state power is defined by the current-sense resistor setting the overcurrent threshold. The selection of inductor shifts the frequency profile across load conditions and line voltages.

- The zero-cross detect (ZCD) pin detects when the inductor current resets to zero.

The frequency of operation across the line voltage is shown in Figure 8.



Figure 8. L6562 Switching Frequency vs. Line Voltage at Low Line

The basic principle of operation of CS1601 can be summarized as follows :

- The CS1601 is a mixed DCM/CRM controller and is a voltage-mode IC which measures the output voltage and computes

AN349REV1 5

a given switching frequency scheme at every line voltage.

- The maximum power is specified by the selection of the inductor. The frequency is computed based on the line voltage.

- The CS1601 measures the line (V

- The EXL Core™ signal processing core computes the correct T

and link voltages and programmed PID coefficients.

- The EXL Core turns off th e ga te once T

- The EXL Core counts until T

- Figure 9 and Figure 10 show the frequency profile across the operating range for a CS1601-based solution.

time is reached or until zero-cross detection (ZCD) — whichever is first.

off

) and link (V

rect

is reached.

) voltages and compares them against a reference.

link

, T

, and switching frequency for the measured line

off

+ 9 hidden pages