LM3421_09 Datasheet, PDF (14/24 Pages) National Semiconductor (TI) – N-Channel Controllers for Constant Current LED Drivers

SEPIC application. When it is placed into a Buck converter a
current is set charging the RCT pin set up by the PNP tran-
sistor and resistor network (see Figure 13) the Off Time
T_OFF is controlled to be:
This promotes a constant ripple converter were the ripple cur-
rent magnitude is a function of the input voltage. There is no
output capacitor and the Dimming control MOSFET is shunt-
ing the current away from the LEDs. As the converter is
always in continuous conduction mode the duty factor is set
by the input and output voltages. This fact allows us to give
an equation for selecting the frequency setting components
for the Buck converter. To select a timing resistor use this
equation:
In the above equation R_T is in kΩ, C_T is in nF, and f is in MHz.
One could also select the timing resistor by setting their de-
sired ripple current using the following equation:
For this equation R_T is in kΩ, C_T is in nF, L_CHOKE is in µH, and
I_RIPPLE is in A.
The above describes a buck converter with constant ripple
regardless of V_LED but that varies with V_IN. The LM3421/
LM3423 can also be set up in a buck configuration where the
ripple current varies with V_LED but remains constant over
varying V_IN. See Figure 14 for an example of how to imple-
ment constant ripple vs. V_IN.
INDUCTOR SELECTION
The inductor should be selected such that the switching reg-
ulator maintains continuous inductor current conduction over
the input and output operating voltage and current ranges.
The minimum inductor value is shown in the following equa-
tion for the non-Buck topologies:
COMP pin. However, a two pole system results when an out-
put capacitor is used to reduce the ripple current in the LEDs.
Two pole systems can become unstable because the total
phase shift approaches 180 degrees at unity gain crossover.
A zero in the control compensation is needed; this takes the
form of the resistor in series with the compensation capacitor.
The value of this resistor should be designed to provide the
same RC time constant with the compensation capacitor as
the output capacitor has with the dynamic impedance of the
LED string. If additional phase margin is desired, make the
compensation time constant slower than the output time con-
stant (larger value of resistor).
FAST PWM DIMMING CAPABILITY
These devices provide fast PWM LED dimming, thus enabling
constant LED current for optimal color temperature. The
DDRV pin is meant to drive the gate of an external dimming
MOSFET. This drive will follow the PWM signal applied at the
nDIM pin. The active low nDIM pin can be driven with a PWM
signal up to 50kHz; the brightness of the LEDs can be varied
by modulating the duty cycle of this signal. LED brightness is
approximately proportional to the PWM signal duty cycle, so
30% duty cycle equals approximately 30% LED brightness.
This function can be ignored if PWM dimming is not required
by using nDIM solely as a V_IN UVLO input or by tying it directly
to V_CC or V_IN (if less than 60VDC).
If high side dimming is implemented with a PMOS instead of
an NMOS, the polarity of the dimming MOSFET driver must
be reversed. The LM3423’s DPOL pin is used to set the po-
larity of the DIM driver output, DDRV. Tying DPOL to ground
causes the DDRV pin to be pulled up to V_CC during dim op-
eration, and should be used when driving a PMOS dimming
MOSFET. Note that when high side dimming, the high side
PMOS gate protection zener’s breakdown voltage should be
selected to be roughly equal to the V_CC output voltage of ap-
proximately 7V. See Figure 16 for further information. Tying
DPOL to V_CC or leaving it open causes the DDRV pin to be
low during dim operation and should be used when driving an
NMOS dimming MOSFET.
A minimum on-time must be maintained in order for PWM
dimming to operate in the linear region of its transfer function
(see the graphs Averege LED Current vs. PWM DIM Duty
Cycle and 30kHz PWM Dimming (5% Duty Cycle ON)). Be-
cause the controller is disabled during dimming, the PWM
pulse must be long enough such that the energy intercepted
from the input is greater than or equal to the energy being put
into the LEDs. For a boost and buck-boost regulator, the fol-
lowing condition must be maintained:
In the above equation K should be a value between 3 and 5
depending on the most important application requirements. A
lower value of K results in a smaller, lower cost inductor but
also in higher ripple and lower efficiency. A higher value of K
results in a larger, more costly inductor but will have lower
ripple and higher efficiency.
For the Buck topology the inductor value is selected for a de-
sired ripple current as shown in the previous section.
COMPENSATION
The controllers’ error amplifier is a high output impedance,
transconductance amplifier for easy, single-pin compensa-
tion. This controller is a current mode controller and the
control loop feedback is monitoring the average output (LED)
current. As such it would be expected that the compensation
network could comprise a single capacitor to ground on the
In the previous equation, t_PULSE is the length of the PWM pulse
is seconds, I_LED is the average current in the LEDs in am-
peres, V_LED is the LED stack voltage in volts which is also
often referred to as V_OUT or V_BOOST, L in the inductance in
henries, and V_IN is the input voltage in volts.
BUCK HIGH SPEED DIMMING
These devices are able to implement a constant ripple buck
converter. In this mode the PWM control of LED dimming is
performed by shunting the current away from the LEDs and
through a MOSFET. Please refer to Figure 13 for the circuit
details.
www.national.com
14