9/13
Rev. B, Sep 2005
Applications Information (2)
Input Voltage>4.5V
For applications with input voltages above 4.5V which
could exhibit an overload or short-circuit condition, a
2Ω/1nF series snubber is required between the SW1 pin
and GND. A Schottky diode from SW1 to VIN should also
be added as close to the pins as possible. For the higher
input voltages VIN bypassing becomes more critical,
therefore, a ceramic bypass capacitor as close to the VIN
and GND pins as possible is also required.
Operating Frequency Selection
There
are
several
considerations
in
selecting
the
operating frequency of the converter. The first is, what are
the operating sensitive frequency bands that cannot
tolerate any spectral noise? For example, in products
incorporating
RF
Communications,
the
455KHz
IF
frequency is sensitive to any noise, therefore switching
above 600 KHz is desired. Some communications have
sensitivity to 1.1MHz and in that case a 2MHz converter
frequency may be employed. Other considerations are the
physical size of the converter and efficiency. As the
operating frequency goes up, the inductor and filter
capacitors go down in value and size. The trade off is in
efficiency since the switching losses due to gate charge
are going up proportional with frequency.
Additional quiescent current due to the output switches
GATE charge is given by:
Buck :
500e
-12
VIN F
Boost :
250e
-12
(VIN + VOUT) F
Buck/Boost : F (750e
-12
VIN +250e
-12
Vout)
where F = switching frequency
Closing the Feedback Loop (1)
The ML3440 incorporated voltage mode PWM control.
The control to output gain varies with operation region
(Buck, Boost, Buck-Boost), but is usually no greater than
15. The output filter exhibits a double pole response is
given by:
1
fFilter_pole =
2π
L COUT
Hz
Where COUT is the output filter capacitor
The output filter zero is given by:
1
fFilter_zero =
2π
RESR COUT
Hz
where RESR is the capacitor equivalent series resistance.
A troublesome feature in Boost mode is the right-half plane
zero(rhp), and is given by:
VIN
2
f RHPZ =
2π
Iout L VOUT
Hz
The loop gain is typically rolled off before the RHP zero
frequency.
A
simple
Type
1
compensation
network
can
be
incorporated to stabilize the loop but at cost of reduced
bandwidth and slower transient response. To ensure
proper phase margin, the loop requires to be crossed over
a decade before the LC double pole.
The unity-gain frequency of the error amplifier with the
Type 1 compensation is given by:
1
fUG =
2π R1 CP1
Hz
Most applications demand an improved transient response
to allow a smaller output filter capacitor. To achieve a
higher bandwidth, Type 3 compensation is required.
Two zeros are required to compensate for the double-pole
response.
1
fPOLE1 =
2π 32e
3
R1 CP1
Hz
which is extremely close to DC
1
fZERO1 =
2π RZ CP1
Hz
1
fZERO2 =
2π R1 CZ1
Hz
1
fZERO3 =
2π RZ CP2
Hz
Figure 7.
Error Amplifier with Type 1 Compensation