ADP3212A, NCP3218A
http://onsemi.com
25
winding’s average current (20 A) plus the ac core loss. In this
example, 330 nH is used.
Another important factor in the inductor design is the
DCR, which is used for measuring the phase currents. Too
large of a DCR causes excessive power losses, whereas too
small of a value leads to increased measurement error. For
this example, an inductor with a DCR of 0.8 mW is used.
Selecting a Standard Inductor
After the inductance and DCR are known, select a
standard inductor that best meets the overall design goals. It
is also important to specify the inductance and DCR
tolerance to maintain the accuracy of the system. Using 20%
tolerance for the inductance and 15% for the DCR at room
temperature are reasonable values that most manufacturers
can meet.
Power Inductor Manufacturers
The following companies provide surfacemount power
inductors optimized for high power applications upon
request:
Vishay Dale Electronics, Inc.
(605) 6659301
Panasonic
(714) 3737334
Sumida Electric Company
(847) 5456700
NEC Tokin Corporation
(510) 3244110
Output Droop Resistance
The design requires that the regulator output voltage
measured at the CPU pins decreases when the output current
increases. The specified voltage drop corresponds to the
droop resistance (R
O
).
The output current is measured by summing the currents
of the resistors monitoring the voltage across each inductor
and by passing the signal through a lowpass filter. The
summing is implemented by the CS amplifier that is
configured with resistor R
PH(x)
(summer) and resistors R
CS
and C
CS
(filters). The output resistance of the regulator is set
by the following equations:
R
O
+
R
CS
R
PH(x)
R
SENSE
(eq. 6)
C
CS
+
L
R
SENSE
R
CS
(eq. 7)
where R
SENSE
is the DCR of the output inductors.
Either R
CS
or R
PH(x)
can be chosen for added flexibility.
Due to the current drive ability of the CSCOMP pin, the R
CS
resistance should be greater than 100 kW. For example,
initially select R
CS
to be equal to 200 kW, and then use
Equation 7 to solve for C
CS
:
C
CS
+
330 nH
0.8 mW 200 kW
+ 2.1 nF
If C
CS
is not a standard capacitance, R
CS
can be tuned. For
example, if the optimal C
CS
capacitance is 1.5 nF, adjust R
CS
to 280 kW. For best accuracy, C
CS
should be a 5% NPO
capacitor. In this example, a 220 kW is used for R
CS
to
achieve optimal results.
Next, solve for R
PH(x)
by rearranging Equation 6 as
follows:
R
PH(x)
w
0.8 mW
2.1 mW
220 kW + 83.8 kW
The standard 1% resistor for R
PH(x)
is 86.6 kW.
Inductor DCR Temperature Correction
If the DCR of the inductor is used as a sense element and
copper wire is the source of the DCR, the temperature
changes associated with the inductors winding must be
compensated for. Fortunately, copper has a wellknown
Temperature Coefficient (TC) of 0.39%/°C.
If R
CS
is designed to have an opposite but equal
percentage of change in resistance, it cancels the
temperature variation of the inductors DCR. Due to the
nonlinear nature of NTC thermistors, series resistors R
CS1
and R
CS2
(see Figure 30) are needed to linearize the NTC and
produce the desired temperature coefficient tracking.
Figure 30. TemperatureCompensation Circuit Values
ADP3212
CSCOMP
CSSUM
CSREF
+
Place as close as possible
to nearest inductor
To Switch Nodes
Keep This Path As Short
As Possible And Well Away
From Switch Node Lines
R
PH3
To V
OUT
Sense
R
PH2
R
PH1
R
CS2
R
CS1
C
CS1
C
CS2
R
TH
19
18
17
ADP3212A, NCP3218A
http://onsemi.com
26
The following procedure and expressions yield values for
R
CS1
, R
CS2
, and R
TH
(the thermistor value at 25°C) for a
given R
CS
value.
1. Select an NTC to be used based on its type and
value. Because the value needed is not yet
determined, start with a thermistor with a value
close to R
CS
and an NTC with an initial tolerance
of better than 5%.
2. Find the relative resistance value of the NTC at
two temperatures. The appropriate temperatures
will depend on the type of NTC, but 50°C and
90°C have been shown to work well for most types
of NTCs. The resistance values are called A (A is
R
TH
(50°C)/R
TH
(25°C)) and B (B is
R
TH
(90°C)/R
TH
(25°C)). Note that the relative
value of the NTC is always 1 at 25°C.
3. Find the relative value of R
CS
required for each of
the two temperatures. The relative value of R
CS
is
based on the percentage of change needed, which
is initially assumed to be 0.39%/°C in this
example.
The relative values are called r
1
(r
1
is 1/(1+ TC ×
(T
1
25))) and r
2
(r
2
is 1/(1 + TC × (T
2
25))),
where TC is 0.0039, T
1
is 50°C, and T
2
is 90°C.
4. Compute the relative values for r
CS1
, r
CS2
, and r
TH
by using the following equations:
r
CS2
+
(AB) r
1
r
2
* A (1B) r
2
) B (1A) r
1
A (1 * B) r
1
* B (1 * A) r
2
* (A * B)
(eq. 8)
r
CS1
+
(1 * A)
1
1*r
CS2
*
A
r
1
*r
CS2
r
TH
+
1
1
1*r
CS2
*
1
r
CS1
5. Calculate R
TH
= r
TH
× R
CS
, and then select a
thermistor of the closest value available. In
addition, compute a scaling factor k based on the
ratio of the actual thermistor value used relative to
the computed one:
k +
R
TH(ACTUAL)
R
TH(CALCULATED)
(eq. 9)
6. Calculate values for R
CS1
and R
CS2
by using the
following equations:
R
CS1
+ R
CS
k r
CS1
(eq. 10)
R
CS2
+ R
CS
ǒ
(1 * k) ) (k r
CS2
)
Ǔ
For example, if a thermistor value of 100 kW is selected
in Step 1, an available 0603size thermistor with a value
close to R
CS
is the Vishay NTHS0603N04 NTC thermistor,
which has resistance values of A = 0.3359 and B = 0.0771.
Using the equations in Step 4, r
CS1
is 0.359, r
CS2
is 0.729,
and r
TH
is 1.094. Solving for r
TH
yields 241 kW, so a
thermistor of 220 kW would be a reasonable selection,
making k equal to 0.913. Finally, R
CS1
and R
CS2
are found
to be 72.1 kW and 166 kW. Choosing the closest 1% resistor
for R
CS2
yields 165 kW. To correct for this approximation,
73.3 kW is used for R
CS1
.
C
OUT
Selection
The required output decoupling for processors and
platforms is typically recommended by Intel. For systems
containing both bulk and ceramic capacitors, however, the
following guidelines can be a helpful supplement.
Select the number of ceramics and determine the total
ceramic capacitance (C
Z
). This is based on the number and
type of capacitors used. Keep in mind that the best location
to place ceramic capacitors is inside the socket; however, the
physical limit is twenty 0805size pieces inside the socket.
Additional ceramic capacitors can be placed along the outer
edge of the socket. A combined ceramic capacitor value of
200 mF to 300 mF is recommended and is usually composed
of multiple 10 mF or 22 mF capacitors.
Ensure that the total amount of bulk capacitance (C
X
) is
within its limits. The upper limit is dependent on the VID
OTF output voltage stepping (voltage step, V
V
, in time, t
V
,
with error of V
ERR
); the lower limit is based on meeting the
critical capacitance for load release at a given maximum load
step, DI
O
. The current version of the IMVP6.5
specification allows a maximum V
CORE
overshoot
(V
OSMAX
) of 10 mV more than the VID voltage for a
stepoff load current.
C
X(MIN)
w
ȧ
ȧ
ȡ
Ȣ
L DI
O
n
ǒ
R
O
)
V
OSMAX
DI
O
Ǔ
V
VID
* C
Z
ȧ
ȧ
ȣ
Ȥ
(eq. 11)
where k + ln
ǒ
V
ERR
V
V
Ǔ
(eq. 12)
C
X(MAX)
v
L
n k
2
R
O
2
V
V
V
VID
ȧ
ȡ
Ȣ
1 )
ǒ
t
v
V
VID
V
V
n k R
O
L
Ǔ
2
Ǹ
* 1
ȧ
ȣ
Ȥ
* C
Z
ADP3212A, NCP3218A
http://onsemi.com
27
To meet the conditions of these expressions and the
transient response, the ESR of the bulk capacitor bank (R
X
)
should be less than two times the droop resistance, R
O
. If the
C
X(MIN)
is greater than C
X(MAX)
, the system does not meet
the VID OTF and/or the deeper sleep exit specifications and
may require less inductance or more phases. In addition, the
switching frequency may have to be increased to maintain
the output ripple.
For example, if 30 pieces of 10 mF, 0805size MLC
capacitors (C
Z
= 300 mF) are used, the fastest VID voltage
change is when the device exits deeper sleep, during which
the V
CORE
change is 220 mV in 22 ms with a setting error of
10 mV. If k = 3.1, solving for the bulk capacitance yields
ȧ
ȡ
Ȣ
330 nH 27.9 A
2
ǒ
2.1 mW )
10 mV
27.9 A
Ǔ
1.4375 V
* 300 mF
ȧ
ȣ
Ȥ
+ 1.0 mF
C
X(MAX)
v
330 nH 220 mV
2 3.1
2
(2.1 mW)
2
1.4375 V
C
X(MIN)
w
ǒ
1 )
ǒ
22ms 1.4375V 2 3.1 2.1mW
220 mV 490 nH
Ǔ
2
Ǹ
1
Ǔ
300 mF
+ 21 mF
Using six 330 mF Panasonic SP capacitors with a typical
ESR of 7 mW each yields C
X
= 1.98 mF and R
X
= 1.2 mW.
Ensure that the ESL of the bulk capacitors (L
X
) is low
enough to limit the high frequency ringing during a load
change. This is tested using:
L
X
v 300 mF (2.1 mW)
2
2 + 2nH
(eq. 13)
L
X
v C
Z
R
O
2
Q
2
where:
Q is limited to the square root of 2 to ensure a critically
damped system.
L
X
is about 150 pH for the six SP capacitors, which is low
enough to avoid ringing during a load change. If the L
X
of
the chosen bulk capacitor bank is too large, the number of
ceramic capacitors may need to be increased to prevent
excessive ringing.
For this multimode control technique, an all ceramic
capacitor design can be used if the conditions of
Equations 11, 12, and 13 are satisfied.
Power MOSFETs
For typical 20 A per phase applications, the Nchannel
power MOSFETs are selected for two highside switches
and two or three lowside switches per phase. The main
selection parameters for the power MOSFETs are V
GS(TH)
,
Q
G
, C
ISS
, C
RSS
, and R
DS(ON)
. Because the voltage of the
gate driver is 5.0 V, logiclevel threshold MOSFETs must be
used.
The maximum output current, I
O
, determines the R
DS(ON)
requirement for the lowside (synchronous) MOSFETs. In
the APD3212A/NCP3218A, currents are balanced between
phases; the current in each lowside MOSFET is the output
current divided by the total number of MOSFETs (n
SF
).
With conduction losses being dominant, the following
expression shows the total power that is dissipated in each
synchronous MOSFET in terms of the ripple current per
phase (I
R
) and the average total output current (I
O
):
P
SF
+ (1D)
ƪ
ǒ
I
O
n
SF
Ǔ
2
)
1
12
ǒ
n I
R
n
SF
Ǔ
2
ƫ
R
DS(SF)
(eq. 14)
where:
D is the duty cycle and is approximately the output voltage
divided by the input voltage.
I
R
is the inductor peaktopeak ripple current and is
approximately
I
R
+
(1 * D) V
OUT
L f
SW
Knowing the maximum output current and the maximum
allowed power dissipation, the user can calculate the
required R
DS(ON)
for the MOSFET. For 8lead SOIC or
8lead SOIC compatible MOSFETs, the junctionto
ambient (PCB) thermal impedance is 50°C/W. In the worst
case, the PCB temperature is 70°C to 80°C during heavy
load operation of the notebook, and a safe limit for P
SF
is
about 0.8 W to 1.0 W at 120°C junction temperature.
Therefore, for this example (40 A maximum), the R
DS(SF)
per
MOSFET is less than 8.5 mW for two pieces of lowside
MOSFETs. This R
DS(SF)
is also at a junction temperature of
about 120°C; therefore, the R
DS(SF)
per MOSFET should be
less than 6 mW at room temperature, or 8.5 mW at high
temperature.
Another important factor for the synchronous MOSFET
is the input capacitance and feedback capacitance. The ratio
of the feedback to input must be small (less than 10% is
recommended) to prevent accidentally turning on the
synchronous MOSFETs when the switch node goes high.
The highside (main) MOSFET must be able to handle
two main power dissipation components: conduction losses
and switching losses. Switching loss is related to the time for
the main MOSFET to turn on and off and to the current and
voltage that are being switched. Basing the switching speed
on the rise and fall times of the gate driver impedance and
MOSFET input capacitance, the following expression
provides an approximate value for the switching loss per
main MOSFET:
P
S(MF)
+ 2 f
SW
V
DC
I
O
n
MF
R
G
n
MF
n
C
ISS
(eq. 15)
where:
n
MF
is the total number of main MOSFETs.
R
G
is the total gate resistance.
C
ISS
is the input capacitance of the main MOSFET.

NCP3218AMNR2G

Mfr. #:
Manufacturer:
ON Semiconductor
Description:
Switching Controllers 3 PHASE BUCK CONTROLLER
Lifecycle:
New from this manufacturer.
Delivery:
DHL FedEx Ups TNT EMS
Payment:
T/T Paypal Visa MoneyGram Western Union

Products related to this Datasheet