The SC1185 combines a synchronous voltage mode
controller with two low-dropout linear regulators
providing most of the circuitry necessary to
implement three DC/DC converters for powering
advanced microprocessors such as Pentium
The SC1185 switching section features an integrated
5 bit D/A converter, pulse by pulse current limiting,
integrated power good signaling, and logic compatible
shutdown. The SC1185 switching section operates at
a fixed frequency of 140kHz, providing an optimum
compromise between size, efficiency and cost in the
intended application areas. The integrated D/A converter provides programmability of output voltage
from 2.0V to 3.5V in 100mV increments and 1.30V to
2.05V in 50mV increments with no external components.
The SC1185 linear sections are low dropout regulators supplying 1.5V for GTL bus and 2.5V for nonGTL I/O. The Reference voltage is made available for
external linear regulators.
-0.3 to +7V
PGND to GND± 1V
BST to GND-0.3 to +15V
Operating Temperature RangeT
Junction Temperature RangeT
Storage Temperature RangeT
Lead Temperature (Soldering) 10 secondsT
Thermal Impedance Junction to Ambient
1AGNDSmall Signal Analog and Digital Ground
2GATE1Gate Drive Output LDO1
3LDOS1Sense Input for LDO1
4LDOS2Sense Input for LDO2
5VCCInput Voltage
6REFBuffered Reference Voltage output
Open collector logic output, high if V
(1)
7PWRGOOD
8CS-Current Sense Input (negative)
9CS+Current Sense Input (positive)
10PGNDHPower Ground for High Side Switch
11DHHigh Side Driver Output
12PGNDLPower Ground for Low Side Switch
13DLLow Side Driver Output
14BSTLSupply for Low Side Driver
15BSTHSupply for High Side Driver
16EN
Careful attention to layout requirements are necessary
for successful implementation of the SC1185 PWM
controller. High currents switching at 140kHz are present in the application and their effect on ground plane
voltage differentials must be understood and minimized.
1). The high power parts of the circuit should be laid out
first. A ground plane should be used, the number and
position of ground plane interruptions should be such
as to not unnecessarily compromise ground plane integrity. Isolated or semi-isolated areas of the ground
plane may be deliberately introduced to constrain
ground currents to particular areas, for example the input capacitor and bottom FET ground.
2). The loop formed by the Input Capacitor(s) (Cin), the
Top FET (Q1) and the Bottom FET (Q2) must be kept
10
1
AGND
2
GATE1
3
LDOS1
4Q1
5
6
7
8
9
10
11
12
LDOS2
VCC
REF
PWRGOOD
CSCS+
PGNDH
DH
PGNDL
0.1uF
0.1uF
GATE2
LDVO
VID0
VID1
VID2
VID3
VID4
VOSENSE
EN
BSTH
BSTL
DL
SC1185
as small as possible. This loop contains all the high current, fast transition switching. Connections should be as
wide and as short as possible to minimize loop inductance. Minimizing this loop area will a) reduce EMI, b)
lower ground injection currents, resulting in electrically
“cleaner” grounds for the rest of the system and c) minimize source ringing, resulting in more reliable gate
switching signals.
3). The connection between the junction of Q1, Q2 and
the output inductor should be a wide trace or copper
region. It should be as short as practical. Since this
connection has fast voltage transitions, keeping this
connection short will minimize EMI. The connection between the output inductor and the sense resistor should
be a wide trace or copper area, there are no fast voltage or current transitions in this connection and length
is not so important, however adding unnecessary
impedance will reduce efficiency.
4) The Output Capacitor(s) (Cout) should be located
as close to the load as possible, fast transient load
currents are supplied by Cout only, and connections
between Cout and the load must be short, wide copper areas to minimize inductance and resistance.
5) The SC1185 is best placed over a quiet ground
plane area, avoid pulse currents in the Cin, Q1, Q2
loop flowing in this area. PGNDH and PGNDL should
be returned to the ground plane close to the package.
The AGND pin should be connected to the ground
side of (one of) the output capacitor(s). If this is not
possible, the AGND pin may be connected to the
ground path between the Output Capacitor(s) and the
Cin, Q1, Q2 loop. Under no circumstances should
AGND be returned to a ground inside the Cin, Q1, Q2
loop.
6) Vcc for the SC1185 should be supplied from the 5V
5V
supply through a 10Ω resistor, the Vcc pin should be
decoupled directly to AGND by a 0.1µF ceramic capacitor, trace lengths should be as short as possible.
7) The Current Sense resistor and the divider across
it should form as small a loop as possible, the traces
running back to CS+ and CS- on the SC1185 should
run parallel and close to each other. The 0.1µF capacitor should be mounted as close to the CS+ and
CS- pins as possible.
8) Ideally, the grounds for the two LDO sections
should be returned to the ground side of (one of) the
output capacitor(s).
SWITCHING SECTION
OUTPUT CAPACITORS - Selection begins with the
most critical component. Because of fast transient load
current requirements in modern microprocessor core
supplies, the output capacitors must supply all transient
load current requirements until the current in the output
inductor ramps up to the new level. Output capacitor
ESR is therefore one of the most important criteria. The
maximum ESR can be simply calculated from:
V
R
ESR
t
≤
I
t
Where
=
t
=
t
step current Transient I
For example, to meet a 100mV transient limit with a
10A load step, the output capacitor ESR must be less
than 10mΩ. To meet this kind of ESR level, there are
three available capacitor technologies.
The choice of which to use is simply a cost/performance issue, with Low ESR Aluminum being the
cheapest, but taking up the most space.
INDUCTOR - Having decided on a suitable type and
value of output capacitor, the maximum allowable
value of inductor can be calculated. Too large an inductor will produce a slow current ramp rate and will
cause the output capacitor to supply more of the transient load current for longer - leading to an output voltage sag below the ESR excursion calculated above.
The maximum inductor value may be calculated from:
CR
ESR
L
()
I
t
VV
−≤
OIN
The calculated maximum inductor value assumes 100%
duty cycle, so some allowance must be made. Choosing
an inductor value of 50 to 75% of the calculated maximum will guarantee that the inductor current will ramp
excursion voltage transient MaximumV
Qty.
Rqd.C(µF)
ESR
(mΩ)
fast enough to reduce the voltage dropped across the
ESR at a faster rate than the capacitor sags, hence ensuring a good recovery from transient with no additional
excursions.
We must also be concerned with ripple current in the
output inductor and a general rule of thumb has been to
allow 10% of maximum output current as ripple current.
Note that most of the output voltage ripple is produced
by the inductor ripple current flowing in the output capacitor ESR. Ripple current can be calculated from:
V
I
L
RIPPLE
IN
=
fL4
⋅⋅
OSC
Ripple current allowance will define the minimum permitted inductor value.
POWER FETS - The FETs are chosen based on several
criteria with probably the most important being power
dissipation and power handling capability.
TOP FET - The power dissipation in the top FET is a
combination of conduction losses, switching losses and
bottom FET body diode recovery losses.
a) Conduction losses are simply calculated as:
2
RIP
OCOND
δ⋅⋅=
)on(DS
where
V
O
≈δ
cycleduty =
V
IN
b) Switching losses can be estimated by assuming a
switching time, if we assume 100ns then:
2
−
10VIP
⋅⋅=
INOSW
or more generally,
f)tt(VI
⋅+⋅⋅
P
=
SW
4
c) Body diode recovery losses are more difficult to estimate, but to a first approximation, it is reasonable to assume that the stored charge on the bottom FET body
diode will be moved through the top FET as it starts to
turn on. The resulting power dissipation in the top FET
will be:
fVQP
⋅⋅=
OSCINRRRR
To a first order approximation, it is convenient to only
consider conduction losses to determine FET suitability.
For a 5V in; 2.8V out at 14.2A requirement, typical FET
losses would be:
BOTTOM FET - Bottom FET losses are almost entirely
due to conduction. The body diode is forced into conduction at the beginning and end of the bottom switch conduction period, so when the FET turns on and off, there
is very little voltage across it, resulting in low switching
losses. Conduction losses for the FET can be determined by:
2
OCOND
)1(RIP
δ−⋅⋅=
)on(DS
For the example above:
FET type
R
DS(on)
(mΩ)
IRL34025151.33D
IRL220310.50.93D
(W)Package
P
D
2
PAK
2
PAK
Si4410201.77SO-8
INPUT CAPACITORS - since the RMS ripple current in
the input capacitors may be as high as 50% of the output current, suitable capacitors must be chosen accordingly. Also, during fast load transients, there may
be restrictions on input di/dt. These restrictions require
useable energy storage within the converter circuitry,
either as extra output capacitance or, more usually,
additional input capacitors. Choosing low ESR input
capacitors will help maximize ripple rating for a given
size.
Each of the package types has a characteristic thermal
impedance, for the TO-220 package, thermal impedance
is mostly determined by the heatsink used. For the surface mount packages on double sided FR4, 2 oz printed
circuit board material, thermal impedances of 40
for the D
2
PAK and 80oC/W for the SO-8 are readily
o
C/W
achievable. The corresponding temperature rise is detailed below:
Temperature rise (oC)
FET typeTop FETBottom FET
IRL3402567.653.2
IRL220347.637.2
Si4410180.8141.6
It is apparent that single SO-8 Si4410 are not adequate for
this application, but by using parallel pairs in each position, power dissipation will be approximately halved and
temperature rise reduced by a factor of 4.