SNVS682D November   2010  – December 2015 LM3444

PRODUCTION DATA.  

  1. Features
  2. Applications
  3. Description
  4. Revision History
  5. Pin Configuration and Functions
  6. Specifications
    1. 6.1 Absolute Maximum Ratings
    2. 6.2 ESD Ratings
    3. 6.3 Recommended Operating Conditions
    4. 6.4 Thermal Information
    5. 6.5 Electrical Characteristics
    6. 6.6 Typical Characteristics
  7. Detailed Description
    1. 7.1 Overview
    2. 7.2 Functional Block Diagram
    3. 7.3 Feature Description
      1. 7.3.1 Theory of Operation
      2. 7.3.2 Valley-Fill Circuit
      3. 7.3.3 Valley-Fill Operation
      4. 7.3.4 Buck Converter
      5. 7.3.5 Overview Of Constant Off-Time Control
      6. 7.3.6 Thermal Shutdown
    4. 7.4 Device Functional Modes
  8. Application and Implementation
    1. 8.1 Application Information
      1. 8.1.1  Determining Duty-Cycle (D)
      2. 8.1.2  Calculating Off-Time
      3. 8.1.3  Setting the Switching Frequency
      4. 8.1.4  Inductor Selection
      5. 8.1.5  Setting the LED Current
      6. 8.1.6  Valley Fill Capacitors
      7. 8.1.7  Determining the Capacitance Value of the Valley-Fill Capacitors
      8. 8.1.8  Determining Maximum Number of Series Connected LEDs Allowed
      9. 8.1.9  Output Capacitor
      10. 8.1.10 Switching MOSFET
      11. 8.1.11 Recirculating Diode
    2. 8.2 Typical Application
      1. 8.2.1 Design Requirements
      2. 8.2.2 Detailed Design Procedure
      3. 8.2.3 Application Curve
  9. Power Supply Recommendations
  10. 10Layout
    1. 10.1 Layout Guidelines
    2. 10.2 Layout Example
  11. 11Device and Documentation Support
    1. 11.1 Device Support
      1. 11.1.1 Third-Party Products Disclaimer
    2. 11.2 Community Resources
    3. 11.3 Trademarks
    4. 11.4 Electrostatic Discharge Caution
    5. 11.5 Glossary

パッケージ・オプション

メカニカル・データ(パッケージ|ピン)
サーマルパッド・メカニカル・データ
発注情報

8 Application and Implementation

NOTE

Information in the following applications sections is not part of the TI component specification, and TI does not warrant its accuracy or completeness. TI’s customers are responsible for determining suitability of components for their purposes. Customers should validate and test their design implementation to confirm system functionality.

8.1 Application Information

8.1.1 Determining Duty-Cycle (D)

Equation 4 shows the duty-cycle (D).

Equation 4. LM3444 30127529.gif

Equation 5 shows the duty-cycle with efficiency considered.

Equation 5. LM3444 30127530.gif

For simplicity, choose efficiency from 75% to 85%.

8.1.2 Calculating Off-Time

The off-time of the LM3444 is set by the user and remains fairly constant as long as the voltage of the LED stack remains constant. Calculating the off-time is the first step in determining the switching frequency of the converter, which is integral in determining some external component values.

PNP transistor Q3, resistor R4, and the LED string voltage define a charging current into capacitor C11. A constant current into a capacitor creates a linear charging characteristic.

Equation 6. LM3444 30127531.gif

Resistor R4, capacitor C11 and the current through resistor R4 (iCOLL), which is approximately equal to VLED/R4, are all fixed. Therefore, dv is fixed and linear, and dt (tOFF) can now be calculated as shown in Equation 7.

Equation 7. LM3444 30127532.gif

Common equations for determining duty-cycle and switching frequency in any buck converter are shown in Equation 8.

Equation 8. LM3444 30127533.gif

Therefore, Equation 9 shows:

Equation 9. LM3444 30127534.gif

With efficiency of the buck converter in mind, Equation 10 shows:

Equation 10. LM3444 30127535.gif

Substituting and rearranging the equations, Equation 11 shows:

Equation 11. LM3444 30127536.gif

Off-time and switching frequency can now be calculated using the previous equations.

8.1.3 Setting the Switching Frequency

Selecting the switching frequency for nominal operating conditions is based on tradeoffs between efficiency (better at low frequency) and solution size and cost (smaller at high frequency).

The input voltage to the buck converter (VBUCK) changes with both line variations and over the course of each half-cycle of the input line voltage. The voltage across the LED string, however, remains constant, and therefore the off-time remains constant.

The on-time, and therefore the switching frequency, varies as the VBUCK voltage changes with line voltage. A good design practice is to choose a desired nominal switching frequency knowing that the switching frequency decreases as the line voltage drops, and increases as the line voltage increases (Figure 14).

LM3444 30127510.gif Figure 14. Graphical Illustration of Switching Frequency vs VBUCK

The off-time of the LM3444 can be programmed for switching frequencies ranging from 30 kHz to over 1 MHz. A trade-off between efficiency and solution size must be considered when designing the LM3444 application.

The maximum switching frequency attainable is limited only by the minimum on-time requirement (200 ns).

Worst case scenario for minimum on time is when VBUCK is at its maximum voltage (AC high line) and the LED string voltage (VLED) is at its minimum value, as shown in Equation 12.

Equation 12. LM3444 30127538.gif

The maximum voltage seen by the Buck Converter is given by Equation 13.

Equation 13. LM3444 30127539.gif

8.1.4 Inductor Selection

The controlled off-time architecture of the LM3444 regulates the average current through the inductor (L2), and therefore the LED string current. The input voltage to the buck converter (VBUCK) changes with line variations and over the course of each half-cycle of the input line voltage. The voltage across the LED string is relatively constant, and therefore the current through R4 is constant. This current sets the off-time of the converter and therefore the output volt-second product (VLED × off-time) remains constant. A constant volt-second product makes it possible to keep the ripple through the inductor constant as the voltage at VBUCK varies.

LM3444 30127540.gif Figure 15. LM3444 External Components of the Buck Converter

Use Equation 14 to calculate an ideal inductor.

Equation 14. LM3444 30127541.gif

Given a fixed inductor value, L, Equation 14 states that the change in the inductor current over time is proportional to the voltage applied across the inductor.

During the on-time, the voltage applied across the inductor is given in Equation 15.

Equation 15. VL(ON-TIME) = VBUCK - (VLED + VDS(Q2) + IL2 × R3)

Because the voltage across the MOSFET switch (Q2) is relatively small, as is the voltage across sense resistor R3, we can approximately simplify this as shown in Equation 16,

Equation 16. VL(ON-TIME) = VBUCK - VLED

During the off-time, the voltage seen by the inductor is given by Equation 17.

Equation 17. VL(OFF-TIME) = VLED

The value of VL(OFF-TIME) is relatively constant, because the LED stack voltage remains constant. If we rewrite the equation for an inductor inserting what we know about the circuit during the off-time, Equation 18 shows that we get:

Equation 18. LM3444 30127542.gif

Rearranging this gives Equation 19.

Equation 19. LM3444 30127543.gif

From this, we can see that the ripple current (Δi) is proportional to off-time (tOFF) multiplied by a voltage, which is dominated by VLED divided by a constant (L2).

These equations can be rearranged to calculate the desired value for inductor L2, as shown in Equation 20.

Equation 20. LM3444 30127544.gif

The off time can be calculated using Equation 21:

Equation 21. LM3444 30127545.gif

Substituting toff in Equation 21 results in Equation 22:

Equation 22. LM3444 30127546.gif

See Typical Application to better understand the design process.

8.1.5 Setting the LED Current

The LM3444 constant off-time control loop regulates the peak inductor current (IL2). The average inductor current equals the average LED current (IAVE). Therefore the average LED current is regulated by regulating the peak inductor current.

LM3444 30127525.gif Figure 16. Inductor Current Waveform in CCM

Knowing the desired average LED current, IAVE, and the nominal inductor current ripple, ΔiL, the peak current for an application running in continuous conduction mode (CCM) is defined in Equation 23.

Equation 23. LM3444 30127548.gif

The LED current would then be calculated using Equation 24.

Equation 24. LM3444 30127549.gif

This is important to calculate because this peak current multiplied by the sense resistor R3 determines when the internal comparator is tripped. The internal comparator turns the control MOSFET off once the peak sensed voltage reaches 750 mV.

Equation 25. LM3444 30127550.gif

Current Limit: The trip voltage on the PWM comparator is 750 mV. However, if there is a short circuit or an excessive load on the output, higher than normal switch currents cause a voltage greater than 1.27 V on the ISNS pin which trip the I-LIM comparator. The I-LIM comparator resets the RS latch, turning off Q2. It also inhibits the Start Pulse Generator and the COFF comparator by holding the COFF pin low. A delay circuit prevents the start of another cycle for 180 µs.

8.1.6 Valley Fill Capacitors

Determining voltage rating and capacitance value of the valley-fill capacitors:

The maximum voltage seen by the valley-fill capacitors is calculated by Equation 26.

Equation 26. LM3444 30127551.gif

This is, of course, if the capacitors chosen have identical capacitance values and split the line voltage equally. Often a 20% difference in capacitance could be observed between like capacitors. Therefore a voltage rating margin of 25% to 50% should be considered.

8.1.7 Determining the Capacitance Value of the Valley-Fill Capacitors

The valley-fill capacitors must be sized to supply energy to the buck converter (VBUCK) when the input line is less than its peak divided by the number of stages used in the valley fill (tX). The capacitance value must be calculated for the maximum LED current.

LM3444 30127552.gif Figure 17. Two Stage Valley-Fill VBUCK Voltage

From Figure 17 and the equation for current in a capacitor, i = C × dV/dt, the amount of capacitance needed at VBUCK is calculated as follows.

At 60 Hz, and a valley-fill circuit of two stages, the hold-up time (tX) required at VBUCK is calculated as follows. The total angle of an AC half cycle is 180° and the total time of a half AC line cycle is 8.33 ms. When the angle of the AC waveform is at 30° and 150°, the voltage of the AC line is exactly ½ of its peak. With a two-stage valley-fill circuit, this is the point where the LED string switches from power being derived from AC line to power being derived from the hold up capacitors (C7 and C9). 60° out of 180° of the cycle or 1/3 of the cycle the power is derived from the hold up capacitors (1/3 × 8.33 ms = 2.78 ms). This is equal to the hold up time (dt) from the previous equation, and dv is the amount of voltage the circuit is allowed to droop. From Determining Maximum Number of Series Connected LEDs Allowed, we know the minimum VBUCK voltage is about 45 V for a
90-VAC to 135-VAC line. At 90-VAC low-line operating condition input, ½ of the peak voltage is 64 V. Thus, with some margin, the voltage at VBUCK can not droop more than about 15 V (dv). (i) is equal to (POUT/VBUCK), where POUT is equal to (VLED × ILED). Total capacitance (C7 in parallel with C9) can now be calculated. See Typical Application for further calculations of the valley-fill capacitors.

8.1.8 Determining Maximum Number of Series Connected LEDs Allowed

The LM3444 is an off-line buck topology LED driver. A buck converter topology requires that the input voltage (VBUCK) of the output circuit must be greater than the voltage of the LED stack (VLED) for proper regulation. One must determine what the minimum voltage observed by the buck converter is before the maximum number of LEDs allowed can be determined. The following two variables must be determined to accomplish this:

  1. AC line operating voltage. This is usually 90 VAC to 135 VAC for North America. Although the LM3444 can operate at much lower and higher input voltages, a range is needed to illustrate the design process.
  2. The number of stages implemented in the valley-fill circuit (1, 2, or 3).

In this example, the most common valley-fill circuit is used (two stages).

LM3444 30127554.gif Figure 18. AC Line

Figure 18 shows the AC waveform. One can easily see that the peak voltage (VPEAK) is always given by Equation 27.

Equation 27. LM3444 30127553.gif

The voltage at VBUCK with a valley-fill stage of two looks similar to the waveforms in Figure 17.

The purpose of the valley-fill circuit is to allow the buck converter to pull power directly off of the AC line when the line voltage is greater than its peak voltage divided by two (two-stage valley-fill circuit). During this time, the capacitors within the valley fill circuit (C7 and C8) are charged up to the peak of the AC line voltage. Once the line drops below its peak divided by two, the two capacitors are placed in parallel and deliver power to the buck converter. One can now see that if the peak of the AC line voltage is lowered due to variations in the line voltage, the DC offset (VDC) lowers. VDC is the lowest value that voltage VBUCK encounters.

Equation 28. LM3444 30127557.gif

Example:

Line voltage = 90 VAC to 135 VAC

Valley-fill = two stage

Equation 29. LM3444 30127558.gif

Depending on what type and value of capacitors are used, some derating should be used for voltage droop when the capacitors are delivering power to the buck converter. With this derating, the lowest voltage the buck converter sees is about 42.5 V in this example.

To determine how many LEDs can be driven, take the minimum voltage the buck converter sees (42.5 V) and divide it by the worst-case forward voltage drop of a single LED.

Example: 42.5 V / 3.7 V = 11.5 LEDs (11 LEDs with margin)

8.1.9 Output Capacitor

A capacitor placed in parallel with the LED or array of LEDs can be used to reduce the LED current ripple while keeping the same average current through both the inductor and the LED array. With a buck topology, the output inductance (L2) can now be lowered, making the magnetics smaller and less expensive. With a well designed converter, you can assume that all of the ripple is seen by the capacitor, and not the LEDs. One must ensure that the capacitor you choose can handle the RMS current of the inductor. See the manufacturer data sheets to ensure compliance. Usually an X5R or X7R capacitor from 1 µF and 10 µF of the proper voltage rating is sufficient.

8.1.10 Switching MOSFET

The main switching MOSFET should be chosen with efficiency and robustness in mind. As shown in Equation 30, the maximum voltage across the switching MOSFET equals:

Equation 30. LM3444 30127559.gif

The average current rating should be greater than what is given in Equation 31.

Equation 31. IDS-MAX = ILED(-AVE)(DMAX)

8.1.11 Recirculating Diode

The LM3444 buck converter requires a recirculating diode D10 (see Figure 8) to carry the inductor current during the MOSFET Q2 off-time. The most efficient choice for D10 is a diode with a low forward drop and near-zero reverse recovery time that can withstand a reverse voltage of the maximum voltage seen at VBUCK. For a common 110 VAC ± 20% line, the reverse voltage could be as high as 190 V, as shown in Equation 32.

Equation 32. LM3444 30127560.gif

As shown in Equation 33, the current rating must be at least:

Equation 33. ID = (1 - DMIN) × ILED(AVE)

Or as shown in Equation 34:

Equation 34. LM3444 30127561.gif

8.2 Typical Application

The following design example illustrates the process of calculating external component values.

LM3444 30127569.gif Figure 19. LM3444 Design Example 1 Input = 90 VAC to 135 VAC
VLED = 7 × HB LED String Application at 400 mA

8.2.1 Design Requirements

Known:

  1. Input voltage range (90 VAC to 135 VAC)
  2. Number of LEDs in series = 7
  3. Forward voltage drop of a single LED = 3.6 V
  4. LED stack voltage = (7 × 3.6 V) = 25.2 V

Choose:

  1. Nominal switching frequency, fSW-TARGET = 350 kHz
  2. ILED(AVE) = 400 mA
  3. Δi (usually 15% to 30% of ILED(AVE)) = (0.30 × 400 mA) = 120 mA
  4. Valley-fill stages (1, 2, or 3) = 2
  5. Assumed minimum efficiency = 80%

8.2.2 Detailed Design Procedure

Calculate:

  1. Calculate minimum voltage VBUCK, as shown in Equation 35, which yields:
  2. Equation 35. LM3444 30127558.gif
  3. Calculate maximum voltage VBUCK, as shown in Equation 36, which yields:
  4. Equation 36. LM3444 30127563.gif
  5. Calculate tOFF at VBUCK nominal line voltage, as given by Equation 37.
  6. Equation 37. LM3444 30127564.gif
  7. Calculate tON(MIN) at high line to ensure that tON(MIN) > 200 ns, as given by Equation 38.
  8. Equation 38. LM3444 30127565.gif
  9. Calculate C11 and R4 in steps 6 through 9.
  10. Choose current through R4 (from 50 µA to 100 µA): 70 µA as given by Equation 39.
  11. Equation 39. LM3444 30127566.gif
  12. Use a standard value of 365 kΩ.
  13. Calculate C11 as given by Equation 40.
  14. Equation 40. LM3444 30127567.gif
  15. Use standard value of 120 pF.
  16. Calculate ripple current: 400 mA × 0.30 = 120 mA
  17. Calculate inductor value at tOFF = 3 µs as given by Equation 41.
  18. Equation 41. LM3444 30127568.gif
  19. Choose C10: 1 µF, 200 V
  20. Calculate valley-fill capacitor values:
    VAC low line = 90 VAC, VBUCK minimum equals 60 V. Set droop for 20-V maximum at full load and low line as shown in Equation 42.
  21. Equation 42. LM3444 30127531.gif

    where

    • i equals POUT/VBUCK (270 mA)
    • dV equals 20 V
    • dt equals 2.77 ms
    • CTOTAL equals 37 µF

    Therefore, C7 = C9 = 22 µF

8.2.3 Application Curve

LM3444 30127505.gif Figure 20. Efficiency vs Input Voltage

Table 1. Bill of Materials

QTY DESIGNATOR DESCRIPTION MANUFACTURER MANUFACTURER PART NUMBER
1 U1 IC, CTRLR, DRVR-LED, VSSOP TI LM3444MM
1 BR1 Bridge Rectifiier, SMT, 400 V, 800 mA DiodesInc HD04-T
1 L1 Common mode filter DIP4NS, 900 mA, 700 µH Panasonic ELF-11090E
1 L2 Inductor, SHLD, SMT, 1 A, 470 µH Coilcraft MSS1260-474-KLB
2 L3, L4 Diff mode inductor, 500 mA 1 mH Coilcraft MSS1260-105KL-KLB
1 L5 Bead Inductor, 160 Ω, 6 A Steward HI1206T161R-10
3 C1, C2, C15 Cap, Film, X2Y2, 12.5 MM, 250 VAC, 20%, 10 nF Panasonic ECQ-U2A103ML
1 C4 Cap, X7R, 0603, 16 V, 10%, 100 nF Murata GRM188R71C104KA01D
2 C5, C6 Cap, X5R, 1210, 25 V, 10%, 22 µF Murata GRM32ER61E226KE15L
2 C7, C9 Cap, AL, 200 V, 105C, 20%, 33 µF UCC EKXG201ELL330MK20S
1 C10 Cap, Film, 250 V, 5%, 10 nF Epcos B32521C3103J
1 C12 Cap, X7R, 1206, 50 V, 10%, 1 µF Kemet C1206F105K5RACTU
1 C11 Cap, C0G, 0603, 100 V, 5%, 120 pF Murata GRM1885C2A121JA01D
1 D1 Diode, ZNR, SOT23, 15 V, 5% OnSemi BZX84C15LT1G
2 D2, D13 Diode, SCH, SOD123, 40 V, 120 mA NXP BAS40H
4 D3, D4, D8, D9 Diode, FR, SOD123, 200 V, 1 A Rohm RF071M2S
1 D10 Diode, FR, SMB, 400 V, 1 A OnSemi MURS140T3G
1 D12 TVS, VBR = 144 V Fairchild SMBJ130CA
1 R2 Resistor, 1206, 1%, 100 kΩ Panasonic ERJ-8ENF1003V
1 R3 Resistor, 1210, 5%, 1.8Ω Panasonic ERJ-14RQJ1R8U
1 R4 Resistor, 0603, 1%, 576 kΩ Panasonic ERJ-3EKF5763V
2 R6, R7 Resistor, 0805, 1%, 1 MΩ Rohm MCR10EZHF1004
2 R8, R10 Resistor, 1206, 0 Ω Yageo RC1206JR-070RL
1 RT1 Thermistor, 120 V, 1.1 A, 50 Ω at 25°C Thermometrics CL-140
2 Q1, Q2 XSTR, NFET, DPAK, 300 V, 4 A Fairchild FQD7N30TF
1 Q3 XSTR, PNP, SOT23, 300 V, 500 mA Fairchild MMBTA92
1 J1 Terminal Block 2 pos Phoenix Contact 1715721
1 F1 Fuse, 125 V, 1.25 A bel SSQ 1.25