7.3.1.1 Control Bank Mapping

Control of the LM3630A device current sinks is not done directly, but through the programming of Control Banks. The current sinks are then assigned to the programmed Control Bank (see Figure 72). Both current sinks can be assigned to Control Bank A or LED1 can use Control Bank A while LED2 uses Control Bank B. Assigning LED1 to Control Bank A and LED2 to Control Bank B allows for better LED current matching. Assigning each current sink to different control banks allows for each current sink to be programmed with a different current or have the PWM input control a specific current sink.

Figure 72. Control Diagram

7.3.1.2 PWM Input Polaritiy

The PWM Input can be set for active high (default) or active low polarity. With active low polarity the LED current is a function of the negative duty cycle at PWM.

7.3.1.3 HWEN Input

HWEN is the global hardware enable to the LM3630A. HWEN must be pulled high to enable the device. HWEN is a high-impedance input so it cannot be left floating. When HWEN is pulled low the LM3630A is placed in shutdown and all the registers are reset to their default state.

7.3.1.4 SEL Input

SEL is the select pin for the serial bus device address. When this pin is connected to ground, the seven-bit device address is 36H. When this pin is tied to the VIN power rail, the device address is 38H.

7.3.1.5 INTN Output

The INTN pin is an open-drain active-low output signal which indicates detected faults. The signal asserts low when either OCP, OVP, or TSD is detected by the LED driver. The Interrupt Enable register must be set to connect these faults to the INTN pin.

7.3.1.6 Boost Converter

The high-voltage boost converter provides power for the two current sinks (ILED1 and ILED2). The boost circuit operates using a 10-μH to 22-μH inductor and a 1-μF output capacitor. The selectable 500-kHz or 1-MHz switching frequency allows for the use of small external components and provides for high boost converter efficiency. Both LED1 and LED2 feature an adaptive voltage regulation scheme where the feedback point (LED1 or LED2) is regulated to a minimum of 300 mV. When there are different voltage requirements in both high-voltage LED strings, because of different programmed voltages or string mismatch, the LM3630A regulates the feedback point of the highest voltage string to 300 mV and drop the excess voltage of the lower voltage string across the lower strings current sink.

7.3.1.7 Boost Switching Frequency Select

The LM3630A’s boost converter can have a 500-kHz or 1-MHz switching frequency. For a 500-kHz switching frequency the inductor value must be between 10 μH and 22 μH. For the 1-MHz switching frequency the inductor can be between 10 μH and 22 μH. Additionally, there is a Frequency Shift bit which offsets the frequency approximately 10%. For the 500 kHz setting, shift = 0. The boost frequency is shifted to 560 kHz when Shift = 1. For the 1-MHz setting, Shift = 0. The boost frequency is shifted to 1120 kHz when shift = 1.

7.3.1.8 Adaptive Headroom

Reference Figure 73 and Figure 74 for the following description.

The adaptive headroom circuit controls the boost output voltage to provide the minimal headroom voltage necessary for the current sinks to provide the specified ILED current. The headroom voltage is fed back to the Error Amplifier to dynamically adjust the Boost output voltage. The error amplifier's reference voltage is adjusted as the brightness level is changed, because the currents sinks require less headroom at lower I_LED currents than at higher I_LED currents. Note that the VHR Min block dynamically selects the LED string that requires the higher boost voltage to maintain the I_LED current; this string has the lower headroom voltage. In Figure 74 this is LED string 2. The headroom voltage on LED string 1 is higher, but this is due to LED string 2 have an overall higher forward voltage than LED string 1. LED strings that have closely matched forward voltages have closely matched headroom voltages and better overall efficiency.

In a single string LED configuration the Feedback enable must be enabled for only that string (LED1 or LED2). The adaptive headroom circuit is control by that single string. In a two string LED configuration the Feedback enable must be enabled for both strings (LED1 and LED2). The VHR Min block then dynamically selects the LED string to control the adaptive headroom circuit.

Figure 73. Adaptive Headroom Block Diagram

Figure 74. Typical Headroom Voltage Curve

7.3.1.9 Current Sinks

LED1 and LED2 control the current up to a 40-V LED string voltage. Each current sink has 5-bit full-scale current programmability and 8-bit brightness control. Either current sink has its current set through a dedicated brightness register and can additionally be controlled via the PWM input.

7.3.1.10 Current String Biasing

Each current string can be powered from the LM3630A device’s boost or from an external source. When powered from an external source the feedback input for either current sink can be disabled in the Configuration Register so it no longer controls the boost output voltage.

7.3.1.11 Full-Scale LED Current

The LM3630A device’s full-scale current is programmable with 32 different full-scale levels. The full-scale current is the LED current in the control bank when the brightness code is at max code (0xFF). The 5-bit full-scale current vs code is given by Equation 1:

With a maximum full-scale current of 28.5 mA.

7.3.1.12 Brightness Register

Each control bank has its own 8-bit brightness register. The brightness register code and the full-scale current setting determine the LED current depending on the programmed mapping mode.

7.3.1.13 Exponential Mapping

In exponential mapping mode the brightness code to backlight current transfer function is given by Equation 2:

Equation 2.

where

• I_{LED_FULLSCALE} is the full-scale LED current setting
• Code is the backlight code in the brightness register
• DPWM is the PWM input duty cycle

Figure 75 and Figure 76 show the approximate backlight code to LED current response using exponential mapping mode. Figure 75 shows the response with a linear Y axis, and Figure 76 shows the response with a logarithmic Y axis. In exponential mapping mode the current ramp (either up or down) appears to the human eye as a more uniform transition then the linear ramp. This is due to the logarithmic response of the eye.

Figure 75. Exponential Mapping Mode (Linear Scale)

Figure 76. Exponential Mapping Mode (Log Scale)

7.3.1.14 Linear Mapping

In linear mapping mode the brightness code to backlight current has a linear relationship and follows Equation 3:

Equation 3.

where

• I_{LED_FULLSCALE} is the full scale LED current setting
• Code is the backlight code in the brightness register
• DPWM is the PWM input duty cycle

Figure 77 shows the backlight code-to-LED current response using linear-mapping mode. The Configuration Register must be set to enable linear mapping.

Figure 77. Linear Mapping Mode

7.3.2.1 Open LED String (LED1 And LED2)

An open LED string is detected when the voltage at the input to either LED1 or LED2 has fallen below 200 mV, and the boost output voltage has hit the OVP threshold. This test assumes that the LED string that is being detected for an open is being powered from the boost output (Feedback Enabled). For an LED string not connected to the boost output, and connected to another voltage source, the boost output would not trigger the OVP flag. In this case an open LED string would not be detected.

7.3.2.2 Shorted LED String

The LM3630A features an LED short fault flag indicating if either of the LED strings have experienced a short. There are two methods that can trigger a short in the LED strings:

1. An LED current sink with feedback enabled, and the difference between OVP input and the LED current sink input voltage goes below 1 V.
2. An LED current sink is configured with feedback disabled (not powered from the boost output) and the difference between V_IN and the LED current sink input voltage goes below 1 V.

7.3.2.3 Overvoltage Protection (Manufacturing Fault Detection and Shutdown)

The LM3630A provides an overvoltage Protection (OVP) mechanism specifically for manufacturing test where a display may not be connected to the device. The OVP threshold on the LM3630A has 4 different programmable options (16 V, 24 V, 32 V, and 40 V). The manufacturing protection is enabled in the Fault Status register bit 0. When enabled, this feature causes the boost converter to shutdown anytime the selected OVP threshold is exceeded. The OVP_fault bit in the Fault Status register is set to one. The boost converter does not resume operation until the LM3630A is reset with either a write to the Software Reset bit in the Software Reset register or a cycling of the HWEN pin. The reset clears the fault.

7.3.3.1 Overvoltage Protection (Inductive Boost Operation)

The overvoltage protection threshold (OVP) on the LM3630A has 4 different programmable options (16 V, 24 V, 32 V, and 40 V). OVP protects the device and associated circuitry from high voltages in the event the feedback enabled LED string becomes open. During normal operation, the LM3630A device’s inductive boost converter boosts the output up so as to maintain at least 300 mV at the active current sink inputs. When a high-voltage LED string becomes open the feedback mechanism is broken, and the boost converterinadvertently over boosts the output. When the output voltage reaches the OVP threshold the boost converter stops switching, thus allowing the output node to discharge. When the output discharges to V_OVP – 1 V the boost converter begins switching again. The OVP sense is at the OVP pin, so this pin must be connected directly to the inductive boost output capacitor’s positive terminal.

For current sinks that have feedback disabled the over voltage sense mechanism is not in place to protect from potential over-voltage conditions. In this situation the application must ensure that the voltage at LED1 or LED2 doesn’t exceed 40 V.

The default setting for OVP is set at 24 V. For applications that require higher than 24 V at the boost output the OVP threshold has to be programmed to a higher level at power up.

7.3.3.2 Current Limit

The switch current limit for the LM3630A device’s inductive boost is set at 1 A. When the current through the NFET switch hits this over current protection threshold (OCP) the device turns the NFET off and the energy of the inductor is discharged into the output capacitor. Switching is then resumed at the next cycle. The current limit protection circuitry can operate continuously each switch cycle. The result is that during high output power conditions the device can continuously run in current limit. Under these conditions the device inductive boost converter stops regulating the headroom voltage across the high voltage current sinks. This results in a drop in the LED current.

7.3.3.3 Thermal Shutdown

The LM3630A contains thermal shutdown protection. In the event the die temperature reaches 140°C, the boost power supply and current sinks shut down until the die temperature drops to typically 125°C.

7.3.4.1 Initialization Timing With HWEN Tied to V_IN

If the HWEN input is tied to V_IN, then the t_WAIT time starts when V_IN crosses 2.5 V as shown in Figure 78. The initial I²C transaction can occur after the t_WAIT time expires. Any I²C transaction during the t_WAIT period are NAK'ed.

Figure 78. Initialization Timing With HWEN Is Tied to V_IN

7.3.4.2 Initialization Timing With HWEN Driven by GPIO

If the HWEN input is driven by a GPIO then the t_WAITtime starts when HWEW crosses 1.2 V as shown in Figure 79. The initial I²C transaction can occur after the t_WAIT time expires. Any I²C transaction during the t_WAIT period are NAK'ed.

Figure 79. Initialization Timing With HWEN Driven by a GPIO

7.3.4.3 Initialization After Software Reset

The time between the I²C transaction that issues the software reset, and the subsequent I²C transaction (that is, to configure the LM3630A) must be at greater or equal to the t_WAIT period of 1 ms. Any I²C transaction during the t_WAIT period are NAK'ed.

7.4.1.1 Start-Up/Shutdown Ramp

The LED current turn on time from 0 to the initial LED current set-point is programmable. Similarly, the LED current shutdown time to 0 is programmable. Both the startup and shutdown times are independently programmable with 8 different levels. The start-up times are independently programmable from the shutdown times, but not independently programmable for each Control bank. For example, programming a start-up or shutdown time, programs the same ramp time for each control bank. The start-up time is used when the device is first enabled to a non-zero brightness value. The shutdown time is used when the brightness value is programmed to zero. If HWEN is used to disable the device, the action is immediate and the Shutdown time is not used. The zero code does take a small amount of time which is approximately 0.5 ms.

7.4.1.2 Run-Time Ramp

Current ramping from one brightness level to the next is programmable. There are 8 different ramp up times and 8 different ramp down times. The ramp up time is independently programmable from the ramp down time, but not independently programmable for each Control Bank. For example, programming a ramp up time or a ramp down time programs the same ramp time for each control bank. The run time ramps are used whenever the device is enabled with a non-zero brightness value and a new non-zero brightness value is written.

7.4.2.1 PWM Input

The PWM input can be assigned to any control bank. When assigned to a control bank, the programmed current in the control bank also becomes a function of the duty cycle at the PWM input. The PWM input is sampled by a digital circuit which outputs a brightness code that is equivalent to the PWM input duty cycle. The resultant brightness value is a combination of the maximum current setting, the brightness registers, and the equivalent PWM brightness code.

7.4.2.2 PWM Input Frequency

The specified input frequency of the PWM signal is 10 kHz to 80 kHz. The recommended frequency is 30 kHz or greater. The PWM input sampler operates beyond those frequency limits. Performance changes based on the input frequency used. Using frequencies outside the specified range is not recommended. Lower PWM input frequency increases the likelihood that the output of the sampler may change and that a single brightness step may be visible on the screen. This may be visible at low brightness because the step change is large relative to the output level.

7.4.2.3 Recommended Settings

For best performance of the PWM sampler it is recommended to have a PWM input frequency of at least 30 kHz. The Filter Strength (register 50h) must be set to 03h. The Hysteresis 1 bit must be set in register 05h to 1 when setting the maximum current for bank A. For example if max current is 20 mA, register 05h is set to 14h, change that to 94h for 1 bit hysteresis and a smooth min-to-max brightness transition.

7.4.2.4 Adjustments to PWM Sampler

The digital sampler has controls for hysteresis and minimum output brightness which allow the optimization of sampler output. The default hysteresis mode of the PWM sampler requires detecting a two code change in the input to increase brightness. Reducing the hysteresis to change on 1 code allows a smoother brightness transition when the brightness control is swept across the screen in a system. The filter strength bits affect the speed of the output transitions from the PWM sampler. A lower bound to the brightness is enabled by default which limits the minimum output of the PWM sampler to an equivalent code of 6 when the LEDs are turned on. A detected code of 1 is forced to off. A minimum 2% PWM input duty cycle is recommended. Input duty cycles of 1% or less causes delayed off-to-on transitions.

7.4.2.5 Minimum T_ON Pulse Width

The minimum T_ON pulse width required to produce a non-zero output is dependent upon the LM3630A settings. The default setting of the LM3630A requires a minimum of 0.78% duty cycle for the output to be turned on. Because the lower bound feature is enabled, a value of 0.78% (equivalent brightness code 2) up to 2.35% (equivalent brightness code 6) all produce an output equivalent to brightness code 6. At 30-kHz PWM input frequency, the minimum pulse width required to turn on the LEDs is 0.78% × 33 µS = 260 ns.

Because of the hysteresis on the PWM input, this pulse width may not be sufficient to turn on the LEDs. It is recommended that a minimum pulse width of 2% be used. 2% × 33 µS = 660 ns at 30 kHz input frequency.

Disabling the lower bound as described allows a smaller minimum pulse width.

7.5.1.1 Data Validity

The data on SDA line must be stable during the HIGH period of the clock signal (SCL). In other words, state of the data line can only be changed when SCL is LOW.

Figure 83. Data Validity Diagram

A pullup resistor between the VIO line of the controller and SDA must be greater than [(V_IO – V_OL) / 3 mA] to meet the V_OL requirement on SDA. Using a larger pullup resistor results in lower switching current with slower edges, while using a smaller pullup results in higher switching currents with faster edges.

7.5.1.2 Start and Stop Conditions

START and STOP conditions classify the beginning and the end of the I²C session. A START condition is defined as SDA signal transitioning from HIGH to LOW while SCL line is HIGH. A STOP condition is defined as the SDA transitioning from LOW to HIGH while SCL is HIGH. The I²C master always generates START and STOP conditions. The I²C bus is considered to be busy after a START condition and free after a STOP condition. During data transmission, the I²C master can generate repeated START conditions. First START and repeated START conditions are equivalent, function-wise.

Figure 84. Start and Stop Conditions

7.5.1.3 Transferring Data

Every byte put on the SDA line must be eight bits long, with the most significant bit (MSB) transferred first. Each byte of data has to be followed by an acknowledge bit. The acknowledge related clock pulse is generated by the master. The master releases the SDA line (HIGH) during the acknowledge clock pulse. The LM3630A pulls down the SDA line during the 9th clock pulse, signifying an acknowledge. The LM3630A generates an acknowledge after each byte is received.

After the START condition, the I²C master sends a chip address. This address is seven bits long followed by an eighth bit which is a data direction bit (R/W). The LM3630A address is 36h. For the eighth bit, a “0” indicates a WRITE and a “1” indicates a READ. The second byte selects the register to which the data is written. The third byte contains data to write to the selected register.

Figure 85. I²C-Compatible Chip Address (0x36), SEL = 0

Figure 86. I²C-Compatible Chip Address (0x38), SEL = 1

*Code 0 results in approximately 0.5 ms ramp time.

*Code 0 results in approximately 0.5 ms ramp time.

The interrupt status register is cleared upon a read of the register. If the condition that caused the interrupt is still present, then the bit is set to one again and another interrupt is signaled on the INTN output pin. The interrupt status register is not cleared if the device is in sleep mode (Control: SLEEP_STATUS = 1). To disconnect the interrupt condition from the INTN pin during sleep mode, disable the fault connection in the Interrupt Enable register. An interrupt condition sets the status bit and causes an event on the INTN pin only if the corresponding bit in the Interrupt Enable register is one and the Global Enable bit is also one.