Fairchild RC5042, RC5040 User Manual

Application Note 42

Implementing the RC5040 and RC5042
DC-DC Converters on Pentium

®

Pro Motherboards

www.fairchildsemi.com

Introduction

This document describes how to implement a switching volt­age regulator using an RC5040 or an RC5042 high speed
controller, a power inductor, a Schottky diode, appropriate
capacitors, and external power MOSFETs. This regulator
forms a step down DC-DC converter that can deliver up to

14.5A of continuous load current at voltages ranging from

2.1V to 3.5V. A specific application circuit, design consider­ations, component selection, PCB layout guidelines and per­formance evaluation procedures are covered in detail.

In the past 10 years, microprocessors have ev olved at such an
exponential rate that a modern chip can rival the computing
power of a mainframe computer. Such evolution has been
possible because of the increasing numbers of transistors that
processors integrate. Pentium CPUs, for example, integrate
well over 5 million transistors on a single piece of silicon.

To integrate so many transistors on a piece of silicon, their
physical geometry has been reduced to the sub-micron level.
As a result of each geometry reduction, the corresponding
operational voltage for each transistor has also been reduced.
This changing voltage for the CPU demands the design of a
programmable power supply—a design that is not com­pletely re-engineered with every change in CPU voltage.

The operational voltage of CPUs has shown a downwards
trend for the past 5 years: from 5V for the x386 and x486, to

3.3V for Pentium, and 3.1V for Pentium Pro. Furthermore,
emerging chip technologies may require operating voltages
as low as 2.5V. With this trend in mind, Raytheon Electron­ics has designed the RC5040 and RC5042 controllers. These
controllers integrate the necessary programmability to
address the changing power supply requirements of lower
voltage CPUs.

Previous generations of DC-DC converter controllers were
designed with fixed output voltages adjustable only with a
set of external resistors. In a high volume production envi­ronment (such as with personal computers), however, a CPU
voltage change requires a CPU board re-design to accommo­date the new voltage requirement. The integrated 4-bit DAC
in the RC5040 and the RC5042 reads the voltage ID code
from the Pentium Pro microprocessor and configures the sys­tem to provide the appropriate voltage. In this manner, the
PC board does not have to be re-designed each time the CPU
voltage changes. The CPU can thus automatically configure
its own required voltage.

Pentium Pro and OverDrive®
Processor Power Requirements

Use Intel’s AP-523 Application Note, Pentium® Pro
Processor Power Distribution Guidelines, November 1995

(order number 242764-001), as a basic reference. The speci­fications contained in this document have been modified
slightly from the original Intel document to include updated
specifications for Pentium Pro microprocessors. Please con­tact Intel Corporation for specific details.

Input V oltages

Available inputs are +5V ±5% and +12V ±5%. Raytheon
Electronics’ DC-DC converters may use either or both
inputs. Their input voltage requirements are listed in Table 1.

Table 1. Input Voltage Requirements

Controller

Part #

RC5040
RC5042

RC5043 +5V ±5% 12V ±5% 12V ±5%

V

CC

+5V ±5% +5V ±5% +5V ±5% or

Pentium Pro DC Power Requirements

Refer to Table 2 for the power supply specifications for
Pentium Pro and Overdrive Processors. For a motherboard
design without a standard Voltage Regulator Module (VRM)
socket, the on-board DC-DC converter must supply a mini­mum ICCP current of 13.9A at 2.5V and 12.4A at 3.3V. For a
flexible motherboard design, the on-board converter must be
able to supply 14.5A maximum ICCP.

DC V oltage Regulation

As indicated in Table 2, the voltage level supplied to the
CPU must be within ±5% of its nominal setting. Voltage
regulation limits must include:

• Output load ranges specified in Table 2

• Output ripple/noise

• DC output initial voltage set point

• Temperature and w arm up drift (Ambient +10°C to +60°C
at full load with a maximum rate of change of 5°C per 10
minutes minimum but no more than 10°C per hour)

• Output load transient with:
Slew rate >30A/µs at the converter pins
Range: 0.3A – ICCP Max (as defined in Table 2).

MOSFET

Drain

MOSFET

Gate Bias

12V ±5%

Rev. 1.1.0

AN42 APPLICATION NOTE

Table 2. Intel Pentium Pro and OverDrive Processor Power Specifications

Voltage

Specification

CPU Model & Features

150MHz – 256K L2 Cache 3.1 ± 5% 9.9 29.2
166MHz – 512K L2 Cache 3.3 ± 5% 11.2 35.0
180MHz – 256K L2 Cache 3.3 ± 5% 10.1 31.7
200MHz – 256K L2 Cache 3.3 ± 5% 11.2 35.0
200MHz – 512K L2 Cache 3.3 ± 5% 12.4 37.9

OverDrive Processors

150 MHz 2.5 ± 5% 11.2 26.7
180 MHz 12.5 29.7
200 MHz 13.9 32.9

Flexible Motherboard

Notes:

1. Maximum power values are measured at typical V

2. Flexible motherboard specifications are recommendations only. Actual specifications are subject to change.

2

VCCP (VDC)

2.4-3.5 ± 5% 14.5 45.0

P to take into account the thermal time constant of the CPU package.

CC

Maximum

Current

ICCP (A)

Maximum Thermal

Design Power

(W)

1

Output Ripple and Noise

Ripple and noise are defined as periodic or random signals
over the frequency band of 20MHz at the output pins. Output
ripple and noise requirements of ±1.0% must be met
throughout the full load range and under all specified input
voltage conditions.

Efficiency

The efficiency of the DC-DC converter must be greater
than 80% at high current draw and greater than 40% at low
current draw.

Processor Voltage Identification

The Pentium Pro package has four voltage identification
pins, VID3–VID0, that can be used for automatic selection
of the power supply voltage. These pins are internally uncon­nected or are shorted to ground (VSS). The logic status of the
pins defines the voltage required by the processor. The VID
codes have been implemented to support voltage specifica­tion variations on future Pentium Pro processors. These
codes are presented in Table 3. A ‘1’ refers to an open pin
and a ‘0’ refers to a short to ground. The VCCP power supply
should supply the voltage that is requested or disable itself.

Table 3. Voltage Identification Codes for Pentium Pro

Data Bits V

VID3 VID2 VID1 VID0 (VDC)

1 1 1 1 No CPU
1 1 1 0 2.1
1 1 0 1 2.2
1 1 0 0 2.3
1 0 1 1 2.4
1 0 1 0 2.5
1 0 0 1 2.6
1 0 0 0 2.7
0 1 1 1 2.8
0 1 1 0 2.9
0 1 0 1 3.0
0 1 0 0 3.1
0 0 1 1 3.2
0 0 1 0 3.3
0 0 0 1 3.4
0 0 0 0 3.5

CC

P

I/O Controls

In addition to the voltage identification pins, several signals
exist to control the DC-DC converter or to provide feedback
from the converter to the CPU. These are Power-Good
(PWRGD), Output Enable (OUTEN), and Upgrade Present
(UP). These signals are discussed later.

2

APPLICATION NOTE AN42

I

L

VINV

OUT

–( )T

ON

L1

-----------------------------------------------=

V

OUT

V

IN

T

ON

T

S

-----------

 

 

=

RC5040 and RC5042 Description

Simple Step-Down Converter

S1

V

IN

Figure 1. Simple Buck DC-DC Converter

Figure 1 illustrates a step-down DC-DC converter with no
feedback control. The basic step-down converter serves as
the basis for deriving the design equations for the RC5040
and RC5042. From Figure 1, the basic operation begins by
closing the switch S1, so that the input voltage VIN is
impressed across inductor L1. The current flowing through
this inductor is given by the following equation:

where T

is the duty cycle (the time when S1 is closed).

ON

When S1 opens, the diode D1 conducts the inductor
current and the output current is delivered to the load accord­ing to the following equation:

V

OUTTSTON

--------------------------------------------=

I

L

–( )

L1

where TS is the overall switching period and (TS – TON) is
the time during which S1 is open.

By solving these equations you can obtain the basic relation­ship for the output voltage of a step-down converter:

D1

L1

+

C1 RL Vout

–

65-AP42-01

The RC5040 and RC5042 Controllers

The RC5040 is a programmable synchronous-mode DC-DC
converter controller. The RC5042 is a non-synchronous ver­sion of the RC5040. When designed with the appropriate
external components, either device can be configured to
deliver more than 14.5A of output current. During heavy
loading conditions, these controllers function as current­mode PWM step-down regulators. Under light loads, they
function in PFM (pulse frequency modulation) or pulse skip­ping mode. The controllers sense the load level and switch
between the two operating modes automatically, thus opti­mizing efficiency under all loads. The key differences
between the RC5040 and RC5042 are listed in Table 4.

Table 4. RC5040 and RC5042 Differences

RC5040 RC5042

Operation Synchronous Non-Synchronous
Package 20-pin SOIC 16-pin SOIC
Output Enable/

Disable

Refer to the RC5040 Block Diagram illustrated in Figure 2.
The control loop of the regulator contains two main sections:
the analog control block and the digital control block. The
analog block consists of signal conditioning amplifiers feed­ing into a set of comparators which provide the inputs to the
digital block. The signal conditioning section accepts inputs
from the IFB (current feedback) and VFB (voltage feedback)
pins and sets two controlling signal paths. The voltage con­trol path amplifies the VFB signal and presents the output to
one of the summing amplifier inputs. The current control
path takes the difference between the IFB and VFB and pre­sents the result to another input of the summing amplifier.
These two signals are then summed together with the slope
compensation input from the oscillator. This output is then
presented to a comparator, which provides the main PWM
control signal to the digital control block.

Yes No

In order to obtain a more accurate approximation for V
we must also include the forward voltage VD across diode
D1 and the switching loss, VSW. After taking into account
these factors, the new relationship becomes:

T

ON

-----------

V

OUT

Where VSW = IL • R

VINVDVSW–+( )

.

DS,ON

V

–=

T

D

S

OUT

The additional comparators in the analog control section sets
the threshold for when the RC5040 enters PFM mode during

,

light loads and the point when the current limit comparator
disables the output drive signals to the MOSFETs.

The digital control block is designed to take the comparator
inputs along with the main clock signal from the oscillator
and provide the appropriate pulses to the HIDRV and
LODRV pins that control the external power MOSFETs. The
digital section was designed utilizing high speed Schottky
transistor logic, thus allowing the RC5040 to operate at clock
speeds as high as 1MHz.

3

AN42 APPLICATION NOTE

Main Control Loop

OSCILLATOR

–
+

VREF

4-BIT

DAC

VID0
VID1
VID2
VID3

1.24V

REFERENCE

Figure 2. RC5040 Block Diagram

High Current Output Drivers

The RC5040 contains two identical high current output
drivers that use high speed bipolar transistors in a push-pull
configuration. Each driver is capable of deliv ering 1A of cur­rent in less than 100ns. Each driver’s power and ground are
separated from the chip power and ground for additional
switching noise immunity. The HIDRV driver’s power sup­ply, VCCQP, is boot-strapped from a flying capacitor as
illustrated in Figure 3. Using this configuration, C12 is
charged from VCC via the Schottky diode DS2 and boosted
when the FET is turned on. This scheme provides a VCCQP
voltage equal to 2•VCC – VDS(DS2), or approximately 9.5V
when VCC = 5V. This voltage is sufficient to provide the 9V
gate drive to the MOSFET that is required to achieve a low

DS(ON). Since the low side synchronous FET is referenced to

R
ground (see Figure 4), boosting the gate drive voltage is not
needed and the VCCP power pin can be tied to VCC.
Refer to Typical Operating Characteristics of the RC5040
data sheet for a full load VCCQP waveform.

Internal Voltage Reference

The reference used in the RC5040 is a precision band-gap
voltage reference, with internal resistors precisely trimmed
to provide a near zero temperature coefficient, TC. Added to
the reference voltage is the output from a 4-bit DAC. The
DAC is provided meet Pentium Pro specifications, requiring
a programmable converter output via a 4-bit voltage identifi­cation (VID) code. This code scales the output voltage from

2.0V (no CPU) to 3.5V in 100mV increments. To guarantee
stable operation under all loads, a 10KΩ pull-up resistor and

0.1µF of decoupling capacitance should be connected to the
VREF pin. No load should be imposed on this pin.

RC5040

–
+

–
+

–
+

DIGITAL

CONTROL

POWER

GOOD

PWRGD

+5V
VIN

VO

65-5040-01

Power Good (PWRGD)

The RC5040 and RC5042 Power Good function has been
designed according to Intel’s Pentium Pro DC-DC converter
specification. The Power Good function provides a constant
voltage monitor on the VFB pin. The internal circuitry of the
converter compares the VFB signal to the VREF voltage and
outputs an active-low interrupt signal to the CPU when the
power supply voltage exceeds ±7% of its nominal setpoint.
The Power Good flag provides no other control function to
the RC5040.

Output Enable (OUTEN)

Intel specifications state that the DC-DC converter should
accept an open collector signal for controlling the output
voltage. A logic LOW for this signal disables the output volt­age. When disabled, the PWRGD output is in the low state.
This feature is available for the RC5040 only.

Upgrade Present (UP#)

Intel specifications state that the DC-DC converter must
accept an open collector signal that indicates the presence of
an upgrade processor. The typical state is high (for a stan­dard P6 processor). When the signal is low or in theground
state (for the OverDrive processor), the output voltage must
be disabled unless the converter can supply the OverDrive
processor’s power requirements. When disabled, the
PWRGD output must be in the low state. Because the
RC5040 and RC5042 can supply the OverDrive processor
requirements, the UP# signal is not required.

4

APPLICATION NOTE AN42

Over-Voltage Protection

The RC5040 and RC5042 constantly monitor the output
voltage for protection against over voltage. If the voltage at
the VFB pin exceeds 20% of the selected program voltage,
an over-voltage condition is assumed, and the controller dis­ables the output drive signal to the external MOSFET(s).

Short Circuit Protection

A current sense methodology is implemented to disable the
output drive signal to the MOSFET(s) when an over-current
condition is detected. The voltage drop created by the output
current flowing across a sense resistor is presented to an
internal comparator. When the voltage developed across the
sense resistor exceeds the comparator threshold voltage,
the controller disables the output drive signal to the
MOSFET(s).

The DC-DC converter returns to normal operation after the
fault has been removed, for either an over voltage or a short
circuit condition.

Oscillator

The RC5040 oscillator section is implemented using a
fixed current capacitor charging configuration. An external
capacitor (CEXT) is used to preset the oscillator frequency
between 200KHz and 1MHz. This allows maximum flexibil­ity in setting the switching frequency and in choosing exter­nal components.

In general, a lower operating frequency increases the peak
ripple current flowing through the output inductor, allowing
the use of a larger inductor value. Operation at lower fre­quencies increases the amount of energy storage that the
bulk output capacitors must provide during load transients
that occur due to the slower loop response of the controller.

In addition, note that the efficiency losses due to switching
are relatively fixed per switching cycle. Therefore, as the
switching frequency increases, the contribution toward effi­ciency due to switching losses also increases.

RC5040 has an optimal operating frequency of 650KHz.
This frequency allows the use of smaller inductive and
capacitive components while optimizing peak efficiency
under all operating conditions.

Design Considerations and
Component Selection

Application Circuits

Figure 3 illustrates a typical non-synchronous application
using the RC5040. Figure 4 shows a typical synchronous
application using the RC5040, and Figure 5 shows a typical
non-synchronous application using the RC5042.

VREF

GND

VCC

C4

0.1µF

R7

10K

C7

0.1µF

L2

2.6µH

VID3
VID2
VID1
VID0

OUTEN

C1

1000µF

C2

1000µF

12
13
14
15
16
17
18
19
20

R1
R2
R3
R4

C3

1000µF

RC5040

C

39pF

10K
10K
10K
10K

EXT

C5

0.1µF

1011

9
8
7
6
5
4
3
2
1

C10

0.1µF

R5

10K

C6

4.7µF

VCC

DS2

1N5817

C12
1µF

2SK1388

R6

10K

M1

C11

0.22µF

C8

0.1µF

DS1

MBR1545CT

VCC

PWRGD

C9

0.1µF

M2
2SK1388

L1

1.3µH

R

SENSE

8mΩ

F

µ

1500

C13

65-AP42-03

1500µF

C14

F

µ

1500

C15

VO

Figure 3. Non-Synchronous DC-DC Converter Application Schematic Using RC5040

5

AN42 APPLICATION NOTE

VREF

GND

VCC

C4

0.1µF

R7

10K

C7

0.1µF

VID3
VID2
VID1
VID0

OUTEN

L2

2.6µH

C1

1000µF

12
13
14
15
16
17
18
19
20

R1
R2
R3

R4

C2

1000µF

RC5040

C
39pF

10K
10K
10K
10K

EXT

C3

1000µF

1011

9
8
7
6
5
4
3
2
1

C10

0.1µF

C5

0.1µF

R5

10K

C6

4.7µF

VCC

DS2

1N5817

C12

1µF

2SK1388

2SK1388

M3

R6

10K

M1

0.22µF

C11

C8

0.1µF

1N5817

VCC

PWRGD

DS1

C9

0.1µF

M2
2SK1388

L1

1.3µH

R

SENSE

8mΩ

65-AP42-04

1500µF

C13

1500µF

C14

VO

1500µF

C15

Figure 4. Synchronous DC-DC Converter Application Schematic Using RC5040

VCC

0.1µF

VREF

GND

L2

2.6µH

C4

C1

1000µF

R7
10K

C7

0.1µF

VID3

VID2
VID1
VID0

9
10
11
12
13
14
15
16

R1

R2
R3
R4

RC5042

C3

C2
1000µF 1000µF

8
7

4

2
1

C

EXT

39pF

10K
10K
10K
10K

C10

C5

0.1µF
DS2

1N5817

C12

6
5

C6

4.7µF

1µF

M1

2SK1388

3

R6

10K

C8

0.1µF

DS1

MBR1545CT

VCC

C9

0.1µF

M2
2SK1388

L1

1.3µH

R

SENSE

8mΩ

F

µ

1500

C13

65-AP42-05

1500µF

C14

VO

1500µF

C15

PWRGD

VCC

C11

0.22µF

0.1µF

Figure 5. Non-Synchronous DC-DC Converter Application Schematic Using RC5042

6

APPLICATION NOTE AN42

MOSFET Selection

This application requires the use of N-channel, Logic Level
Enhancement Mode Field Effect Transistors. The desired
characteristics of these components are:

• Low Static Drain-Source On-Resistance
R

• Low gate drive voltage, VGS ≤ 4.5V

Table 5. MOSFET Selection Table

Manufacturer & Model # Conditions

Fuji
2SK1388

Siliconix
SI4410DY

National Semiconductor
NDP706AL

< 37 mΩ (lower is better)

DS,ON

V

= 4V

GS

ID = 17.5A
V

= 4.5V

GS

ID = 5A
V

= 5V

GS

ID = 40A

• Power package with low thermal resistance

• Drain current rating of 20A minimum

• Drain-Source voltage > 15V.

The on-resistance (R

) is the main parameter for MOS-

DS,ON

FET selection. It determines the MOSFET’s power dissipa­tion, thus significantly affecting the efficiency of the
converter. Several suitable MOSFETs are shown in Table 5.

R

1

DS, ON

(mΩ)

Package

TJ = 25°C 25 37 TO-220 Φ
TJ = 125°C 37 —
TJ = 25°C 16.5 20 SO-8
TJ = 125°C 28 34

(SMD)

TJ = 25°C 13 15 TO-220 Φ

NDP706AEL TJ = 125°C 20 24
National Semiconductor V
NDP603AL TJ = 125°C 42 54 Φ
National Semiconductor V
NDP606AL TJ = 125°C 33 40 Φ
Motorola V
MTB75N03HDL TJ = 125°C 9.3 14 (D2 PAK) Φ
Int. Rectifier V
IRLZ44 TJ = 125°C — 46 Φ
Int. Rectifier V
IRL3103S TJ = 125°C 31 Φ

Note:

1. R

turers and are approximations only.

) values at Tj = 125°C for most devices were extrapolated from the typical operating curves supplied by the manufac-

DS(ON

= 4.5V

GS

ID = 10A

= 5V

GS

ID = 24A

= 5V

GS

ID = 37.5A

= 5V

GS

ID = 31A

= 4.5V

GS

ID = 28A

TJ = 25°C 31 40 TO-220 Φ

TJ = 25°C 22 25 TO-220 Φ

TJ = 25°C 6 9 TO-263 Φ

TJ = 25°C — 28 TO-220 Φ

TJ = 25°C — 19 TO-220 Φ

Thermal

ResistanceTyp. Max.

= 75

JA

Φ

= 50

JA

= 62.5

JA

Φ

= 1.5

JC

= 62.5

JA

= 2.5

JC

= 62.5

JA

= 1.5

JC

= 62.5

JA

= 1.0

JC

= 62.5

JA

= 1.0

JC

= 62.5

JA

= 1.0

JC

Two MOSFETs in Parallel

We recommend two MOSFETs used in parallel instead of a
single MOSFET. The following significant advantages are
realized using two MOSFETs in parallel:

• Significant reduction of power dissipation.
Maximum current of 14A with one MOSFET:
P

MOSFET

(14)2(0.050*)(3.3+0.4)/(5+0.4-0.35) = 7.2 W
With two MOSFETs in parallel:

P

MOSFET

(14/2)

* Note: R

= (I2 R

= (I2 R

2

(0.037*)(3.3+0.4)/(5+0.4-0.35) = 1.3W/FET

increases with temperature. Assume R

DS,ON

at 25°C. R
when using a single MOSFET. When using two MOSFETs in
parallel, the temperature effects should not cause the R
above the listed maximum value of 37mΩ.

DS,ON

)(Duty Cycle) =

DS,ON

)(Duty Cycle) =

DS,ON

can easily increase to 50mΩ at high temperature

DS,ON

DS,ON

= 25mΩ

to rise

• No added heat sink required.

With the power dissipation down to around one watt and
with MOSFETs mounted flat on the motherboard, no
external heat sink is required. The junction-to-case
thermal resistance for the MOSFET package (TO-220) is
typically at 2°C/W and the motherboard serves as an
excellent heat sink.

• Higher current capability.

With thermal management under control, this on-board
DC-DC converter can deliver load currents up to 14.5A
with no performance or reliability concerns.

7

AN42 APPLICATION NOTE

MOSFET Gate Bias

The MOSFET(s) can be biased using one of two methods:
Charge Pump or 12V Gate Bias.

Charge Pump (or Bootstrap)

Figure 6 employs a charge pump to provide the MOSFET
gate bias. The charge pump capacitor, CP, is used as a flying
capacitor to boost the voltage of the RC5040 or RC5042 out­put driver . When the MOSFET switches off, the source of the
MOSFET is at -0.6V. VCCQP is charged through the Schot­tky diode to 4.5V. Thus, the capacitor CP is charged to 5V.
When the MOSFET turns on, the source of the MOSFET is
at approximately 5V. The capacitor voltage follows, and
hence provides a voltage at VCCQP equal to 10V. The Schot­tky is required to provide the charge path when the MOSFET
is off, and then reverses bias when the VCCQP goes to 10V.
The capacitor CP needs to be a high Q and high frequency
capacitor. A 1µF ceramic capacitor is recommended here.

+5V

DS2

VCCQP

HIDRV

PWM/PFM

Control

CP

M1

DS1

L1

RS

CB

VO

12V Gate Bias

Figure 7 illustrates how an external 12V source can be used
to bias VCCQP. A 47 Ω resistor is used to limit the transient
current into the VCCQP pin, and a 1µF capacitor filter is
used to filter the VCCQP supply. This method provides a
higher gate bias voltage (VGS) to the MOSFET, and there­fore reduces the R
MOSFET . Figure 8 illustrates how R

and resulting power loss within the

DS,ON

decreases dra-

DS,ON

matically as VGS increases. A 6.2V Zener (DS2) is used to
clamp the voltage at V

to a maximum of 12V and

CCQP

ensure that the absolute maximum voltage of the IC is not
exceeded.

Warning: The 12V Gate Bias method applies only to the
RC5042. The RC5040 has not been designed to accept an
external 12V gate bias voltage, and may be damaged if
this method is used.

+5V

VCCQP

HIDRV

47Ω

D1

6.2V

M1

DS1

L1

RS

VO

CB

PWM/PFM

Control

+12V

Figure 6. Charge Pump Configuration

0.1

0.09

0.08

0.07

(Ω)

0.06

0.05

0.04

DS,ON

R

0.03

0.02

0.01
0

1.5 2 2.5 3 3.5 4 5 6 7 8 9 10 11

Figure 8. R

65-AP42-06

Gate-Source Voltage, V

vs. VGS for Selected MOSFETs

DS,ON

GS

65-AP42-07

Figure 7. 12V Gate Bias Configuration

R(DS)Fuji
R(DS)Fuji
R(DS)706A
R(DS)-706AEL

(V)

8

APPLICATION NOTE AN42

(7) PD

CAPIRMS

2

ESR×=

(8) PD

IC

V

CCICC

×=

Converter Efficiency

Losses due to parasitic resistance in the switches, inductor,
and sense resistor dominate at high load-current levels. The
major loss mechanisms under heavy loads, in order of
importance, are:

• MOSFET I2R Losses

• Inductor Losses

Efficiency of the converter under heavy loads can be calculated as follows:

• Sense Resistor Losses

• Gate-Charge Losses

• Diode-Conduction Losses

• Transition Losses

• Input Capacitor Losses

• Losses Due to the Operating Supply Current of the IC.

Efficiency

P

where

LOSS

-------------­p

Design Equations:

P

(1) PD

(2) PD

(3) PD

(4) PD

(5) PD

(6) PD

MOSFETIOUT

INDUCTORIOUT

RSENSEIOUT

GATE

DIODE

TRAN

OUT

IN

PD

q

GATE

VfID× 1 DutyCycle–( )=

V

-------------------------------------------------------------=

IN

---------------------------------------------------------= =
I

OUTVOUTPLOSS

PD

MOSFET

+ + + + + + +=

2

R

× DutyCycle×= DutyCycle

DS,ON

2

R

×=

2

R

×=

SENSE

f 5V××=

+×

INDUCTOR

INDUCTOR

, where q

,

PD

RSENSE

, where

is the gate charge and f is the switching frequency

GATE

×

I

OUTVOUT

2

C

× I

RSS

I

DRIVE

× f×

LOAD

, where C

RSS

PD

GATE

PD

DIODE

+

V

OUTVD

------------------------------------------=
V

INVDVSW

–+

PD

TRAN

PD

CAP

PD

IC

is the reverse transfer capacitance of the high-side MOSFET.

Example:

DutyCycle

PD

PD

PD

PD

PD

PD

PD

PD

MOSFET

INDUCTOR

RSENSE

GATE

DIODE

TRAN

CAP

0.2W=

IC

3.3 0.5+

------------------------------ 0.73= =
5 0.5 0.3–+

1020.030× 0.73× 2.19W= =

1020.010× 1W= =

1020.0065× 0.65W= =

CV f× 5V× 1.75nf 9 1–( )V 650Khz 5V××× 0.045W= = =

0.5 10 1 0.73–( )× 1.35W= =

52400pf× 10× 650khz×

----------------------------------------------------------------

0.7A

0.010W∼=

7.5 2.5–( )20.015× 0.37W= =

9

AN42 APPLICATION NOTE

PD

LOSS

Efficiency

2.19W 1.0W 0.65W 0.045W 1.35W 0.010W 0.37W 0.2W+ + + + + + + 5.815W= =

3.3 10×

---------------------------------------

3.3 10 5.815+×

85%≈=∴

Selecting the Inductor

Selecting the right inductor component is critical in the
DC-DC converter application. The inductor’s critical param­eters to consider are inductance (L), maximum DC current
(IO), and coil resistance (Rl).

The inductor core material is crucial in determining the
amount of current it can withstand. As with all engineering
designs, tradeoffs exist between various types of core mate­rials. In general, Ferrites are popular due to their low cost,
low EMI properties, and high frequency (>500KHz) charac­teristics. Molypermalloy powder (MPP) materials exhibit
good saturation characteristics, low EMI, and low hysteresis
losses; however, they tend to be expensive and more effec­tively utilized at operating frequencies below 400KHz.

Another critical parameter is the DC winding resistance of
the inductor. This value should typically be as low as possi­ble because the power loss in DC resistance degrades the
efficiency of the converter by P

LOSS

of the inductor is a function of the oscillator duty cycle
(TON) and the maximum inductor current (IPK). IPK can be
calculated from the relationship:

VINVSW– VD–

 

I

PKIMIN

------------------------------------------

+=

 

L

Where TON is the maximum duty cycle and VD is the
forward voltage of diode DS1.

2

= I

x Rl. The value

O

T

ON

Table 6. RC5040 and RC5042 Short Circuit Comparator
Threshold Voltage

Short Circuit Comparator

V

threshold

(mV)

Typical 120

Minimum 100

Maximum 140

When designing the external current sense circuitry, pay
careful attention to the output limitations during normal
operation and during a fault condition. If the short circuit
protection threshold current is set too low , the con v erter may
not be able to continuously deliver the maximum CPU load
current. If the threshold level is too high, the output driver
may not be disabled at a safe limit and the resulting power
dissipation within the MOSFET(s) may rise to destructive
levels.

The design equation used to set the short circuit threshold
limit is as follows:

V

th

--------

R

SENSE

ISCI

≥ I

inductor

where I

pk

I

load, max

I

and I

, where: I

SC

Load, max

are peak ripple currents and

min

SC

= output short circuit current=

----------------------------+=

IpkI

–( )

2

min

is the maximum output load current.

The inductor value can be calculated using the following
relationship:

=

 

I

PKIMIN

Where VSW (R

–

DS,ON

T

ON

x IO) is the drain-to-source voltage of

– VO–

V

INVSW

 

------------------------------------------

L

M1 when it is turned on.

Implementing Short Circuit Protection

Intel currently requires all power supply manufacturers
to provide continuous protection against short circuit
conditions that may damage the CPU. To address this
requirement, Raytheon Electronics has implemented a cur­rent sense methodology on the RC5040 and RC5042 con­trollers. This methodology limits the power delivered to the
load during an overcurrent condition. The voltage drop cre­ated by the output current flowing across a sense resistor is
presented to one terminal of an internal comparator with
hysterisis. The other comparator terminal has a threshold
voltage, nominally 120mV. Table 6 states the limits for the
comparator threshold of the switching regulator:

10

You must also take into account the current (Ipk –I

min

), or
the ripple current flowing through the inductor under normal
operation. Figure 9 illustrates the inductor current waveform
for the RC5040 and RC5042 DC-DC converters at maxi­mum load.

I

pk

I

(I

– I

pk

)/2

min

I

min

T

ON

T = 1/f

Figure 9. Typical DC-DC Converter

Inductor Current Waveform

T

OFF

s

I

LOAD, MAX

t

The calculation of this ripple current is as follows:

V

– V

IpkI

–( )

min

---------------------------­2

INVSW

-------------------------------------------------

–( )

L

OUT

V

+( )

OUTVD

--------------------------------------------­– VD+( )

V

INVSW

T×=

APPLICATION NOTE AN42

IPKI

where:

• V

• V

= Input Voltage to the Converter

IN

= Voltage Across the MOSFET = I

SW

LOAD

x R

DS,ON

• VD = Forward Voltage of the Schottky Diode

• T = The Switching Period of the Converter = 1/fS,
Where fS = Switching Frequency.

For an input voltage of 5V, an output voltage of 3.3V, an
inductor value of 1.3µH, and a switching frequency of
650KHz (using C

= 39pF), the inductor current can be

EXT

calculated as follows:

–( )

I

pkImin

---------------------------­2

--------------------------------------------------------------

3.3 0.5+( )

5.0 14.5 0.037×– 0.5+( )

5.0 14.5 0.037×– 3.3–( )

--------------------------------------------------------------

1.3 10

6–

×

1

-----------------------

× 1.048A=

650 103×

×=

ISCI

≥ I

inductor

Load, max

For continuous operation at 14.5A, the short circuit detection
threshold must be at least 15.5A.

The next step is to determine the value of the sense resistor.
Including tolerance, the sense resistor value can be approxi­mated as follows:

R

SENSE

----------------

1 TF–( )×

I

SC

V

th,min

where TF = Tolerance Factor for the sense resistor.

Several different types of sense resistors exist. Table 7
describes tolerance, size, power capability, temperature
coefficient and cost of various sense resistors.

–( )

min

-----------------------------+ 14.5 1+ 15.5A= = =
2

V

th,min

-----------------------------------

1.0 I

+

Load,max

1 TF–( )×= =

Therefore, for a continued load current of 14.5A, the peak
current through the inductor, I

Table 7. Comparison of Sense Resistors

Motherboard

Description

Trace Resistor

Tolerance
Factor (TF)

Size
(L x W x H)

2" x 0.2" x 0.001"

(1 oz Cu trace)

Power capability > 50A/in 1 watt

, is found to be:

pk

Discrete Iron

Alloy

Resistor (IRC)

±29% ±5%

(±1% available)

0.45" x 0.065" x

0.200"

Discrete Metal

Strip Surface

Mount Resistor

(Dale)

Discrete MnCu

Alloy Wire

Resistor

±1% ±10% ±10%

0.25" x 0.125" x

0.025"

0.200" x 0.04" x

0.160"

1 watt 1 watt 1 watt

Discrete

CuNi Alloy

Wire Resistor

(Copel)

0.200" x 0.04" x

0.100"

(3W and 5W

available)

Temperature

+4,000 ppm +30 ppm ±75 ppm ±30 ppm ±20 ppm

Coefficient
Cost

@10,000 piece

Low

included in

$0.31 $0.47 $0.09 $0.09

motherboard

Refer to Appendix A for Directory of component suppliers

Based on the Tolerance in the above table, for embedded PC
trace resistor and for I

R

SENSE

100mV

---------------------------------

1.0A 14.5A+

-----------------------------------------

1.0A I

load,max

V

th,min

+

Load, max

1 29%–( )× 4.6mΩ=

For a discrete resistor and I

= 14.5A:

1 TF–( )× = =

load, max

= 14.5A:

R

SENSE

100mV

---------------------------------

1.0A 14.5A+

-----------------------------------------

1.0A I

+

Load, max

1 5%–( )× 6.1mΩ=

1 TF–( )× = =

V

th,min

For user convenience, Table 8 lists the recommended values
for sense resistor at various load currents using an embedded
PC trace resistor or discrete resistor.

11

AN42 APPLICATION NOTE

i

Table 8. R

I

Load,max

(A)

for various load currents

sense

R

SENSE

PC Trace

Resistor (mΩ)

R

SENSE

Discrete

Resistor (mΩ)

10.0 6.5 8.6

11.2 5.8 7.8

12.4 5.3 7.1

13.9 4.8 6.4

14.0 4.7 6.3

14.5 4.6 6.1

Discrete Sense Resistor

Discrete iron alloy resistors come in a variety of tolerances
and power ratings, and are ideal for precision implementa­tions. Either an MnCu alloy wire resistor or an CuNi alloy
wire resistor is ideal for a low cost implementation.

Embedded Sense Resistor (PC Trace Resistor)

Embedded PC trace resistors have the advantage of almost
zero cost implementation. However, the value of the PC
trace resistors have large variations. Embedded resistors
have 3 major error sources: the sheet resistivity of the inner
layer, the mismatch due to L/W, and the temperature varia­tion of the resistor. When laying out embedded sense resis­tors, consider all error sources described as follows:

• Sheet resistivity.

For 1 ounce copper, the thickness variation is typically
between 1.15 mil and 1.35 mil. Therefore, the error due to
sheet resistivity is (1.35 – 1.15)/1.25 = 16%.

• Mismatch due to L/W.

The error in L/W is dictated by the geometry and the
power dissipation capability of the sense resistor. The
sense resistor must be able to handle the load current and,
therefore, requires a minimum width, calculated as
follows:

I

L

----------=

W

0.05

where W is the minimum width required for proper po wer
dissipation (mils), and IL is the load current in Amps.

For a load current of 15A, the minimum width required is
300mils, which reflects a 1% L/W error.

• Thermal Considerations.

The I2R power losses cause the surface temperature of the
resistor to increase along with its resistance value. In
addition, ambient temperature variations add the change
in resistor value:

R R

where R20 is the resistance at 20°C, α

1 α20T 20–( )]+[=

20

= 0.00393/ °C,T

20

is the operating temperature, andR is the desired value.
For temperature T = 50°C, the %R change = 12%.

Table 9 is a summary of tolerances for the Embedded PC
Trace Resistor.

Table 9. Summary PC Trace Resistor Tolerance

Tolerance due to sheet resistivity variation 16%
Tolerance due to L/W error 1%
Tolerance due to temperature variation 12%
Total Tolerance for PC Trace Resistor 29%

Design rules for using an embedded resistor

The basic equation for laying an embedded resistor is:

L

t

R ρ

×=

L

------------­W t×

W

where ρ is the Resistivity (W-mil), L is the Length (mils), W
is the Width (mils), and t is the Thickness (mils).

For 1oz copper, t = 1.35 mils, ρ = 717.86 µΩ-mil,
1 L/1 W = 1 Square ( ■ ).

For example, you can layout a 5.30mΩ embedded sense
resistor. From Equations above,

I

----------

W

0.05

R W× t×

L

-----------------------

= = =

10

L

ρ

---------- 200mil= = =

0.05

0.00530 200× 1.35×

--------------------------------------------------- 2000m

717.86

L/W = 10 ■.
Therefore, to model 5.30mΩ enbedded resistor, you need

W = 200 mils, and L = 2000 mils. See Figure 10.

1 1 1 1 1 1 1 1 1 1

W = 200 mils

L = 2000

Figure 10. 5.30mΩ Sense Resistor (10 ■)

You can also implement the sense resistor in the following
manner. Each corner square is counted as 0.6 square since
the current flowing through the corner square does not flow
uniformly, concentrated towards the inside edge. This is
shown in Figure 11.

1 1 1 1 1 1

.6 .6

1 1

.8

Figure 11. 5.30mΩ Sense Resistor (10 ■)

A Resign Example Combining an Embedded Resistor
with a Discrete Resistor

For low cost implementation, the embedded PC trace resistor
is the most desirable alternative, but, as discussed earlier, the
wide tolerance (±29%) presents a challenge. In addition,
changing CPU requirements may force the maximum load

12

APPLICATION NOTE AN42

Embedded Sense Resistor

IFBH

MnCu Discrete

IFBL

Resistor

Figure 12. Short Circuit Sense Resistor Design Using PC Trace Resistor and Optional Discrete Sense Resistor

R21 R22

Output Power
Plane (Vout)

R-∆r

R

R+∆r

currents to change. Therefore, combining an embedded
resistor with a discrete resistor may be a desirable option.
This section discusses a design that provides flexibility and
addresses wide tolerances. Refer to Figure 12.

In this design, the user has the option to choose either an
embedded or a discrete MnCu sense resistor. To use the dis­crete sense resistor, populate R21 with a shorting bar (zero
Ohm resistor) for a proper Kelvin connection and add the
MnCu sense resistor. To use the embedded sense resistor,
populate R22 with a shorting bar for a Kelvin connection.
The embedded sense resistor allows you to choose a plus or a
minus delta resistance tap to offset any large sheet resisti vity
change.

In this design, the center tap yields 6mΩ, and the left or the
right tap yield 6.7 or 5.3 mΩ, respectively.

RC5040 and RC5042 Short Circuit Current
Characteristics

The RC5040 and RC5042 have a short circuit current char­acteristic that includes a hysteresis function. This function
prevents the DC-DC converter from oscillating in the event
of a short circuit. Figure 13 shows the typical characteristic
of the DC-DC converter using a 6.5 mΩ sense resistor.

3.5

3.0

2.5

2.0

1.5

1.0

Output Voltage

0.5
0

0 5 10 15 20 25

Output Current

The converter exhibits at normal load regulation until the
voltage across the resistor reaches the internal short circuit
threshold of 120mV. At this point, the internal
comparator trips and signals the controller to turn off the
gate drive to the power MOSFET. This causes a drastic
reduction in the output voltage as the load regulation col­lapses into the short circuit control mode. The output voltage
does not return to its nominal value until the output short cir­cuit current is reduced to within the safe range for the DC­DC converter.

Power Dissipation Consideration During a
Short Circuit Condition

The RC5040 and RC5042 controllers respond to an output
short circuit by drastically changing the duty cycle of the
gate drive signal to the power MOSFET. In doing this, the
power MOSFET is protected from over-stress and eventual
destruction. Figure 14A shows the gate drive signal of a typ­ical RC5040 operating in continuous mode with a load cur­rent of 10A. The duty cycle is then set by the ratio of the
input voltage to the output voltage. If the input voltage is 5V
and the output voltage is 3.1V, the ratio of Vout/ Vin is 62%.
Figure 14B shows the result of the RC5040 going into its
short circuit mode when the duty cycle is around 20%. Cal­culating the power on the MOSFET at each condition on the
graph in Figure 13 shows how the protection scheme works.
The power dissipated in the MOSFET at normal operation
for a load current of 14.5A, is given by:

˙

2

14.5

 

PDI2RON× DutyCycle

for each MOSFET.

The power dissipated in the MOSFET at short circuit
condition for a peak short current of 20A, is given by:

2

20

 

P

------

D

 

.037× .2 × 0.74W==

2

for each MOSFET.

----------

 

2

.037×

.62 1.2W=×=×=

Figure 13. RC5040/RC5042 Short Circuit Characteristic

Thus, the MOSFET is not being over-stressed during a short
circuit condition.

13

AN42 APPLICATION NOTE

Figure 14A. V

Operation Condition with V

Output Waveform for Normal

CCQP

= 3.3V@10A

out

Figure 14B. V

Output Shorted to Ground

The Schottky diode has a power dissipation consideration
during the short circuit condition. During normal operation,
the diode dissipates power when the power MOSFET is off.
The power dissipation is given by:

P

D Diode,

IFVF× 1 DutyCycle–( )× = =

14.5 0.5V× 1 0.62–( ) 2.75W=×

In short circuit mode, the duty cycle is dramatically reduced
to approximately 20%. The forward current during a short
circuit condition decays exponentially through the inductor.
The power dissipated on the diode during the short circuit
condition, is approximated by:

-----------–

I

F ending,

I

F ave,

20A 7.9A+( ) 2⁄ 14A≈ ≈

L R⁄

Isce

× 20A e

Output Waveform for

CCQP

1

1.5us

-------------–

1.3us

× 7.9A≈==

P

D Diode,

I

VF× 1 DutyCycle–( )× = =

F ave,

14 0.45× 0.8× 5W≈

Thus for the Schottky diode, the thermal dissipation during
a short circuit is greatly magnified and requires that the
thermal dissipation of the diode be properly managed by the
appropriate choice of a heat sink. In order to protect the
Schottky from being destroyed in the event of a short, we
should limit the junction temperature to less than 130°C.
Using the equation for maximum junction temperature,
we can arrive at the thermal resistance required below:

T

P

--------------------------------=

D

–

J max( )TA

R

ΘJA

Assuming that the ambient temperature is 50°C, we get:

R

ΘJA

T

--------------------------------

–

J max( )TA

P

D

130 50–

--------------------- 16°C W⁄= = =

5

Thus we need to provide for a heat sink that will give the
Schottky diode a thermal resistance of at least 16°C/W or
lower in order to protect the device during an indefinite
short.

In summary, with proper heat sink, the Schottky diode is not
being over stressed during a short circuit condition.

Schottky Diode Selection

The application circuits of Figures 3, 4, and 5 show two
Schottky diodes, DS1 and DS2. In synchronous mode, DS1
is used in parallel with M3 to prevent the lossy diode in the
FET from turning on. In non-synchronous mode, DS1 is
used as a flyback diode to provide a constant current path for
the inductor when M1 is turned off.

The Schottky diode DS2 serves a dual purpose. As config­ured in Figures 3, 4, and 5, DS2 allows the VCCQP pin on
the RC5040 to be bootstrapped up to 9V using capacitor
C12. When the lower MOSFET M3 is turned on, one side of
capacitor C12 is connected to ground while the other side of
the capacitor is being charged up to voltage VIN – VD
through DS2. The voltage that is then applied to the gate of
the MOSFET is VCCQP – VSAT, or typically around 9V.
DS2 also provides correct sequencing of the various supply
voltages by assuring that VCCQP is not enabled before the
other supplies.

A vital selection criteria for DS1 and DS2 is that they exhibit
a very low forward voltage drop, as this parameter can
directly affect the regulator efficiency. Table 10 lists several
suitable Schottky diodes. Note that the MBR2015CTL has a
very low forward voltage drop. This diode is ideal for appli­cations where output voltages less than 2.8V are required.

14

APPLICATION NOTE AN42

ESR

DF

2

πfC

-------------=

Table 10. Schottky Diode Selection Table

Manufacturer

Model # Conditions

Philips
PBYR1035

Motorola
MBR2035CT

Motorola
MBR1545CT

Motorola
MBR2015CTL

IF = 20A; Tj=25°C

IF = 20A; Tj=125°C

IF = 20A; Tj=25°C

IF = 20A; Tj=125°C

IF = 15A; Tj=25°C

IF = 15A; Tj=125°C

IF = 20A; Tj=25°C

= 20A; Tj=150°C

I

F

Forward Voltage

V

F

< 0.84v
< 0.72v

< 0.84v
< 0.72v

< 0.84v
< 0.72v

< 0.58v
< 0.48v

Output Filter Capacitors

Output ripple performance and transient response are
functions of the filter capacitors. Since the 5V supply of a PC
motherboard may be located several inches away from the
DC-DC converter, the input capacitance can play an impor­tant role in the load transient response of the RC5040.
The higher the input capacitance, the more charge storage is
available for improving the current transfer through the
FET(s). Capacitors with low Equivalent Series Resistance
(ESR) are best for this type of application and can influence
the converter's efficiency if not chosen carefully. The input
capacitor should be placed as close to the drain of the FET as
possible to reduce the effect of ringing caused by long trace
lengths.

ESR is the resonant impedance of the capacitor, and it is dif­ficult to quantify. Since the capacitor is actually a complex
impedance device having resistance, inductance, and capaci­tance, it is natural for it to have a resonant frequency. As a
rule, the lower the ESR, the better suited the capacitor is for
use in switching power supply applications. Many manufac­turers do not supply ESR data, but a useful estimate can be
obtained using the following equation:

With this in mind, correct calculation of the output capaci­tance is crucial to the performance of the DC-DC converter.
The output capacitor determines the overall loop stability,
output voltage ripple, and load transient response. The calcu­lation is as follows:

C µF( )

--------------------------------------=
∆V I

ESR×–

O

∆T×

I

O

where ∆V is the maximum voltage deviation due to load
transients, ∆T is the reaction time of the power source, and
I

is the output load current. ∆V is the loop response time of

O

the RC5040 and RC5042, approximately 8µs.

For IO = 10A and ∆V = 165mV, the bulk capacitance
required can be approximated as follows:

C µF( )

∆T×

O

-------------------------------------­∆V I

ESR×–

O

10A 8µs×

--------------------------------------------------------- 1454µF= = =
165mV 10A 11mΩ×–

I

Input filter

The DC-DC converter design should include an input induc­tor between the system +5V supply and the converter input
as described below. This inductor will serve to isolate the
+5V supply from noise occurring in the switching portion of
the DC-DC converter and also to limit the inrush current into
the input capacitors during power up. An inductor value of
around 2.5µH is recommended, as illustrated in Figure 15.

5V Vin

0.1µF

2.5µH

1000µF, 10V
Electrolytic

65-AP42-17

Figure 15. Input Filter

Bill of Materials

where DF is the dissipation factor of the capacitor, f is the
operating frequency, and C is the capacitance in farads.

The Bill of Materials for the application circuits of Figures 2
through 4 is presented in Table 11.

Table 11. Bill of Materials for a 14.5A Pentium Pro Motherboard Application

C4, C5, C7, C8, C9,
C10

C6 Panasonic

Cext Panasonic

C12
C1, C2, C3 United Chemicon

C11 Panasonic

Panasonic
ECU-V1H104ZFX

ECSH1CY475R

ECU-V1H121JCG

LXF16VB102M

ECU-V1H224ZFX

0.1µF 50V capacitor

4.7µF 16V capacitor

39pF capacitor

1000µF 6.3V electrolytic
capacitor 10mm x 20mm

0.22µF 50V capacitor

ESR<0.047Ω

15

AN42 APPLICATION NOTE

Table 11. Bill of Materials for a 14.5A Pentium Pro Motherboard Application

C13, C14, C15 Sanyo

6MV1500GX

DS1
(note 1)

DS2 General Instruments 1N5817 Schottky Diode 1A, 20V
L1 Skynet 320-8107 1.3µH inductor
L2* Skynet

M1, M2, M3
(note 2)

Rsense COPEL

R1, R2, R3, R4, R6, R7Panasonic ERJ-6ENF10.0KV 10K 5% Resistors

U1 Raytheon

Refer to Appendix A for Directory of component suppliers.
Notes:

1. In synchronous mode using the RC5040, a 1A schottky diode (1N5817) may be substituted for the MBR1545CT.

2. MOSFET M3 is only required for the RC5040 synchronous application.

Motorola
MBR1545CT

320-6110
Fuji

2SK1388

A.W.G. #18

RC5042M or RC5040M

1500µF 6.3V electrolytic
capacitor 10mm x 20mm

Shottky Diode Vf<0.72V @ If = 15A

2.5µH inductor *Optional – will help re­duce ripple on 5v line

N-Channel Logic Level
Enhancement Mode MOSFET

6 milliohm CuNi Alloy Wire
resistor

DC-DC Converter for Pentium
Pro

R

DS(ON)

V

PCB Layout Guidelines and Considerations

PCB Layout Guidelines

• Placement of the MOSFETs relative to the RC5040 is
critical. The MOSFETs (M1 & M2), should be placed
such that the trace length of the HIDRV pin to the FET
gate is minimized. A long lead length causes high
amounts of ringing due to the inductance of the trace and
the large gate capacitance of the FET. This noise radiates
all over the board, and because it is switching at a high
voltage and frequency, it is very difficult to suppress.

Figure 16 shows an example of proper MOSFET
placement in relation to the RC5040. It also shows an
example of problematic placement for the MOSFETs.

In general, noisy switching lines should be kept away
from the quiet analog section of the RC5040. That is,
traces that connect to pins 12 and 13 (HIDRV and
VCCQP) should be kept far away from the traces that
connect to pins 1 through 5, and pin 16.

• Place the 0.1µF decoupling capacitors as close to the
RC5040 and RC5042 pins as possible. Extra lead length
negates their ability to suppress noise.

• Each VCC and GND pin should have its own via to the
appropriate plane on the board to add isolation between
pins

• The CEXT timing capacitor should be surrounded with a
ground trace. The placement of a ground or power plane
underneath the capacitor provides further noise isolation,
and helps to shield the oscillator from the noise on the
PCB. This capacitor should be placed as close to pin 1 as
possible.

• Group the MOSFETs, inductor, and Schottky diode as
close together as possible. This minimizes ringing derived
from the inductance of the trace and the large gate
capacitance of the FET . Place the input bulk capacitors as
close to the drains of MOSFETs as possible. In addition,
place the 0.1µF decoupling capacitors right on the drain
of each MOSFET . This helps to suppress some of the high
frequency switching noise on the DC-DC converter input.

• The traces that run from the RC5040 IFB (pin 4) and VFB
(pin 5) pins should be run next to each other and be Kelvin
connected to the sense resistor. Running these lines
together helps to reject some of the common mode noise
to the RC5040 feedback input. Run the noisy switching
signals (HIDRV & VCCQP) on one layer, and use the
inner layers for power and ground only. If the top layer is
being used to route all of the noisy switching signals, use
the bottom layer to route the analog sensing signals VFB
and IFB.

ESR < 0.047 Ω

< 37m ohm

< 4V, ID > 20A

GS

16

APPLICATION NOTE AN42

Good layout

1115RC5040 RC5040
12
13

14

16
17
18
19
20

10

9
8

7
6
5
4
3
2
1

“Quiet” Pins=

Figure 16. Example of Proper MOSFETs Placements

PC Motherboard Layout and Gerber File

A reference design for motherboard implementation of the
RC5040 and RC5042 along with the Layout Gerber File and
Silk Screen are presented below. The actual PCAD Gerber

Bad layout

11
12
13

14
15
16
17
18
19
20

10

9
8

7
6
5
4
3
2
1

File can be obtained from Raytheon Electronics Semicon­ductor Division’s Marketing Department at [phone].

17

AN42 APPLICATION NOTE

18

APPLICATION NOTE AN42

Guidelines for Debugging and
Performance Evaluations

Debugging Your First Design Implementation

Use the following procedure to help you debug your design
implementation:

1. Note the VID pins settings. They tell you what voltage is
to be expected.

2. Do not connect any load to the circuit. While monitoring
the output voltage, apply power to the part with current
limiting at the power supply. Do this to make sure that
no catastrophic shorts occur.

3. Ιf proper voltage is not achieved, follow the procedures
in the Troubleshooting section.

5. Apply load at 1A increments; an active load (HP6060B
or equivalent) is suggested.

6. In case of poor regulation, refer to the procedures in the
Troubleshooting section.

Troubleshooting

1. If no voltage is registered at the output and the circuit is
not drawing current, look for openings in the connec­tions. Check the circuitry versus the schematic, and the
power supply pins at the device to ascertain that volt­age(s) had been applied.

2. If no voltage is registered at the output and the circuit is
drawing excessive current (>100mA) with no load,
check for possible shorts. Trace the path of the excessi ve
current to determine if the controller is at fault or if the
excessive current is due to peripheral components.

4. After there is proper voltage, increase the current limit­ing of the power supply to 16A.

3. If the output voltage comes near to, but is not, what is
expected, check the VID inputs at the device pins. The
part is factory set to correspond to the VID inputs.

19

AN42 APPLICATION NOTE

4. Premature shut down can be caused by an inappropriate
value of sense resistor. See the Sense Resistor section.

5. A poor load regulation can have many causes. You
should first check the voltages and signals at the critical
pins.

6. The VREF pin should be at the voltage set by the VID
pins. If the power supply pins are correct and the VID
pins are correct, the VREF should be at the correct volt­age.

7. Next check the oscillator pin. A saw tooth wave at the
frequency set by the external capacitor should be seen.

8. When the VREF and CEXT pins are determined to be
correct and the output voltage is still incorrect look at
the waveform at VCCQP. This pin should be swinging
from ground to +12V (in the +12V application) and
from slightly below +5V to about +10V (charge pump
application). If the VCCQP pin is noisy , with ripples and
overshoots, then the noise may cause the converter to
function improperly.

9. Next, look at the HIDRV pin. This pin directly dri ves the
gate of the FET. It should provide a gate drive (Vgs) of
about 5V when turning the FET on. A careful study of
the layout is recommended. See the PCB Layout Guide-
lines and Considerations section.

10. Experience shows that the most frequent errors are using
incorrect components, improper connections, and poor
layout.

Performance Evaluation

This section shows the results of a random sample evalua­tion. Use these results as a reference guide for evaluating the
RC5040 DC-DC converter for Pentium Pro motherboards.

Load Regulation

VID I

(A) V

load

0100 0.5 3.0904

1.0 3.0825

2.0 3.0786

3.0 3.0730

4.0 3.0695

5.0 3.0693

6.0 3.0695

7.0 3.0695

8.0 3.0694

9.0 3.0694

9.9 3.0691

Load Regulation 0.5A – 9.9A 0.70%

VID I

(A) V

load

0010 0.5 3.2805

1.0 3.2741

2.0 3.2701

3.0 3.2642

4.0 3.2595

5.0 3.2597

6.0 3.2606

7.0 3.2611

8.0 3.2613

9.0 3.2611

10.0 3.2607

11.0 3.2599

12.0 3.2596

12.4 3.2596

Load Regulation 0.5A – 12.4A 0.64%

out

out

(V)

(V)

20

APPLICATION NOTE AN42

VID I

(A) V

load

out

(V)

1010 0.5 2.505

1.0 2.504

2.0 2.501

3.0 2.496

4.0 2.493

5.0 2.493

6.0 2.492

7.0 2.492

8.0 2.491

9.0 2.490

10.0 2.489

11.0 2.488

12.0 2.486

13.0 2.485

13.9 2.484

Load Regulation 0.5 - 13.9A 0.84%

Note:

Load regulation is expected to be typically around 0.8%. The
load regulation performance for this device under evaluation is
excellent.

Output V olta ge Load Transients Due to Load Current Step

This test is performed using Intel P6.0/P6S/P6T Voltage
Transient Tester.

Low to High
Current Step

0.5A-9.9A - 76.0mV Refer to
Attachment
A for Scope
Picture

High to Low
Current Step

9.9A-0.5A + 70mV Refer to
Attachment
B for Scope
Picture

Low to High
Current Step

0.5A-12.4A - 97.6mV Refer to
Attachment
C for Scope
Picture

High to Low
Current Step

12.4A-0.5A + 80.0mV Refer to
Attachment
D for Scope
Picture

Low to High
Current Step

0.5A-13.9A - 99.2mV Refer to
Attachment
E for Scope
Picture

High to Low
Current Step

13.9A-0.5A + 105.2mV Refer to
Attachment
F for Scope
Picture

Note:

Excellent transient voltage response. Transient voltage is rec­ommended to be less than 4% of the output voltage. The per­formance of the device under evaluation is significantly better
than a typical VRM.

Input Ripple and Power on Input Rush Current

= 9.9A Input Ripple

I

load

Voltage = 15mV

Refer to Attachment
G for Scope Picture

Power on Input Rush Current is not measured on the mother­board because we did not want to cut the 5V trace and insert
current probe in series with the supply. However, with the
input filter design, the Input Rush Current will be well within
specification.

21

AN42 APPLICATION NOTE

Component Case Temperature

Device Description

Q3A MOSFET

Case Temperature

(°C)

I

= 9.9A

load

57 63 56.3

Case Temperature

(°C)

I

= 12.4A

load

Case Temperature

(°C)

I

=13.9A

load

K1388

Q3B MOSFET

58 64 66.6

K1388

L1 Inductor,

53 56 61.2

Unknown

Q2 Schottky Diode

66 70 87

2048CT
IC Raytheon RC5040 52 54 58
Cin Input Capacitor 1000µF 38.2 36.8 39
Cout Output Capacitor

35 34.8 38.2

1500µF

Note:

Case temperatures are all within guidelines. Our guideline is that case temperatures for all components should be below 105°C
@25°C Ambient.

Comments:
Excellent input ripple voltage. Input ripple voltage is recommended to be less than 5% of the output voltage.

Evaluation Summary:

The on-board DC-DC converter is fully functional. It has
excellent load regulation, transient response, and input
voltage ripple.

Attachment A Attachment B

22

APPLICATION NOTE AN42

Attachment C

Attachment D

Attachment E

Attachment F

Attachment G

Summary

This application note covers for implementation of a DC-DC
converter on a Pentium Pro motherboard using the RC5040
and RC5042. The detailed discussion includes Pentium Pro
processor power requirements, RC5040 and RC5042
description, design considerationsn and component selec­tions, layout guidelines and considerations, guidelines for
debugging, and performance evaluations.

RC5040/RC5042 Evaluation Board

Raytheon Electronics provides an evaluation board for the
purpose of verifying system level performance of the
RC5040 and RC5042. The evaluation board serv es as a guide
as to what can be expected in performance with the supplied
external components and PCB layout. Please call Raytheon
Electronics Marketing Department at [phone] for an
evaluation board.

23

AN42 APPLICATION NOTE

Appendix A: Directory of Component Suppliers

Dale Electronics, Inc.
E. Hwy. 50, PO Box 180
Yankton, SD 57078-0180
PH: [phone]

Fuji Electric
Collmer Semiconductor Inc.
14368 Proton Rd.
Dallas, Texas 75244
PH: [phone]

General Instrument
Power Semiconductor Division
10 Melville Park Road
Melville, NY 11747
PH: [phone]

Hoskins Manufacturing Co.
(Copel Resistor Wire)
10776 Hall Road
Hamburg, MI 48139-0218
PH: [phone]

Intel Corp.
5200 NE Elam Young Pkwy.
Hillsboro, OR. 97123
PH: [phone] Tech. Support
for Power Validator

International Rectifier
233 Kansas St.
El Segundo, CA 90245
PH: [phone]

IRC Inc.
PO Box 1860
Boone, NC 28607
PH: [phone]

Motorola Semiconductors
PO Box 20912
Phoenix, Arizona 85036
PH:[phone]

National Semiconductor
2900 Semiconductor Drive
Santa Clara, CA 95052-8090
PH: [phone]

Nihon Inter Electronics Corp.
Quantum Marketing Int’l, Inc.
12900 Rolling Oaks Rd.
Caliente, CA 93518
PH: [phone]

Panasonic Industrial Co.
6550 Katella Avenue
Cypress, CA 90630
PH: [phone]

Pulse Engineering
12220 World Trade Drive
San Diego, CA 92128
PH: [phone]

Sanyo Energy USA
2001 Sanyo Avenue
San Diego, CA 92173
PH: [phone]

Siliconix
Temic Semiconductors
2201 Laurelwood Road
Santa Clara, CA 95056-1595
PH: [phone]

Sumida Electric USA
5999 New Wilke Road Suite #110
Rolling Meadows, IL 60008
PH: [phone]

Xicon Capacitors
PO Box 170537
Arlington, Texas 76003
PH:[phone]

LIFE SUPPORT POLICY

FAIRCHILD’S PRODUCTS ARE NOT AUTHORIZED FOR USE AS CRITICAL COMPONENTS IN LIFE SUPPORT DEVICES
OR SYSTEMS WITHOUT THE EXPRESS WRITTEN APPROVAL OF THE PRESIDENT OF FAIRCHILD SEMICONDUCTOR
CORPORATION. As used herein:

1. Life support devices or systems are devices or systems
which, (a) are intended for surgical implant into the body,
or (b) support or sustain life, and (c) whose failure to
perform when properly used in accordance with

2. A critical component in any component of a life support
device or system whose failure to perform can be
reasonably expected to cause the failure of the life support
device or system, or to affect its safety or effectiveness.

instructions for use provided in the labeling, can be
reasonable expected to result in a significant injury of the
user.

Fairchild Semiconductor
Corporation

Americas

Customer Response Center
Tel:[phone]

www.fairchildsemi.com

Fairchild Semiconductor
Europe

Fax: [phone]

Email: [e-mail]
Deutsch Tel: [phone]
English Tel: [phone]
Italy Tel: [phone]

Fairchild Semiconductor
Hong Kong Ltd.

13th Floor, Straight Block,
Ocean Center, 5 Canto Rd.
Tsimshatsui, Kowloon
Hong Kong
Tel:[phone]
Fax:[phone]

National Semiconductor
Japan Ltd.

Tel:[phone]
Fax:[phone]

2/98 0.0m

 1998 Fairchild Semiconductor Corporation

Stock#AN30000042