determined by the input voltage and load current, with
the worst case occurring at V
IN
= 2 x V
OUT
:
Therefore, when V
IN
is 2 x V
OUT
:
I
RMS
= I
LOAD
/ 2
Output Filter Capacitor Value
The output filter capacitor values are generally deter-
mined by the ESR and voltage-rating requirements
rather than actual capacitance requirements for loop
stability. In other words, the low-ESR electrolytic capac-
itor that meets the ESR requirement usually has more
output capacitance than is required for AC stability.
Use only specialized low-ESR capacitors intended for
switching-regulator applications, such as AVX TPS,
Sprague 595D, Sanyo OS-CON, or Nichicon PL series.
To ensure stability, the capacitor must meet both mini-
mum capacitance and maximum ESR values as given
in the following equations:
C
OUT
> V
REF
(1 + V
OUT
/ V
IN(MIN)
) / V
OUT
x R
SENSE
x f
R
ESR
< R
SENSE
x V
OUT
/ V
REF
where R
ESR
can be multiplied by 1.5, as discussed
below.
These equations are worst case, with 45 degrees of
phase margin to ensure jitter-free, fixed-frequency
operation, and provide a nicely damped output
response for zero to full-load step changes. Some cost-
conscious designers may wish to bend these rules with
less-expensive capacitors, particularly if the load lacks
large step changes. This practice is tolerable if some
bench testing over temperature is done to verify
acceptable noise and transient response.
No well-defined boundary exists between stable and
unstable operation. As phase margin is reduced, the
first symptom is timing jitter, which shows up as blurred
edges in the switching waveforms where the scope
does not quite sync up. Technically speaking, this jitter
(usually harmless) is unstable operation, since the duty
factor varies slightly. As capacitors with higher ESRs
are used, the jitter becomes more pronounced, and the
load-transient output voltage waveform starts looking
ragged at the edges. Eventually, the load-transient
waveform has enough ringing on it that the peak noise
levels exceed the allowable output voltage tolerance.
Note that even with zero phase margin and gross insta-
bility, the output voltage noise never gets much worse
than I
PEAK
x R
ESR
(under constant loads).
Designers of RF communicators or other noise-sensi-
tive analog equipment should be conservative and stay
within the guidelines. Designers of notebook computers
and similar commercial-temperature-range digital sys-
tems can multiply the R
ESR
value by a factor of 1.5
without hurting stability or transient response.
The output voltage ripple, which is usually dominated
by the filter capacitor’s ESR, can be approximated as
I
RIPPLE
x R
ESR
. There is also a capacitive term, so the
full equation for ripple in continuous-conduction mode
is V
NOISE(p-p)
= I
RIPPLE
x [R
ESR
+ 1 / (2 x p x f x
C
OUT
)]. In Idle Mode, the inductor current becomes
discontinuous, with high peaks and widely spaced
pulses, so the noise can actually be higher at light load
(compared to full load). In Idle Mode, calculate the out-
put ripple as follows:
Selecting Other Components
MOSFET Switches
The high-current N-channel MOSFETs must be logic-
level types with guaranteed on-resistance specifica-
tions at VGS = 4.5V. Lower gate-threshold
specifications are better (i.e., 2V max rather than 3V
max). Drain-source breakdown voltage ratings must at
least equal the maximum input voltage, preferably with
a 20% derating factor. The best MOSFETs have the
lowest on-resistance per nanocoulomb of gate charge.
Multiplying R
DS(ON)
by Q
g
provides a good figure of
merit for comparing various MOSFETs. Newer MOSFET
process technologies with dense cell structures gener-
ally perform best. The internal gate drivers tolerate
>100nC total gate charge, but 70nC is a more practical
upper limit to maintain best switching times.
In high-current applications, MOSFET package power
dissipation often becomes a dominant design factor.
I
2
R power losses are the greatest heat contributor for
both high-side and low-side MOSFETs. I
2
R losses are
distributed between Q1 and Q2 according to duty fac-
tor as shown in the equations below. Generally, switch-
ing losses affect only the upper MOSFET, since the
Schottky rectifier usually clamps the switching node
before the synchronous rectifier turns on. Gate-charge
losses are dissipated by the driver and do not heat the
MOSFET. Calculate the temperature rise according to
package thermal-resistance specifications to ensure
V
0.02 x R
R
0.0003 x L x 1/V 1/ V V
R x C
NOISE(p p)
ESR
SENSE
OUT IN OUT
SENSE
2
F
=+
+−
()
[]
()
II VVVV
RMS LOAD OUT IN OUT IN
()
/
MAX1636
Low-Voltage, Precision Step-Down
Controller for Portable CPU Power
______________________________________________________________________________________ 19
MAX1636
Low-Voltage, Precision Step-Down
Controller for Portable CPU Power
20 ______________________________________________________________________________________
that both MOSFETs are within their maximum junction
temperature at high ambient temperature. The worst-
case dissipation for the high-side MOSFET occurs at
both extremes of input voltage, and the worst-case dis-
sipation for the low-side MOSFET occurs at maximum
input voltage.
Duty = (V
OUT
+ V
Q2
) / (V
IN
- V
Q1
)
PD (upper FET) = I
LOAD
2
x R
DS(ON)
x Duty + V
IN
x
I
LOAD
x f x [(V
IN
x C
RSS
) / I
GATE
+ 20ns]
PD (lower FET) = I
LOAD
2
x R
DS(ON)
x (1 - Duty)
where on-state voltage drop V
Q
= I
LOAD
x R
DS(ON)
,
C
RSS
= MOSFET reverse transfer capacitance, I
GATE
=
DH driver peak output current capability (1A typ), and
20ns = DH driver inherent rise/fall time. The MAX1636’s
output undervoltage shutdown protects the synchro-
nous rectifier under output short-circuit conditions. To
reduce EMI, add a 0.1µF ceramic capacitor from the
high-side switch drain to the low-side switch source.
Rectifier Clamp Diode
The rectifier is a clamp across the low-side MOSFET
that catches the negative inductor swing during the
60ns dead time between turning one MOSFET off and
each low-side MOSFET on. The latest generations of
MOSFETs incorporate a high-speed silicon body
diode, which serves as an adequate clamp diode if
efficiency is not of primary importance. A Schottky
diode can be placed in parallel with the body diode to
reduce the forward voltage drop, typically improving
efficiency 1% to 2%. Use a diode with a DC current rat-
ing equal to one-third of the load current; for example,
use an MBR0530 (500mA-rated) type for loads up to
1.5A, a 1N5819 type for loads up to 3A, or a 1N5822
type for loads up to 10A. The rectifier’s rated reverse-
breakdown voltage must be at least equal to the maxi-
mum input voltage, preferably with a 20% derating
factor.
Boost-Supply Diode
A signal diode such as a 1N4148 works well in most
applications. If the input voltage can go below +6V,
use a small (20mA) Schottky diode for slightly
improved efficiency and dropout characteristics. Do
not use large power diodes, such as 1N5817 or
1N4001, since high junction capacitance can pump up
VL to excessive voltages.
Low-Voltage Operation
Low input voltages and low input-output differential
voltages each require extra care in their design. Low
absolute input voltages can cause the VL linear regula-
tor to enter dropout and eventually shut itself off. Low
V
IN
- V
OUT
differentials can cause the output voltage to
sag when the load current changes abruptly. The sag’s
amplitude is a function of inductor value and maximum
duty factor (D
MAX
, an Electrical Characteristics
parameter, 98% guaranteed over temperature at f =
200kHz) as follows:
Table 6 is a low-voltage troubleshooting guide. The
cure for low-voltage sag is to increase the output
capacitor’s value. For example, at V
IN
= +5.5V, V
OUT
=
5V, L = 10µH, f = 200kHz, and I
STEP
= 3A, a total
capacitance of 660µF keeps the sag less than 200mV.
Note that only the capacitance requirement increases;
the ESR requirements do not change. Therefore, the
added capacitance can be supplied by a low-cost bulk
capacitor in parallel with the normal low-ESR capacitor.
__________Applications Information
Heavy-Load Efficiency Considerations
The major efficiency-loss mechanisms under loads are
as follows, in the usual order of importance:
P(I
2
R) = I
2
R losses
P(tran) = transition losses
P(gate) = gate-charge losses
P(diode) = diode-conduction losses
P(cap) = capacitor ESR losses
P(IC) = losses due to the IC’s operating supply
current
Inductor core losses are fairly low at heavy loads
because the inductor’s AC current component is small.
Therefore, they are not accounted for in this analysis.
Ferrite cores are preferred, especially at 300kHz, but
powdered cores, such as Kool-Mu, can also work well.
Efficiency = P
OUT
/ P
IN
x 100%
= P
OUT
/ (P
OUT
+ P
TOTAL
) x 100%
P
TOTAL
= P(I
2
R) + P(tran) + P(gate) + P(diode)
+ P(cap) + P(IC)
P = (I
2
R) = (I
LOAD
)
2
x (R
DC
+ R
DS(ON)
+R
SENSE
)
V
IxL
xC V xD V
SAG
STEP
F IN MIN MAX OUT
()
( )
()
=
×−
2
2
where R
DC
is the DC resistance of the coil, R
DS(ON)
is
the MOSFET on-resistance, and R
SENSE
is the current-
sense resistor value. The R
DS(ON)
term assumes identi-
cal MOSFETs for the high-side and low-side switches
because they time-share the inductor current. If the
MOSFETs are not identical, their losses can be estimat-
ed by averaging the losses according to duty factor.
PD(tran) = transition loss = V
IN
x I
LOAD
x f x 3/2 x
[(V
IN
C
RSS
/ I
GATE
) + 20ns]
where C
RSS
is the reverse transfer capacitance of the
high-side MOSFET (a data-sheet parameter), I
GATE
is
the DH gate-driver peak output current (1.5A typ), and
20ns is the rise/fall time of the DH driver (20ns typ).
P(gate) = Q
g
x f x VL
where VL is the internal logic-supply voltage (+5V), and
Q
g
is the sum of the gate-charge values for low-side
and high-side switches. For matched MOSFETs, Q
g
is
twice the data-sheet value of an individual MOSFET. If
V
OUT
is set to less than 4.5V, replace VL in this equa-
tion with V
BATT
. In this case, efficiency can be
improved by connecting VL to an efficient 5V source,
such as the system +5V supply.
P(diode) = diode conduction losses =
I
LOAD
x V
FWD
x t
D
x f
where t
D
is the diode-conduction time (120ns typ), and
V
FWD
is the forward voltage of the diode. This power is
dissipated in the MOSFET body diode if no external
Schottky diode is used.
P(cap) = input capacitor ESR loss = I
RMS
2
x R
ESR
where I
RMS
is the input ripple current as calculated in
the Input Capacitor Value section.
Light-Load Efficiency Considerations
Under light loads, the PWM operates in discontinuous
mode, where the inductor current discharges to zero at
some point during the switching cycle. This makes the
inductor current’s AC component high compared to the
load current, which increases core losses and I
2
R loss-
es in the output filter capacitors. For best light-load effi-
ciency, use MOSFETs with moderate gate-charge
levels and use ferrite, MPP, or other low-loss core mate-
rial. Avoid powdered-iron cores; even Kool-Mu
(aluminum alloy) is not as good as ferrite.
PC Board Layout Considerations
Good PC board layout is required in order to achieve
specified noise, efficiency, and stable performance.
The PC board layout artist must be given explicit
instructions, preferably a pencil sketch showing the
placement of power-switching components and high-
current routing. See the PC board layout in the
MAX1636 evaluation kit manual for examples. A ground
plane is essential for optimum performance. In most
applications, the circuit will be located on a multi-layer
board, and full use of the four or more copper layers
is recommended. Use the top layer for high-current
MAX1636
Low-Voltage, Precision Step-Down
Controller for Portable CPU Power
______________________________________________________________________________________ 21
CONDITION
Low V
IN
-V
OUT
differential, <1V
Maximum duty-cycle limits
exceeded.
Low V
IN
-V
OUT
differential, <1.5V
Dropout voltage is too high
(V
OUT
follows V
IN
as V
IN
decreases)
Reduce operation to 200kHz.
Reduce MOSFET on-resistance and
coil DCR.
SYMPTOM
Low V
IN
-V
OUT
differential, <0.5V
ROOT CAUSE
Normal function of internal
low-dropout circuitry.
Unstable—jitters between
different duty factors and
frequencies
Limited inductor-current
slew rate per cycle.
Increase the minimum input voltage
or ignore.
Low input voltage, <4.5V
VL output is so low that it
hits the VL UVLO threshold.
Low input voltage, <5V
Won’t start under load or
quits before battery is
completely dead
Supply VL from an external source
other than V
IN
, such as the system
+5V supply.
VL linear regulator is going
into dropout and isn’t provid-
ing good gate-drive levels.
SOLUTION
Poor efficiency
Use a small 20mA Schottky diode
for boost diode. Supply VL from an
external source.
Sag or droop in V
OUT
under
step-load change
Increase bulk output capacitance
per formula (see Low-Voltage
Operation section). Reduce inductor
value.
Table 6. Low-Voltage Troubleshooting Chart

MAX1636EAP+

Mfr. #:
Manufacturer:
Maxim Integrated
Description:
LDO Voltage Controllers Precision Step-Down for Portable CPU Pwr
Lifecycle:
New from this manufacturer.
Delivery:
DHL FedEx Ups TNT EMS
Payment:
T/T Paypal Visa MoneyGram Western Union

Products related to this Datasheet