Cirrus Logic AN374 User Manual

AN374

Application Note

Considerations for Light Engine Selection

for a CS1630/31 2-Channel TRIAC Dimmable Circuit

1 Overview of the CS1630

The CS1630/31 is a high-performance offline AC to DC LED controller for dimmable and high color rendering index (CRI) LED replacement lamps and luminaires. It features Cirrus Logic's proprietary digital dimmer compatibility control technology and digital correlated color temperature (CCT) control system that enables two-channel LED color mixing.

The CS1630/31 offers tremendous flexibility to achieve constant CCT control to match an incandescent dimming profile. The digital CCT control system provides the ability to program dimming profiles, such as constant CCT dimming and black body line dimming. The CS1630/ 31 optimizes LED color mixing by temperature compensating the LED current with an external negative temperature coefficient (NTC) thermistor. The IC controller is equipped with power line calibration for remote system calibration and end-of-line programming. The CS1630 provides a register lockout feature for security against potential access to proprietary registers.

During the course of two-channel design, several design tradeoffs are made based on cost, size, and performance. This document considers the requirements when designing a system around the CS1630/31. This document helps answer the following system questions:

• How can the light engine be designed or modified to maximize the benefits of the CS1630/31?

• How can clear specifications be derived for the system and the LED driver that can reduce the light bulb

development time with the CS1630/31?

2 Introduction

This application note provides a guide to designing a solid-state lighting (SSL) LED lamp circuit using Cirrus Logic's CS1630/31. The first half of the document presents an introduction to the CS1630/31 color control system and the criterion for selecting a compatible light engine. The second half of the document supports the design effort by detailing the requirements to curve fit the polynomial gain equations for a robust color correlation temperature solution.

2.1 Definition of Acronyms

Acronym Description

PFC Power Factor Correction

ZCD Zero-current Detection

BOP Boost Overvoltage Protection

COP Clamp Overpower Protection

OVP Second-stage Output Open Circuit Protection and Overvoltage Protection

OCP Second-stage Overcurrent Protection

OLP Second-stage Open Loop Protection

SCP Short Circuit Protection

iOTP Internal Overtemperature Protection

eOTP External Overtemperature Protection

PLC Power Line Calibration

OTP One-time Programmable

LED Light Emitting Diode

TX Transformer

TRIAC

NTC

SSL

CSV Comma-separated Values File

CCT Correlated Color Temperature

DAC Digital-to-Analog Converter

TRIode for Alternating Current, which is an electronic component that can conduct current in either direction when it is triggered. It is formally called a bidirectional triode thyristor.

Negative Temperature Coefficient thermistor

Solid-state lighting. Refers to a type of lighting that uses semiconductor LEDs as a source of illumination rather than electrical filaments, plasma, or gas.

4 AN374REV2

2.2 Definition of Symbols

CH1

----------------





CH2

----------------





Symbol Description

& F

sw1

sw2

TT Second-stage switching period

& TT

CH1

& T1

& T2

& T3

CH2

Second-stage switching frequency

Switching frequency for channel 1 and channel 2

Switching period for channel 1 and channel 2

Second-stage primary FET ‘ON’ time for channel 1 and channel 2

Second-stage secondary rectifier diode conduction time for channel 1 and channel 2

Time the second-stage FET and rectified diode are ‘OFF’ for channel 1 and channel 2

AN374

MODE1

PK1(FB)

MODE1

GAIN

& D

Reflected

CLAMP

PK(FB)

& I

& V

Sense

& T

NTC

PK(BST)

BST

CH1

CH2

& I

CH1

OUT

Red

White

dim

& GAIN

MODE2

PK2(FB)

MODE2

NTC

CH2

MODE2

DTR

Duty ratio for Mode 1 and Mode 2

Input line voltage

Voltage across secondary winding reflected onto primary

Primary clamping voltage above boost output voltage (V

BST

)

Maximum second-stage peak current in primary-side FET

Maximum second-stage peak current in primary-side FET for Mode 1 and Mode 2

Output current for Mode 1 and Mode 2

Output voltage for Mode 1 and Mode 2

Second-stage primary current sense resistor

Negative temperature coefficient resistance and corresponding temperature

Maximum boost inductor current

Second-stage primary inductance

Boost inductance

Boost output voltage

Flyback transformer turns ratio N

P/NS

Channel 1 secondary output VDC (channel 1 LED string supply voltage)

Channel 2 secondary output VDC (channel 2 LED string supply voltage)

Channel 1 and channel 2 LED string current

Load power = Power to the LED string

Output current that flows through the amber/red color LED string

Output current that flows through the white/blue color LED string

The CS1630/31 color control system has the ability to maintain a constant CCT or change CCT as the light dims. OTP configurations allow the selection of the dimming profile. A specific CCT profile can be programmed to the digital mapping device. The mapping is two-dimensional: one current versus temperature profile is generated for each dim level. The CS1630/31 provides two-dimensional mapping for the color LED’s current only, and one-dimensional mapping (current versus dim level) for the other string.

The dim-regulated gain and dim-regulated plus temperature-regulated gain

AN374REV2 5

AN374

dim

NTC

(From Boost)

÷ 4096

÷ 256

Normalize

Saturation

Logic

GAIN

= Q3 × D3 + Q2 × D2 + Q1 × D + Q0

GAIN

DTR

= P30 × T3 + P20 × T2 + P10 × T + P03 × D3 + P02 × D2 + P01 × D +

P21 × T

× D + P12 × T × D2 + P11 × T × D + P00

White(ref)

Red(ref)

White

Red

dim

Temperature

(ADC Fast Filter)

Figure 1. Color Control System

GAIN

DTR

P30 T3 P20 T2 P10++T P03 D3 P02++D2 P01 D P21 T2D++= P12 T D2 P11 T D P00+++

GAIN

Q3= D3 Q2 D2 Q1 D Q0+++

[Eq. 1]

3 Introduction to the Color System

The CS1630/31 is a two-channel TRIAC dimmable LED driver IC designed to change the color temperature of the light output by independently varying the gains of the two different color LED strings to establish levels of color mixing. This feature can be used to make the color temperature versus dim characteristics of the light similar to that of an incandescent light bulb. In many such designs, one of the LED strings is composed of red or amber LEDs, and the other string is composed of cool-white or blue-white LEDs. While the lumen output of white LEDs does not vary significantly across temperature, the lumen output of red LEDs can vary as much as 40% across temperature. To achieve a consistent light output across temperature, the current in the red LED string needs to be compensated with respect to temperature. Depending on the design, either channel can be temperature compensated.

Cirrus Logic, Inc. and its affiliates and subsidiaries generally make no representations or warranties that the combination of Cirrus Logic’s products with light-emitting diodes (“LEDs”), converter materials, and/or other components will not infringe any third-party patents, including any patents related to color mixing in LED lighting applications, such as, for example, U.S. Patent No. 7,213,940 and related patents of Cree, Inc. For more information, please see Cirrus Logic’s Terms and Conditions of Sale, or contact a Cirrus Logic sales representative.

Figure 1 illustrates the block diagram of the color control block inside of the CS1630. The color temperature of the light engine can be modified by changing the gains of each channel based on the current dim level. On the temperature-controlled channel, the currents can be varied according to the temperature sensed by an external NTC.

The required gain value for a particular combination of dim and temperature is obtained using polynomial curves, the coefficients of which are programmed into the CS1630/31 OTP memory. A different polynomial is used for each channel. One of these is a polynomial in two variables, dim and temperature, while the other is a polynomial in dim only. If D and T are assumed to be the normalized dim and temperature values, respectively, between 0 and 1.0, then GAIN

refers to the dim-regulated gain and temperature-regulated gain, and GAINDR refers to the dim-reg-

DTR

ulated gain.

As shown in Equation 1, the gain equation for the white is a third-order polynomial in dim, and the gain equation for the red is a third-order polynomial in both dim and temperature. The color gain is third order to provide good tradeoffs between computational overhead and being able to operate over the largest variety of LEDs across a wide range of temperature. A lower-order polynomial fit, such as a quadratic or a linear fit, would not allow a large range of gain values across the entire operating range. This would limit the sample space of available LEDs, since the gain is an indirect reflection of its variation across temperature and required dim. As a result, a third-order fit allows the system engineer to achieve a large variation in CCT and lumens across dim.

6 AN374REV2

AN374

RedIRed ref

dim GAIN

DTR

=

[Eq. 2]

WhiteIWhite ref

dim GAIN

=

[Eq. 3]

4 Light Engine Choices

This section examines key design considerations for a complete lamp design using the CS1630/31 two-channel TRIAC dimmable driver IC. It provides some of the appropriate design choices based on a comprehensive understanding of the CS1630/31 driver and digital control algorithms.

4.1 Considerations for the Color System

Constraint 1: Translating current versus dim requirements into a fourth-order or lower polynomial

Numerous smooth curves that do not have asymptotes at the origin can be modeled as a polynomial function. The few exceptions are notable functions where the output current equals log converter design requires a light engine that has a color curve that trends to any of these functions, then some design tradeoffs must be considered. Such output current profiles are not widely prevalent in the consumer or commercial lighting applications in which two-channel LEDs are typically used.

Constraint 2: The current versus dim plot should have intercept at origin

The final output currents for red current I

and white current I

Red

are calculated using Equations 2 and 3:

White

(dim) or 1/dim. If the power

These equations show that the current versus the dim will tend to the origin even for non-zero coefficients P00 and Q0.

The dim axis is an imaginary axis with no physical significance to an LED designer. This constraint causes problems that can be substantially mitigated by remapping the dim axis, as demonstrated in section Manipulating Data for Better Curve Fit on page 27.

Constraint 3: Placement of NTC with respect to light engine

The NTC can be placed close to the LEDs when the LEDs represent the junction temperature of the LEDs or it can be placed farther away from the LEDs, where it represents the heat sink temperature or even the ambient temperature inside the bulb. Each placement comes with its own advantages, and Steps 1 and 2 below describe the implications of placing the LEDs in either location.

Step 1) Advantages of placing the NTC close to the LEDs

1. The problem of thermal mass is greatly reduced, which allows using simpler temperature protection and thermal fold back.

2. Since the NTC is a good representation of the junction temperature, it is simpler to make light-system models that can be used for designing the light engine.

3. Temperature protection systems such as thermal fold back and thermal shutdown can be more accurately designed if needed, improving reliability in such applications.

Step 2) Disadvantages of placing the NTC close to the LEDs

1. The

 of the NTC varies across temperature, and a narrow temperature range produces a more accurate

linear approximation. By having a large temperature rise on the NTC, the NTC is no longer accurate, since the CS1630/31 can only accept a single

 value and a single expression for the temperature.

2. If the thermal impedance between the NTC and the LED junction is large, the NTC will not accurately represent the LED junction temperature, and the red current I

compensation may be incorrect. For

Red

example, when the NTC is placed close to the LED strings, then the measured temperature is a close representation of the LED junction temperature. In a tight form factor and at full brightness, the temperature rise from ambient to thermal equilibrium is significant. Typically, LED strings can take up to a few minutes before they reach thermal equilibrium. When the system is started at ambient temperature T

= 25C and full brightness, the NTC temperature and the junction temperature of the LEDs are at

amb

C. Since the curve fitter produces a gain function that is dependent on a combination of the red current

AN374REV2 7

AN374

PK1 FBVBST

CH1

----------------





 200V

5.3s

3543H

---------------------





 299m A== =

[Eq. 4]

MODE1 avgIPK1 FB

CH1

2TT

----------------





 299m

A 5.57

9.0s

2 35.05s

------------------------------





 214m A== =

[Eq. 5]

, NTC, and dim variables, the curve fitter will have to compute gains across a larger number of points,

Red

which creates the risk of increased error. Failure to compensate for these points may cause a color shift during startup until the LED strings reach thermal equilibrium.

Constraint 4: Red currents increase with an increase in ambient temperature

The current I value.

in the red channel should increase monotonically with ambient temperature at a given dim

Red

Ambient Temperature

C300mA

C330mA

C360mA

Red Current Flowing in

Current Compensated Channel

Table 1. Maximum Measurable Switching Period

For example, if the current at ambient temperature T T

= 25C, then the currents are non-monotonic with ambient temperature.

amb

= 50C is less than the current at ambient temperature

amb

Constraint 5: Maximum allowable color gain is 4.0

The recommended maximum gain value that the color system generates is 4.0. For example, the red current at 25

C cannot be four times larger or smaller than any other current at the same dim value.

4.2 Constraints Imposed by Second Stage

Constraints in this section are imposed by the second stage. Calculations of these parameters are detailed in application note AN368 Design Guide for a CS1630/31 2-Channel TRIAC Dimmable SSL Circuit. All variables referenced in this document are computed in application note AN368.

Constraint 1: Currents in either channel should not be zero at non-zero dim values

Calculate peak current I

PK1(FB)

during Mode 1 using Equation 4:

Calculate the average current I

MODE1(avg)

during Mode 1 using Equation 5:

Equations 4 and 5 show that both channel currents share the same peak currents. It is difficult to have zero current in one channel and have a non-zero current in the other channel. As a result, currents must be greater than zero at all non-zero dim values.

Constraint 2: No abrupt change in current ratios at a dim point

It is recommended to avoid any abrupt change in the ratio of two-channel currents because it may cause color flicker. The ratio of the channel current and the actual value of the string current should be a smooth function which does not have any abrupt changes that become visible.

8 AN374REV2

AN374

PK1 FB

PK2 FB

-------------------

MODE1

MODE2

--------------------=

[Eq. 6]

Constraint 3: Ratio of peak currents between the two channels must be less than 4

The DAC reference on the current sense comparator has a 1.4V full-scale threshold. There is a minimum peak current I ratio of the peak currents is 5.6. However, a range is required to dim down to the lower dimming settings. Hence a ratio of 1.5 is considered optimum. Any peak ratio beyond 2.5 is considered a difficult requirement for the driver to guarantee regulation across the dimming range. If the resonant reset time T3 is short and can be neglected, then the ratio of the peak currents can be given as follows:

Refer to application note AN368 for detailed calculation.

Constraint 4: Both strings cannot have identical configuration

Either the output voltage differential should be maintained, or the output current differential should be maintained. If the strings are placed in series, then the currents should never be the same at any point of the dimming curve. If the strings are placed in parallel, then the voltage cannot be the same. The currents can cross in a parallel load configuration. In a commercial lighting space that requires mixing white LEDs of different wavelengths to produce a higher CCT white light (>3000K), it is recommended to have a different number of white LEDs in the two LED strings when using the CS1630/31.

Constraint 5: Maximum differential between string voltages

The overvoltage protection (OVP) threshold is set to 1.25V and the FBSENSE comparator threshold is set at a 200mV. As the dimmer conduction angle is reduced, the flyback operates continuously in DCM, while maintaining valley switching for low losses. At very low conduction angles, the flyback may switch on the third valley. Therefore, a minimum signal of 400mV on the FBAUX pin is recommended, which means that the string voltage of one channel cannot be greater than three times the string voltage of the other channel in a parallel load configuration. The FBAUX pin is used exclusively for overvoltage protection. In the event that a higher string voltage differential is desired, an external circuit must be designed such that different divider ratios are chosen for both modes of operation.

The OVP point is fixed for the higher of the two strings. For example, if one string has an output voltage of 5V and the other string has an output voltage of 15V, the OVP point is still considered to be at greater than 15V. Therefore, both output capacitors must be rated for the OVP voltage, which is in this case greater than 15V.

at full brightness threshold that is usually set at approximately 25%, so the theoretical maximum

PK(FB)

Constraint 6: The implications when determining LED string configuration and topology

Identify the light engine and select an appropriate second stage topology. There are two aspects that need to be considered: first, whether to use the series or parallel LED configuration and second, which power train topology to use.

AN374REV2 9

AN374

Figure 2a. Flyback Series Output Model Figure 2b. Flyback Parallel Output Model

R22

R21

R23

CS1630 /31

FBAUX

GND

FBSEN SE

TX1

BST

C10

C15

GND

_ Q

VCC

D15

R12 D10

C16

Channel 1 LED (White)

Channel 2 LED (Red)

GND

IGND

MODEx

PRI

MODEx

R22

R21

R23

CS1630 /31

FBAUX

GND

FBSEN SE

TX1

BST

C10

C15

GND

_ Q

VCC

D15

R12 D10

C16

Channel 1 LED (Whi te )

Channel 2 LED (Red)

GND

IGND

Step 1) LED string configuration

Determine if a series configuration or a parallel configuration is a viable solution for the identified light engine. Figure 2a illustrates a series light configuration. The two LED strings are arranged in series so that current passes through either one or both LED strings. A MOSFET is used to shunt current around one string on alternating switching cycles. In this configuration, one string is required to have a larger current than the other string. When considering a series design, it is recommended that the current flowing through one of the LED channels be 80% or lower than that of the other LED channel at all times. The LED string that has current flowing continuously is referred to as channel 1 LED (I as channel 2 LED (I

CH2

); I

CH2

 0.8I

CH1

), while the string with the bypass FET is referred to

CH1

Figure 2b illustrates a parallel light configuration. The two LED strings are networked in parallel so that current flows to either the channel 1 LED string or the channel 2 LED string at any given T2 time. The CS1630/31 controller uses the LED forward voltage to detect which LED string is being driven. One LED string must always have a larger forward voltage compared to the other LED string. The LED string with the higher voltage is referred to as channel 1 LED with forward voltage V to as channel 2 LED with a forward voltage of V

CH2

and the LED string with the lower voltage is referred

CH1

A good rule of thumb is that channel 2 LED must always have a forward voltage of 85% or lower than the channel 1 LED to consider a parallel design. Table 2 defines the selection process based on the requirements of the series and parallel output configuration.

< V

CH1

CH2

and I

CH2

< I

CH1

Cross Parallel Configuration Not Functional

CH2

Series or Parallel Configuration Series Configuration

CH1

Table 2. Series vs. Parallel

If there are problems converging on a target design using an existing light engine, the current and voltage profile can be modified by adding, removing, or moving LEDs between the two LED channels. Figure 3 illustrates the parallel and series scenarios that can be configured using the CS163X Customer Application tool.

10 AN374REV2

CH1

and V

CH2

Cross

AN374

Step 2a:

Select Series

vs. Parallel

Step 2b:

Select Flyback,

Buck, or

Tapped Buck

Figure 3. Second Stage String Configuration

Bit STRING in register Config3 at address 7 selects the second stage output channel configuration. When bit STRING is set to ‘1’ a series configuration is selected. Figure 3 illustrates the process used to select the second stage flyback mode using the CS163X system design application.

Step 2) Select power train topology

The CS1630/31 supports three possible power train configurations: tapped buck, buck, and flyback (see Figure 4a, 4b, and 4c). The two most important factors for selecting a power train configuration is whether the output requires Underwriters Laboratory (UL) approved isolation and the input to output voltage ratio. The flyback power train can be either isolated or non-isolated. Buck and tapped-buck designs are expected to always be non-isolated. If isolation is not required, one of the three possible solutions must be selected. If isolation is required, the design will be a flyback.

AN374REV2 11

AN374

Figure 4a. Tapped Buck Series Output Model Figure 4b. Buck Series Output Model

R22

R21

R23

CS1630 /31

FBAUX

GND

FBSENSE

BST

C10

C15

D11

GND

_ Q

VCC

R12 D10

C14

Channel 1 LED (White)

Channel 2 LED (Red)

GND

IGND

R22

R21

R23

CS1630 /31

FBAUX

GND

FBSENSE

BST

R25

C13

C15

C16

D11

GND

_ Q

VCC

R26 D10

R27

C14

Channel 1 LED (Whi te )

Channel 2 LED (Red)

GND

IGND

Figure 4c. Flyback Series Output Model

R22

R21

R23

CS1630 /31

FBAUX

GND

FBSENSE

TX1

BST

C10

C15

GND

VCC

D15

R12 D10

C16

Channel 1 LED (White)

Channel 2 LED (Red)

GND

IGND

MODEx

PRI

MODE x

Flyback topology is enabled by setting bit S2CONFIG to ‘1’ in register Config12 at address 44. A flyback topology is selected as a design guide example in application note AN368 so bit S2CONFIG is set to ‘1’. The flyback transformer input-to-output voltage ratio is used to determine the duty cycle and minimum turn on time T1 for the power FET. Buck topology is enabled by setting bit S2CONFIG to ‘0’ in register Config12 at address

44. If a buck topology is selected, the buck inductor is configured using bits BUCK[3:0] in register Config10 at address 42. Set bits BUCK[3:0] to ‘0’ or ‘1’ for a normal buck topology. Tapped buck topology is enabled by setting bit S2CONFIG to ‘0’ in register Config12 at address 44 and by setting bits BUCK[3:0] in register Config10 at address 42 to the required value N, where N equals the turns ratio between primary and secondary windings.

12 AN374REV2

AN374

Figure 5. Selecting the Second Stage Mode Using the CS1630/ 31 System Design Utility

MODE1

CH1VCH2

+I

CH2

=

[Eq. 7]

Figure 5 illustrates the process used to select the second stage flyback mode using the CS1630 / 31 system design utility.

The CS1630/31 has a minimum required T1 time that is dependent on leading edge blanking time T Blanking time T

is programmable from 150ns to 800 ns and is used to effectively disable the peak current

LEB

comparator from turning off the gate drive too early due to spurious switching noise. In applied systems, a good rule of thumb is to target a minimum duty cycle of 10% or greater.

Step 3) General considerations for all topology

1. In the series configurations, the current in one string will always be higher than the current in the second string across the entire dim and temperature range. There are specific advantages to using the series configuration if the light engine specifications comply with the constraints. In Mode 1, the phase synchronizer MOSFET is ‘OFF’ and the output power P

delivered to the load in a series design is calculated using

MODE1

Equation 7:

AN374REV2 13

LEB

AN374

MODE2

CH1ICH2

–V

CH1

=

[Eq. 8]

MODE1

CH1ICH1

=

[Eq. 9]

MODE2

CH1ICH2

=

[Eq. 10]

In Mode 2 the phase synchronizer MOSFET is turned ‘ON’ and the output power P

MODE2

is:

2. In the parallel configuration, the voltages of the two strings should at least have a 20% difference between the two strings in order for the CS1630/31 to distinguish between the two strings. The most common application of a parallel string is to achieve low CCT at low dim (<2000 K) and maintain ~ 2700 K at full brightness using a combination of mint/blue/white on one channel and amber/red on the other channel. Depending on the configuration, the current I in the white channel at low dim, whereas the current I brightness. When the phase synchronizer MOSFET is ‘OFF’ the output power P

in the red channel may be higher than the current I

Red

may be higher than current I

White

MODE1

delivered to the

Red

White

at full

load in a parallel design is as follows:

and when the phase synchronizer MOSFET is turned ‘ON’ the output power P

MODE2

is as follows:

3. If the power delivered to the load is similar in both configurations, then the magnetic utilization will be the maximum and the inductor size will be the smallest. For an LED configuration, the power can be computed in the four possible scenarios and it can be determined which design would be the optimum if the size of the inductor were the dominant factor in deciding between series and parallel configurations.

4. To quantify magnetic size, consider analyzing the peak current I inductance L

of a system designed to drive a single LED string. For example, examine a design with the

, turns ratio N, and primary

PK(FB)

single string CS1610/11/12/13 driver IC. For a comparable CS1630/31-based system converting the same level of output power, let peak current I be I

PK2(FB)

. It is assumed that I

PK1(FB)

for Mode 1 be I

PK(FB)

> I

PK2(FB)

PK1(FB)

and peak current I

PK(FB)

for Mode 2

. Let turns ratio N be such that the effective reflected voltage be the same so that their peak currents would be comparable between the two systems under comparison.

14 AN374REV2

AN374

Pow er Transf erred in a S ingle S tring Cont roller

Pow er Transf erred in a Tw o String C ontrol ler with

Unequal Balanc e between Two S trings

i(t)

T1 T2

T3T1 T2

CH1

CH2

Figure 6. Current Comparison through the Inductor

OUT dualPMODE1PMODE2

+= 2P

OUT glesin



[Eq. 11]

MODE1

OUT glesin

-----------------------------

1 +

------------------



[Eq. 12]





1 +F

sw1



------------------------------------



[Eq. 13]

Figure 6 illustrates the current comparison through the inductor. In this case, the buck is chosen to explain the point. The same logic can be extended to the flyback topology as well.

In a single-string design, power P

OUT(single)

ered in the second switching period and the total power is equal to two times power P channel design, the total power P

OUT(dual)

delivered in the first switching period TT equals power deliv-

OUT(single)

. In a two-

is approximately equal to:

The above equation yields Equation 12:

where

 = P

MODE2/PMODE1

The size of the inductor depends on the amount of energy stored at a first-order approximation. The energy stored in the inductor of a single string system is equal to P in the inductor of a two string system is equal to P by:

From Equation 13, it can be seen that the inductor for a CS1630/31-based equivalent system will be

 Fsw. The maximum energy stored

. The size factor  of the inductor is defined

MODE1

OUT(single)

F

sw1



times larger.

AN374REV2 15

AN374

SW2

SW1

BST

NV

MODE1

+

BST

NV

MODE2

+

----------------------------------------------------------

MODE1VMODE2



MODE2VMODE1



------------------------------------------

=

[Eq. 14]

The following are disclaimers for this analysis:

a. This analysis works better for a non-isolated design. For isolated flyback designs, the size may be deter-

mined by safety requirements for either triple insulated wire or margin tape spacing. As a result, increased peak current may or may not translate into the size of the inductor easily.

b. The explanation above shows a qualitative trend in understanding the tradeoffs of having power deliv-

ered to two strings unevenly. It is understood that it is not possible to determine a true analogy since the reflected voltage and frequency vary over the two modes of operation in the CS1630/31. This comparison should only be viewed from a qualitative understanding of the system.

5. When considering efficiency, if the power delivered to the load in the two modes is significantly different, then the frequency of operation of switching frequency F the converter operates in CRM at full brightness.

The converter will switch at a very high frequency while converting very little power. The power converter’s efficiency is low, however the power is transferred at two different frequencies and some amount of spreading effects of the spectral energy can be observed. This can potentially be favorable from an EMI perspective.

Step 4) Analyze system tradeoffs for efficiency when using a flyback topology

In a parallel configuration, the turns ratio will be higher for a given reflected voltage, so the leakage inductance will also be higher. As a result, the peak voltage on the drain of the flyback MOSFET must be taken into account. Since the effective output voltage is smaller, the synchronizer diode drop amounts to a much bigger efficiency penalty as opposed to the parallel design case.

In a series design, the leakage inductance will be smaller for the same reflected voltage. The maximum voltage on the rectifier diode is given by V

/ N + V

. The diode must be a higher voltage rated diode, which

OVP

increases cost and may impact efficiency, depending on the reflected voltage.

In a flyback topology, it is difficult to predict the diode losses prior to the development of a system, since diode losses vary significantly with junction temperature. This step is meant to serve as an added consideration when choosing a certain load configuration.

sw1

and F

will be significantly different because

sw2

Step 5) Analyze system tradeoffs for efficiency when using a buck topology

The minimum voltage rating for the power diode is greater than 250V for the CS1630 and greater than 450V for the CS1631. In order to guarantee optimum output current regulation in a buck converter, it is desirable to have a duty cycle of at least 10%, and any duty cycle above 15% is considered optimal for the buck converter.

In a parallel configuration, the effective output voltage is lower, and the diode losses higher. The duty cycle is smaller, since the effective output voltage is smaller. In a series configuration, the diode losses are lower, since the diode drop is still set by the boost voltage V

and not the output voltage.

BST

Constraint 7: Note on Frequency and EMI

The frequency of the double pulse is dependent on the ratio of the power delivered in each mode. Both strings having different power levels provides natural spreading of the spectrum for EMI. However, it decreases the magnetic utilization of the core.

The following are the hard constraints on the frequency of operation:

1. Maximum switching frequency cannot exceed 200kHz.

2. Minimum switching frequency cannot be lower than the value programmed in the OTP register TTMAX at Address 38.

16 AN374REV2

AN374

R22

R21

R23

CS1630 /31

FBAUX

GND

FBSEN SE

BST

C16

C15

D11

R27

Channel 1 LED (Wh ite )

Channel 2 LED (Red)

GND

R25

R26

SYNC

GND

Figure 7. Non-isolated Synchronizer Buck Series Output Model

4.3 Synchronizer Design Considerations

The synchronizer circuit drives an external MOSFET, which may be ground referenced depending on the topology.

Constraint 1: Non-isolated synchronizer circuit considerations

In the case of a non-isolated flyback, the secondary ground of the flyback can be referenced to the primary ground thereby having the least expensive synchronous circuit solution consisting of the synchronous MOSFET and the diode. The MOSFET is directly driven from the SYNC pin of the CS1630/31.

In the case of the buck topology, the outputs are referenced to link voltage and the following two options are possible:

1. Level shifting the SYNC pin output to a higher voltage using an external BJT and a P-channel MOSFET (see Figure 7).

2. Driving the synchronizer externally using flip-flop from the switching node or another winding off of the buck converter.

4.4 Constraints on the Boost Stage

Constraint 1: Maximum power should be power at full AC sine wave

It is recommended to design the dimming curve such that the maximum output power is delivered to the load at 100% brightness when the driver is behind the full AC sine wave.

Constraint 2: Power should be monotonically increasing

The output power in the driver should be close to monotonically increasing. If the available power from the dimmer cut waveform at any conduction angle is smaller than the requested output power, then the link voltage may not be in regulation at that conduction angle. In such a case, the peak currents on the boost inductor need to be increased, and all design calculations mentioned in application note AN368 are no longer valid for that particular design. Increasing peak currents may compromise EMI and efficiency.

AN374REV2 17

5 Design Flowchart

Yes

Pick a light engine

candidate across dim

Compute peak currents for a given transformer

using AN368

Start

Polynomial

approximation

possible?

Non-zero

currents across

dim?

Peak

current 4 times

smaller?

Proceed to choosing

string configuration

No abrupt change in

current

ratio

Nonidentical string configurations

across

dim?

Yes

No fundamental

limitations to using

light engine?

Figure 8. Eliminating Incompatible Light Engines

Figure 8 shows the flowchart for eliminating incompatible light engines.

AN374

18 AN374REV2

Figure 9 shows a way for selecting the correct second stage string configuration.

Yes

Provide completed light

engine design across

dim and temperature

Not compatible

No hard limitations to

using either

string configuration

Start

Parallel string

configuration

Series string configuration

Currents

cross at

any dim?

Voltage

Same?

Voltage

same?

Yes

Flyback topology. Series

and parallel string

configurations are sufficient

Buck?

Proceed to selecting

the topology

The series string configuration has

higher efficiency

Figure 9. Selecting the Correct Second Stage String Configuration

AN374

AN374REV2 19

Figure 10 shows a method for selecting the correct power train topology.

Yes

Start

Flyback

Buck

Isolation

required?

Required

MODEx

/ V

BST

< 10%

Yes

Figure 10. Selecting the Correct Power Train Topology

AN374

20 AN374REV2

AN374

1600

650

2700

Brightness (lm)

CCT (K)

Figure 11. Typical LED Light Engine Specifications

0.02

650

Brightness (lm)

1.0

0.02

2700

1600

CCT (K)

Dim

1.0

Figure 12. Light Engine Specifications Translated to an Imaginary Dim Axis

6 Data Improvements for the Curve Fitting Process

The curve-fitting process consists of translating a typical LED system specification into polynomial gain equations that the curve fitter can then calculate the required current for a given dim and NTC reading. The gains can be used in conjunction with the design of the power stage to complete a system design for a CS1630/31-based LED driver.

6.1 Typical Light Engine System Specifications

A typical light engine system specification is concisely captured in Figure 11. It is essentially a plot of the target lumen output at a given target color correlated temperature (CCT). Points 1, 2, and 3 are some data points of interest. These points have been used to define system behavior. The graph shows that the behavior is expected to be constant across the entire operating temperature of the driver. If the behavior over temperature were different, it can be represented as separate parameterized curves using the same axis.

6.2 Translation into Input Specifications for Calculating Color Gains

The light engine specifications above need to be translated into typical driver specifications that can be used by the second stage of the CS1630/31 based driver.

Step 1) Introduction of an independent dim axis

The curve fitter polynomial equation is an equation across dim and NTC temperature. The independent dim axis can be introduced as shown in Figure 12. At this point, the specifications can be redefined as needed to find a trajectory the light engine needs to traverse in terms of CCT and brightness when the input dim level to the system changes.

AN374REV2 21

Step 2) Translate into LED current profile at room temperature

m1, I

Red

, m2, I

White

Red

/ I

White

Ratio of the 2

Channel’s Lumens

Sum of Each

Channel’s Lumens

Approximated

Using Current

Approximated Using Current

0.02

650

1.0

0.02

2700

1600

Dim Dim Dim

1.0

0.02 1.0

0.02

Ratio

1.0

0.02 1.0

0.02

Ratio

1.0

, P

Red

+ K2, P

White

K1, P

Red

, P

White

Brightness (lm)

Relative Brightness (lm)

CCT (K)

Figure 13. Translate Data and CCT Lumen Dependence on Output Current

0.02

Red

1.0

amb

= 5°C

amb

= 25°C

amb

= 40°C

amb

= 55°C

Figure 14. Red Current Across Dim at Various Ambient Temperatures

The translation is as shown in Figure 13.

AN374

Variables K

and K2 represent the luminous efficacy of the LED, and variables m1 and m2 are arbitrary

constants. The plot in Figure 13 illustrates the relationship between CCT and output currents, lumens, and output power.

The data points 1, 2, and 3 are recreated under thermal equilibrium with the target LED light engine at room temperature to obtain the LED current profiles for each channel across the entire range of the target lumen output. The NTC temperature recorded on the final measurement is used on the driver. At the end of this procedure, the following are obtained: the current versus dim, forward voltage versus dim, NTC resistance versus dim, current ratio versus dim, and the LED light engine power profile data at room temperature.

Step 3) Validate data across various ambient temperatures

Finally, the same step needs to be completed across different ambient temperatures. All the measurements should be performed when the system has reached thermal equilibrium. Once the data is taken, the sample data is shown in Figures 14 and 15.

(A) I

Dim

22 AN374REV2

Note that any variation of the white string should be compensated for by changing the currents in the other

0.02 1.0 Dim

amb

= 5°C

amb

= 25°C

amb

= 40°C

amb

= 55°C

NTC Resistance (:

Figure 15. NTC Codes Across Dim at Various Ambient Temperatures

string so that the correct lumen and CCT are maintained across the entire temperature range of interest. At this point, both the currents of the white LED across dim and the data represented in Figures 14 and 15 should be available for the design.

6.3 Methods to Collect Required Data

The following sections describe the methods to collect the necessary data required for performing curve fitting.

6.3.1 Experimental Measurement in Lab

Experimental measurements are the most robust way to collect data. However, it is time consuming and requires expensive test equipment. It also requires the final light engine that is to be used for the design. The thermal resistance between the LED junction and the NTC should be known. In other words, the NTC should be placed at its final location to best mimic the thermal resistance. The light engine should be placed in an integrating sphere and characterized. For example, specifications with regards to CCT, Cx, Cy, lumens, and CRI can be measured.

Each output string is connected to an independent DC source that can source the required voltage and currents. The currents are experimentally changed until the desired points 1, 2, and 3 are obtained on Figure 11. At each of these points, measure the NTC resistance when it stabilizes to a constant value. If the NTC is not placed, then a thermocouple can be placed close to where the NTC will be placed and the temperature can be recorded.

The process should be repeated at various ambient temperatures. The entire data is collected across the various ambient temperatures to cover the thermal operating point space of the LEDs, which it is forced to traverse during transients (even though the actual ambient temperature is not what was specified in the spreadsheet). For example, it can alleviate the color shift problem at cold startup, as discussed in sectionTranslation into Input Specifications for Calculating Color Gains on page 21.

Another advantage of this approach is that it naturally compensates for any variation in the behavior of the other string across the various ambient temperatures by adjusting the two strings accordingly. Performing this experiment of accurately measuring the data across various ambient temperatures using an integrating sphere is not easy.

AN374

6.3.2 Simulation

Some of the designs can be done on the spreadsheet if at the least the following are available:

• Theoretical models for LED behavior across junction temperature

• Thermal model of the junction to NTC temperature thermal resistance

• NTC model

• Good optics and reflector model with respect to diffuser and LED beam angles

AN374REV2 23

AN374

T

amb

TJT

25 C

–=

[Eq. 15]

T

NTC

TJT

NTC

–=

[Eq. 16]

TJT

amb 0CTamb

[Eq. 17]

NTC

amb 0CTNTC

[Eq. 18]

RT0e



-------

–







=

[Eq. 19]

Red 0CIRed 25C



-------

=

[Eq. 20]

Red 40CIRed 25C



-------

=

[Eq. 21]

6.3.3 Experiment and Approximation

In many cases, collecting data at room ambient is not difficult. However, getting data across various ambient temperatures, particularly at low temperatures, is not easy. In an event where getting data is not feasible at ambient temperatures outside of room temperature, existing 25 based on data sheet numbers of the LED and measured temperature rises. If the requested operating temperature is very large (

T

> 50C), then the errors due to this simple linear approximation may be too

amb

high. Remember that the method is a simple approximation and subject to errors that may vary depending on the system under investigation.

The following method is suggested to extend data to other ambient temperatures. In this case, two ambient temperatures of 0

C and 40C are chosen. These are some of the assumptions that are made. The

temperature rise from ambient is assumed to be constant irrespective of the actual ambient temperature level.

Step 1) Collect red current I

C ambient temperature T

, white current I

Red

amb(25C)

, and NTC data and junction temperature TJ of the LED at

White

from 0% dim to 100% dim.

Step 2) Calculate the following:

C data can be extended to other temperatures

Step 3) Calculate the various NTC temperature T

and junction temperature TJ of the LED at various

NTC

ambient temperatures across the entire dimming range as follows:

Step 4) Calculate the NTC resistance using the equation of the system at 25

C ambient and the expression

At this point, a complete sample space has been recreated using the temperature rise. The remaining steps are focused on recreating the red currents I

across temperature.

Red

Step 5) Refer to the LED data sheet for the Relative Luminous Flux Curve and note the relative luminous flux

at ambient temperature T fluxes at 25

C, 0C, and 40C ambient, then the red currents can be calculated using Equations 20

= 25C, 0C, and 40C. If 1, 2, and 

amb

are the relative luminous

and 21:

Thus by performing this at every dim level, the entire dimming data required can be constructed.

24 AN374REV2

6.4 Improving Data for Feeding the Curve Fitter

NTC Beta 4334 Vref 1.25 ENTER DIM VALUES (BwDim, RedDim), CURRENTS (I_bw, I_red)AND NTC TEMPERATURE (Tntc) in deg. C NTC T0 °C 25 Nbit 8 NTC code IS COMPUTED FROM Tntc AND NTC PARAMETERS NTC Rs at T0 100000 Ifs 8.00E-05 Series R 14000 іNTC PARAMETERS

љљ

BwDim I_bw (A) ї enter 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

0.1 0.068

0.2 0.131 Tamb(room)

computedќ

enter ќ

0.3 0.190 ї enter 5.00 0.042 0.078 0.110 0.138 0.161 0.180 0.196 0.207 0.216 0.221 0.09 0.16 0.23 0.28 0.33 0.37 0.40 0.42 0.44 0.45

0.4 0.246 15.00 0.044 0.083 0.116 0.145 0.170 0.190 0.206 0.218 0.227 0.233 0.09 0.17 0.24 0.30 0.35 0.39 0.42 0.45 0.46 0.48

0.5 0.297 25.00 0.046 0.087 0.122 0.153 0.179 0.200 0.217 0.230 0.239 0.245 0.09 0

.18 0.25 0.31 0.36 0.41 0.44 0.47 0.49 0.50

0.6 0.344 45.00 0.049 0.091 0.128 0.160 0.187 0.210 0.228 0.241 0.251 0.257 0.10 0.19 0.26 0.33 0.38 0.43 0.46 0.49 0.51 0.53

0.7 0.386 65.00 0.051 0.096 0.135 0.168 0.197 0.220 0.239 0.253 0.264 0.270 0.10 0.20 0.28 0.34 0.40 0.45 0.49 0.52 0.54 0.55

0.8 0.425 85.00 0.053 0.101 0.142 0.177 0.207 0.231 0.251 0.266 0.277 0.284 0.11 0.21 0.29 0.36 0.42 0.47 0.51 0.54 0.57 0.58

10.490 Tamb(room) enter ќ

5.00 5.20 7.40 9.60 11.80 14.00 16.20 18.40 20.60 22.80 25.00

15.00 15.20 17.40 19.60 21.80 24.00 26.20 28.40 30.60 32.80 35.00

25.00 25.20 27.40 29.60 31.80 34.00 36.20 38.40 40.60 42.80 45.00

45.00 45.20 47.40 49.60 51.80 54.00 56.20 58.40 60.60 62.80 65.00

65.00 65.20 67.40 69.60 71.80 74.00 76.20 78.40 80.60 82.80 85.00

computedќ

13.5 15.2 17.0 19.0 21.2 23.6 26.1 28.9 31.9 35.1

22.5 25.0 27.6 30.5 33.6 36.9 40.5 44.2 48.2 52.4

35.4 38.8 42.5 46.3 50.4 54.7 59.2 64.0 68.9 73.9

74.4 79.7 85.1 90.6 96.2 102.0 107.7 113.6 119.5 125.3

Hight Temp Code Limit 185 125.9 131.7 137.5 143.3 149.0 154.6 160.0 165.4 170.7 175.8 Low Temp Code Limit 12 176.2 181.1 185.9 190.6 195.0 199.3 203.4 207.4 211.2 214.8

RedDim

RaƟo (String2/String1)

Tntc[C]

NTC code

I_red

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0 0.2 0.4 0.6 0.8 1 1.2

RaƟo

Dim

Current RaƟos Vs Dim

Tamb = 5C

Tamb=15C

Tamb = 25C

Tamb=45C

Tamb=65C

Tamb=85C

0.000

0.100

0.200

0.300

0.400

0.500

0.600

0 0.2 0.4 0.6 0.8 1 1.2

Output Current (mA)

Dim

Currents Vs Dim

I_BW

I_RED 5C

I_RED 15C

I_RED 25C

I_RED 45C

I_RED 65C

I_RED 85C

Step 1:

Enter NTC

Configuration

Step 4:

Enter Current Ratio

vs Dim and Temp

Step 5:

Enter Temperature or

NTC Resistance

Step 6:

Compute NTC Code

Step 3:

Enter Red Current vs

Dim and Temperature

Step 2:

Enter Blue

Current vs Dim

Figure 16. Example Spreadsheet Used to Collect Curve Fitting Data

NTC Beta 4334 Vre f 1.25 NTC T0 °C 25 Nbit 8 NTC Rs at T0 100000 Ifs 8.00E-05 Se ries R 14000

Does Not Change

Figure 17. NTC Specification from Example Spreadsheet

NTC

CODE

4M

Series

RT0e



-------

–













--------------------------------------------------------------------------

[Eq. 22]

This section demonstrates the use of a spreadsheet to manipulate the data collected in the previous section to get a more accurate curve fit. The data is presented in a format compatible with the curve fitter software.

6.4.1 Understanding the Data and Data Format

The final data can be collected in a spreadsheet. An example of the spreadsheet is shown in Figure 16. Steps are provided to the left to describe the process.

enter enter

0.9 0.460

85.00 85.20 87.40 89.60 91.80 94.00 96.20 98.40 100.60 102.80 105.00

AN374

The NTC configuration section is shown in Figure 17. Values for Vref, Nbit and Ifs are internal parameters of the IC that do not change. The beta and initial resistance at the initial temperature can be changed.

The NTC code can be generated using Equation 22. If temperature is specified in the system, then:

where

Temperature T1 and T0 are in degrees Kelvin

AN374REV2 25

AN374

NTC

4M

NTCRSeries

--------------------------------------

[Eq. 23]

enter enter

љљ

0.1 0.068

0.3 0.190

0.4 0.246

0.6 0.344

0.7 0.386

0.8 0.425

0.9 0.460

Enter dim values

in ascending order

At 50% dim,

White

= 0.297 A

Figure 18. Example of Formatted Data for the White/Blue Current Across Dim

ї enter 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

5.00 0.042 0.078 0.110 0.138 0.161 0.180 0.196 0.207 0.216 0.221

15.00 0.044 0.083 0.116 0.145 0.170 0.190 0.206 0.218 0.227 0.233

25.00 0.046 0.087 0.122 0.153 0.179 0.200 0.217 0.230 0.239 0.245

45.00 0.049 0.091 0.128 0.160 0.187 0.210 0.228 0.241 0.251 0.257

85.00 0.053 0.101 0.142 0.177 0.207 0.231 0.251 0.266 0.277 0.284

5.00 5.20 7.40 9.60 11.80 14.00 16.20 18.40 20.60 22.80 25.00

15.00 15.20 17.40 19.60 21.80 24.00 26.20 28.40 30.60 32.80 35.00

65.00 65.20 67.40 69.60 71.80 74.00 76.20 78.40 80.60 82.80 85.00

85.00 85.20 87.40 89.60 91.80 94.00 96.20 98.40 100.60 102.80 105.00

computedќ

13.5 15.2 17.0 19.0 21.2 23.6 26.1 28.9 31.9 35.1

22.5 25.0 27.6 30.5 33.6 36.9 40.5 44.2 48.2 5 2.4

125.9 131.7 137.5 143.3 149.0 154.6 160.0 165.4 170.7 175.8

176.2 181.1 185.9 190.6 195.0 199.3 203.4 207.4 211.2 214.8

RedDim

Tntc[C]

NTC code

I_red

At 30% dim,

Red

= 0.122 A and

NTC

= 29.60°C

Enter dim values in ascending order

Enter temperature values in ascending order

Figure 19. Example of Formatted Data for Red Current and NTC Resistance/Code Across Dim

If resistance is measured, then

The data below shows the white/blue currents across the dim. The format in Figure 18 is useful because it is required when the data is imported into the curve fitter.

BwDim I_bw (A)

0.2 0.131

0.5 0.297

10.490

The blue/white current at 100% will be considered the nominal current at full sine wave. The reference currents are set based on this value.

The red currents and the corresponding temperature data are shown below.

Tamb(room)

Tamb(room) enter ќ

The target currents will be the currents at 100% dim and 25 Figure 19 shows the computed NTC resistance according to Equation 22. The NTC code is the internal computed value used by the CS1630/ 31.

26 AN374REV2

computedќ

65.00 0.051 0.096 0.135 0.168 0.197 0.220 0.239 0.253 0.264 0.270

25.00 25.20 27.40 29.60 31.80 34.00 36.20 38.40 40.60 42.80 45.00

45.00 45.20 47.40 49.60 51.80 54.00 56.20 58.40 60.60 62.80 65.00

35.4 38.8 42.5 46.3 50.4 54.7 59.2 64.0 68.9 73.9

74.4 79.7 85.1 90.6 96.2 102.0 107.7 113.6 119.5 125.3

C ambient temperature. The NTC code data in

AN374

6.4.2 Manipulating Data for Better Curve Fit

As mentioned previously, the curve fitter looks for a polynomial fit with a maximum of fourth order. There are a few quick techniques that can be performed to make a better fit. This section shows some of the common techniques used to improve the quality of data and therefore the quality of the fit. In this section, it is assumed that the data, if attempted to be fitted using a curve fitter, provides a large error, so improvements are necessary.

Step 1) Use a weighted fit

A weighting inequality can be enforced such that some regions can be given higher weights thereby forcing the curve fitter to reduce the error while fitting in those regions. This is particularly useful when the observed error is skewed more towards the region of interest as opposed to regions where the light bulb may not operate regularly.

Step 2) Identify the region of interest

Some data points on the curve fitter have a larger significance to the light bulb as opposed to others. These regions should be identified immediately and the curve fitting should be done in such a way that the errors in these regions are minimized as a first priority.

Consider an example where an incandescent dimming profile is requested with 2% lumen output is required at the lowest dim settings. Since all the major standards and specifications such as UL, LM79 are based on full sine wave operation, that region is important and high accuracy in both lumen output and CCT is desirable. From experience, it has been observed that at a low conduction angle, the light output is so small that the eye is not yet saturated and if the two strings are not mixed correctly along the Planckian locus, the individual components will be visible. It is important for the curve-fitting process to have minimal error in this region. In most of the other regions, some amount of allowance could possibly be made depending on the application.

Step 3) Add trendline in a spreadsheet to see outliers in data

The trendline function in a spreadsheet or any other function in a compatible software helps identify any outliers in the data. The following guidelines describe what to consider while using this trendline functions:

1. Choose a polynomial function of order 2, 3, or 4 for the current versus dim on the red and white currents. A linear fit is acceptable if it produces a lower mean squared error (MSE).

2. Set intercept to origin

3. Display the equation on the chart

4. Display MSE on the chart (R2)

5. The trendline provides physical intuition in understanding where the potential problems may lie in trying to fit the data into a polynomial curve.

Step 4) Dim axis remapping

Many data that may even look fundamentally incompatible with the curve fitter can be made to fit well by remapping the dim axis. Mathematically a straight line with an equation Y = Mx + B can be made to be a line Y = Mx by adding -B/M to every point on the x axis. Thus by shifting the whole axis to the right by B/M, the curve can be made to pass through the origin.

Figure 20 shows another example of the above statement. In this case, by merely changing the dim point on the axis from 0.4 to 0.5, the data that was an outlier is now part of the good data set. From a physics standpoint, this still does not have any impact to the system specifications since the red current and the white current combination remain the same.

AN374REV2 27

AN374

0.05

0.125

0.25

0.375

0.5

0.03

0.075

0.15

0.225

0.3

0.1

0.2

0.3

0.4

0.5

0.6

0 0.2 0.4 0.6 0.8 1 1.2

LED Current (A)

Red LED Current

White LED Current

Figure 20. White and Red LED Current vs Dim

Notes: 1. Based on relative brightness.

Dim

White

(mA) I

Red

(mA)

0.10 0.050 0.030

0.25 0.125 0.075

0.40 0.250 0.150

0.75 0.375 0.225

1.00 0.500 0.300

0.05

0.125

0.25

0.375

0.5

0.03

0.075

0.15

0.225

0.3

0.1

0.2

0.3

0.4

0.5

0.6

0 0.2 0.4 0.6 0.8 1 1.2

Current (A)

White LED Current

Red LED Current

Figure 21. Corrected White and Red LED Current vs Dim

Notes: 1. Based on relative brightness.

Dim

White

(mA) I

Red

(mA)

0.10 0.050 0.030

0.25 0.125 0.075

0.50 0.250 0.150

0.75 0.375 0.225

1.00 0.500 0.300

When re-mapping the dim axis, remember the following points:

• The dim axis needs to be remapped to the white current and the red current axis simultaneously. Otherwise, the data will be altered.

• In the plot in Figure 20 it can be seen that the data point is to the left of the trendline for the red current and the white current. Moving the dim axis helps mitigate the problem under these circumstances.

Dim (%)

• If the red point and the white point were tending in the opposite way, remapping the dim axis to the center could split the error for the same polynomial. One can change the shape of the trendline used for either one of the curves by changing its order and then again attempting to remap the dim axis to see if a better fit can be achieved.

28 AN374REV2

Step 5) Eliminating points outside the region of interest

If remapping the dim axis does not provide sufficient results, then eliminate the points that are outside the region of interest and see if the trendline can provide a more acceptable fit. The order of the polynomial that is used can be different for the red and the white and one can by trial and error see if moving to a higher order polynomial can improve the fit. An example below is shown to better understand the point.

Figure 22 shows raw data for a light engine system and the best fit trendline. It can be seen that the trendline is off at a number of points in the low end. As discussed earlier the curve should be accurate at the low dim range and yet should be accurate at full brightness. The currents should be always > 0 as discussed in the

Dim (%)

AN374

y = 887.97x4- 1549x3+ 733.71x2+ 301.04x

R² = 0.9969

100

150

200

250

300

350

400

0.00 0.20 0.40 0.60 0.80 1.00 1.20

Output Current (mA)

Dim

Figure 22. Best Fit Trendline for a Light Engine

y = 1974x4- 3827.1x3+ 2171.9x2+ 56.207x

R² = 1

100

150

200

250

300

350

400

0 0.2 0.4 0.6 0.8 1 1.2

Output Current (mA)

Dim

Figure 23. Reconstructing the Dim Map

previous sections. If some of the points in the middle region of the dimming curve are removed, then the fit can be significantly improved in the other regions of interests.

By removing the points of interest, the dim map is reconstructed as illustrated in Figure 23. It can be seen that even without remapping the dim axis, the points that were considered critical could be fitted. The error between the requested data and the actual suggested fit is shifted to the region between 25% to 40% dim at which point the bulb is bright enough to not be able to spot the anomalies of being away from the Planckian locus.

Step 6) Weighted fit to increase accuracy in the region of interest.

Once the data has been collected, the weighting function and the corresponding weights can be chosen to increase the weights in a specific region. This forces the curve fitter to minimize the error in this region improving the fit in the region of interest.

AN374REV2 29

AN374

1 0.085 0.005928

3 0.165 0.01882

5 0.26 0.042232

7 0.35 0.071397

9 0.5 0.132022

11 0.7 0.228331

13 0.9 0.331064

Figure 24. Sample CSV File for White Current vs Dim

ABCDE F GHI J K L MNO

2 0 0.011601 0.017686 0.028347 0.04255 0.054017 0.066287 0.082542 0.099565 0.134836 0.16994 0.20272 0.231014 0.252666 0.265514

4 40 0.01623 0.024743 0.039658 0.059529 0.075571 0.092737 0.115478 0.139294 0.188639 0.237751 0.28361 0.323195 0.353486 0.371462

Dim values in ascending order

Ambient temperature values in ascending order

Figure 25. Sample CSV File for Red Current vs Dim

1 0.085 0.117 0.165 0.22 0.26 0.3 0.35 0.4 0.5 0.6 0.7 0.8 0.9 1

3 25 109000 108000 105000 103000 99000 93000 80000 75000 68000 59000 54500 48700 43300 38700

NTC data is calculated using the exact dim and ambient temperature values that are used to derive the red current.

Figure 26. Sample CSV File for NTC Resistance vs Dim

7 Performing Curve Fitting

7.1 Using the CS1630/31 Application Software to Perform Curve Fit

The CS1630/31 application software and graphical user interface (GUI) are designed to perform all necessary operations to program the one-time programmable (OTP) registers. In this section, an overview of the utility is provided to help understand how to perform the curve fit. The curve fitting utility is incorporated in the application software and can be downloaded from the Cirrus web site. The instructions for its installation can be found in application note AN369: CS1630/31 Device Programmer User Guide. The curve fitter utility is found under the System Design tab. In order to perform the curve fitting process, it is not necessary to be connected to an IC, and it can be done offline.

The curve fitter utility contains the following three tabs:

• White Gain: Curve fitting for white gain terms. In this section, enter the white current versus dim data. The data can be manually entered in the fields. Right click in the data section to load a comma-separated values (CSV) file. Format the CSV so that the dim values are in ascending order and the currents are entered in Amperes (see Figure 24).

Dim values in ascending order

2 0.117 0.01028

4 0.22 0.031378

6 0.3 0.054425

8 0.4 0.090116

10 0.6 0.178605

12 0.8 0.279663

14 1 0.380998

• Red Gain: Surface curve fitting tool for the red gain. Enter the red current versus dim across ambient temperatures here. The data can be entered manually. Right click in the data section to load a CSV file. Format the CSV so that the ambient temperature values are increasing when descending across the row and the currents are entered in amperes (see Figure 25).

1 0.085 0.117 0.165 0.22 0.26 0.3 0.35 0.4 0.5 0.6 0.7 0.8 0.9 1

3 25 0.014113 0.021515 0.034485 0.051764 0.065714 0.080641 0.100416 0.121125 0.164034 0.20674 0.246617 0.281039 0.307379 0.32301

• NTC Data: Thermal equilibrium NTC data. The NTC data can be entered as a temperature value or as an NTC resistance value. The NTC data and the red gain data should have the exact same dim and temperature values. They should be a one-to-one correspondence. The data can be entered manually. Right click in the data section to load a CSV file. Format the CSV file so that the NTC data is calculated using the exact dim and ambient temperature values that were used for the red current.

ABCDEFGH I J KLMNO

2 0 415756.2 411154.6 397406.6 388288.9 370170.5 343293.7 286373.6 264992.1 235572.1 198692.9 180684.3 157932.4 137249.8 120044.4

30 AN374REV2

4 40 54213.34 53756.93 52385.42 51469.14 49631.8 46863.29 40808.96 38457.99 35143.39 30837.63 28664 25839.6 23184.32 20900.52

Once the data has been entered, the curve fitter can be understood by performing the following steps:

Step 1:

Enter the white

LED currents

and dim.

Step 2:

Enter the

Weight Factor.

Step 4:

Adjust results.

Step 5:

Click the UPDATE

REGISTERS button.

Step 3:

Click the Calculate

Coefficients button.

Error graph give ratio of fit to the data.

Figure 27. SYSTEM DESIGN - White Gain Tab

Step 1) White Gain setting

The entry procedure that is required to preform a curve fit for the White Gain polynomial is detailed in Figure 27.

AN374

AN374REV2 31

1. Enter white LED currents (in Amperes) and dim. Right click to see options to load data from a CSV file.

2. Enter the Weight Factor. Values > 1 favor high dim levels in the fit and values < 1 favor low dim levels.

3. Click the Calculate Coefficients button.

4. Adjust results. Polynomial coefficients can be modified over and above the fit results.

5. Click the UPDATE REGISTERS button to update the Color System tab under REGISTERS with Q3, Q2, Q1, and Q0. Values are not written to the chip; only the tab is populate.

AN374

Step 1:

Enter the color LED

currents for different dims.

Select graph to view to check quality of fit.

Select 3D view perspective.

Click Expand to display the graph in a larger window

Step 2:

Select “Use Mask”.

Step 3:

Enter the Weight Factor.

Step 5:

Adjust Results.

Step 6:

Click the UPDATE

REGISTERS button.

Step 4:

Click the Calculate

Coefficients button.

Figure 28. SYSTEM DESIGN - Color Gain Tab

Step 2) Red Gain setting

The entry procedure that is required to preform a curve fit for the Color Gain polynomial is detailed in Figure 28.

1. Enter the color LED currents (in Amperes) for different dims. Repeat for different ambient temperatures. Right click to see options to load data from CSV file.

2. Select “Use Mask” to mask the interpolated NTC code regions not present in the NTC data in the surface to be fit. Set Mask Low Level to a value relative to 1 to use as weights for the interpolated NTC code regions on the surface.

3. Enter Weight Factor. Values > 1 favor high dim levels in the fit and values < 1 favors low dim levels.

4. Click the Calculate Coefficients button.

5. Adjust Results. Polynomial coefficients can be modified over and above the fit results.

6. Click the UPDATE REGISTERS button to update the Color System tab under REGISTERS with Q3, Q2, Q1, and Q0. Values are not written to the chip; only the tab is populated.

32 AN374REV2

AN374

Figure 29. SYSTEM DESIGN - NTC Data Tab

Step 3) NTC setting

In Figure 29, the value can be entered as an NTC parameter, which corresponds to temperature or in resistances themselves. In addition, the resistance in series to the NTC should be specified. Check application note AN368 for schematic and typical connection to the eOTP pin.

Once the gains are computed, note that the UPDATE REGISTERS button does write the registers to the IC. At this point, connect the I the bottom of the tab page, and click the RUN button. See application note AN369 for more information about the CS1630/31 application software.

C connections and click the Write All Registers button from the drop-down list in

7.2 Register Bits in OTP Map Affected by the Curve Fitter

The following bits need to be set based on the data collected:

• The second stage minimum dim value may be changed as a result of remapping the dim axis (S2DIM). Refer to application note AN368 for more information about how to recalculate it.

• Bits HI_SAT[2:0] and LOW_SAT[2:0]: By default, HI_SAT[2:0] is set at the eOTP threshold. If the eOTP threshold is much higher than the highest NTC temperature in the operating range, then the value can be clipped to the highest temperature with a little margin on top for accuracy considerations. The codes can be directly specified if the spreadsheet is constructed in the manner specified previously.

AN374REV2 33

• Registers CH1CUR and CH2CUR: Channel 1 Reference targets should be the current at 100% brightness

on the white string. Channel 2 reference targets should be the currents at 100% brightness at 25 temperature. The currents can be translated into their corresponding digital representation as shown in application note AN368.

7.3 Using Commercially Available Curve Fitting Software

There are other curve fitting tools that are commercially available, and they can be used for performing the curve fit. The criteria for using a curve fit utility are as follows:

1. There is a consistent (I

2. There is an I

3. All the remaining bits mentioned above will have to be manually entered.

There are many commercially available curve-fitting software. Some of the advantages of using advanced curve fitters are as follows:

• Ability to generate weighting methodology based on temperature codes. For example, the fitting requirements for the cold ambient temperature data can be relaxed greatly as that is the least likely operating space region for most designs.

• Advanced weighting error function

Some examples of curve fitting tools that match the criterion above are CurveExpert Professional and ZunZun.com Online Curve Fitting. Cirrus Logic does not sponsor or endorse these third party products, any use of which is done at your own risk. The third party products may include or reflect trademarks or logos of companies other than Cirrus Logic, Inc. All such trademarks or logos are the property of their respective owners, who are not affiliated in any way with Cirrus Logic.

White

, I

Red

, NTC) combination across the entire region of interest

Red

combination across the entire region of interest at room ambient temperature.

AN374

C ambient

7.4 Process of Converting a Given Gain Coefficients to CS1630/31 OTP Map

As mentioned previously, the color polynomial coefficients can have a maximum range of [-8, 8] and the actual gain coefficients themselves can have a range of [0, 4]. If the coefficients have an absolute value greater than 8, check for the following items as a process of debug.

• Currents should be in Amperes.

• NTC resistances should be in Ohms.

• Normalize the whole equation such that the polynomial coefficients are now less than 8.

Step 1) Get the white gain equation in the normalized form as mentioned in the previous bullet.

Step 2) Take two’s compliment of the coefficient Q0, Q1, Q2, Q3. Refer to the CS1630/31 data sheet to see

exactly the way the bits are arranged (for more information about this data sheet, see Further Reading on page 1).

Step 3) Take two’s compliment of the color gain coefficients. Refer to the CS1630/31 data sheet for conversion

into corresponding OTP map.

Step 4) At dim = 100%, 50%, calculate the blue/white gains and the red gains.

Step 5) Calculate the currents computed by the color gain and compute the error.

Step 6) On the fitting curve, look for the MSE at that point and see if the error calculated matches with the error

seen on the curve fitter output.

34 AN374REV2

8 Revision History

Revision Date Changes

REV1 NOV 2012 Initial release.

REV2 FEB 2013 Corrected typographical errors.

AN374

AN374REV2 35

Cirrus Logic AN374 User Manual

Specifications and Main Features

Frequently Asked Questions

User Manual

1 Overview of the CS1630

2 Introduction

2.1 Definition of Acronyms

2.2 Definition of Symbols

3 Introduction to the Color System

4 Light Engine Choices

4.1 Considerations for the Color System

4.2 Constraints Imposed by Second Stage

4.3 Synchronizer Design Considerations

4.4 Constraints on the Boost Stage

5 Design Flowchart

6 Data Improvements for the Curve Fitting Process

6.1 Typical Light Engine System Specifications

6.2 Translation into Input Specifications for Calculating Color Gains

6.3 Methods to Collect Required Data

6.3.1 Experimental Measurement in Lab

6.3.2 Simulation

6.3.3 Experiment and Approximation

6.4 Improving Data for Feeding the Curve Fitter

6.4.1 Understanding the Data and Data Format

6.4.2 Manipulating Data for Better Curve Fit

7 Performing Curve Fitting

7.1 Using the CS1630/31 Application Software to Perform Curve Fit

7.2 Register Bits in OTP Map Affected by the Curve Fitter

7.3 Using Commercially Available Curve Fitting Software

7.4 Process of Converting a Given Gain Coefficients to CS1630/31 OTP Map

8 Revision History