Modular architecture for high bandwidth computers

BACKGROUND OF THE INVENTION

This application relates to an architecture for a computer system, and specifically, to an architecture that allows partitioning of electronic components on printed circuit boards (also called “cards”) of the computer system in such a way that any printed circuit board may be easily replaced without having to redesign or replace any other printed circuit board in the related system. In addition, the printed circuit boards and backplanes can be arranged in various appropriate configurations.

Conventional computer systems contain electronic components that are located on printed circuit boards (PCBs). PCBs are also called “cards,” “daughtercards,” or “motherboards.” Conventional computers contain the majority of their components on a main PCB called a “motherboard.” The motherboard usually contains at least a processor, memory, and a peripheral controller. The motherboard usually also contains various bus logic chips, buffers, bus protocol circuitry, and memory management chips.

Some conventional systems include additional PCBs in addition to the motherboard. These PCBs contain electronics used by the motherboard, where the electronics are of a type compatible with the motherboard. Such electronics may include controllers for add-on peripherals, video circuitry, sound circuitry, etc. Other conventional systems contain a memory subsystem in low-bandwidth pluggable modules (called single in-line memory modules or “SIMMS”) on one or more separate PCBs.

The electronic elements on a motherboard are connected to one another on the motherboard by one or more “busses” and by lines carrying various control signals. Busses transmit addresses, data, control signals, etc. between electronic components. A motherboard is connected to other PCBs by one or more “connectors.” Each connector has “pins,” some of which transmit signals that are passed between the motherboard and the other PCBs and some of which are connected to power or ground. Signal paths called “traces” connect the connectors on the PCBs, backplanes, and/or motherboards.

Conventional connectors that are used to connect PCBs cannot achieve a density much higher than eighty contacts per linear inch. This low density limits the number of pins that can be located on a connector and limits the possible width of busses connecting the motherboard to other PCBS. In addition, when a connector contains a relatively small number of pins, signals are often multiplexed on at least some of the pins. When two signals are multiplexed on a single pin, for example, the signals are transmitted at different times over the single pin.

Multiplexed signals add electronic overhead and slow the operational speed of the system. As an alternative to narrow busses and multiplexed signals, some conventional systems simply use very large connectors. Such a size increase causes timing problems. Similarly, undesirable effects such as noise, signal disturbances, propagation delay, and cross-talk increase along with connector size. Some connector pins must be used for power and ground signals. It is desirable to have a relationship of 2:1 or 3:1 between signal and power/ground. Yet, such a relationship is not possible within the limitations of conventional low density connectors. Thus, the pin-out limits and size of conventional connector technology places limitations on the types of electronic components that can be located on boards other than the motherboard.

Currently, it is becoming desirable for computer systems to be able to use whatever processor works best for a certain task to be performed. For example, a first type of processor might work best for graphics processing while a second type of processor might be the best choice to act as a network server. It is desirable to be able to have a system that can interchange various types of processors according to, e.g., the task to be performed by the computer system. It is also desirable when changing a processor used by the system to be able to use an operating system and associated application software that are optimal for the new processor.

Conventional computer systems usually include a processor on the motherboard. Some conventional systems often allow a user to substitute processors by unplugging a first type of processor chip from the motherboard and replacing it with a second type of processor chip. Such substitution, however, can only be performed between processor chips having identical bus sizes and similar architectures. Specifically, both processor chips must be compatible with the other electronics on the motherboard.

In conventional systems, the architecture of a computer system is dictated by a type of processor used in the computer system. Thus, for example, a processor chip on a motherboard cannot be upgraded to a processor chip having a different architecture. To use a processor having a different architecture, the other components on the motherboard must be redesigned to operate with the new processor. Similarly, a type of peripheral controller used in a system determines a type of peripheral bus used in the system and a type of peripheral cards accepted by the system. To use a new type of peripheral bus in a system, the motherboard must be redesigned to accept a corresponding new type of peripheral controller.

The evolution of the personal computer has been marked by significant increases in processor speed. Bus widths have continued to increase for every new generation of processor. It is now common to integrate memory management and peripheral support functions into “chip sets.” The introduction of a new processor or chip set has previously required that the computer's motherboard be redesigned to benefit fully from the increased functionality and bandwidth of the new processor. The high speeds and dense packages dictate that the processor, the chip set, and the bus that interconnects them be placed on a single motherboard. The use of a motherboard limits the extent to which an existing system can be upgraded when new technologies become available because a motherboard is designed to operate only with certain bus widths, memory management schemes, peripheral busses and expansion slots.

In general, therefore, it is desirable to make the components of a computer system as modular as possible. When most of the components of a computer system are located on a motherboard, the motherboard will necessarily be large. Manufacture of these large boards is more complex than manufacture of small boards and, therefore, large boards are more difficult and costly to manufacture. In general, the effects of the many small tolerances required by a large motherboard combine to cause manufacturing problems for large boards, resulting in a lower yield of usable boards during the manufacturing process. Large boards also must be thicker than small boards to avoid warpage and to facilitate routing of tracer.

In addition, the larger a board is, the more components are located on the board. Large boards are also more difficult and costly to repair than small boards because, for example, if a single component on a motherboard is faulty, the entire board must be removed from the computer for repair or replacement. As stated above, although it is desirable to have modular components in a computer system, the pin-out limits of conventional connectors make modularity impracticable.

SUMMARY OF THE INVENTION

The present invention overcomes the problems and disadvantages of the prior art by altering the way electronic parts are partitioned among printed circuit boards (PCBs). The present invention eliminates the use of a traditional motherboard. Instead, the electronic components of the computer are partitioned between a plurality of PCBs in such a way that all components relating to the processor are placed on a first PCB, all components of the computer relating to memory are placed on at least a second PCB, all the electronic components of the computer relating to peripheral control are placed on at least a third PCB, etc. The plurality of PCBs are connected through a backplane. Thus, the PCBs containing the processor, the memory, the peripheral controller, or any other components can be replaced or upgraded easily by PCBs that contain elements performing the same general function. The present invention solves the problem of pin-out limitations between the PCBs by using a new type of high-density connector, which eliminates conventional I/O limitations and eases timing constraints.

In addition, in the present invention, a backplane or backplanes can be configured in a variety of ways using high-density connectors. The PCBs may be mounted on the backplanes using a variety of mounting techniques.

In accordance with the purpose of the invention, as embodied and broadly described herein, the invention is a computer system, comprising: a local bus; a memory bus; a first backplane including a first PCB containing a microprocessor thereon and connected to the local bus and to the memory bus, a second PCB containing a memory thereon and connected to the memory bus, and a third PCB having a peripheral controller thereon and connected to the local bus; and a second backplane having a peripheral connector and connected to the first backplane.

In accordance with the purpose of the invention, as embodied and broadly described herein, the invention is a computer system, comprising: a local bus; a memory bus; a first PCB containing a microprocessor thereon and connected to the local bus and the memory bus through a first high-density connector system; a second PCB containing a memory thereon and connected to the memory bus through a second high-density connector system; and a third PCB having a peripheral controller thereon and connected to the local bus through a third high-density connector system, where at least one of the first, second, and third high-density connector systems having a density of at least 100 contacts per linear inch.

Objects and advantages of the invention will be set forth in part in the description which follows and in part will be obvious from the description or may be learned by practice of the invention. The objects and advantages of the invention will be realized and attained by means of the elements and combinations particularly pointed out in the appended claims.

BRIEF DESCRIPTION OF THE DRAWINGS

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate several embodiments of the invention and, together with the description, serve to explain the principles of the invention.

FIG. 1

is a block diagram of principal PCBS in an embodiment of the present invention;

FIG. 2

is a detailed block diagram of an embodiment of the present invention showing main signals between the PCBs of

FIG. 1

;

FIG. 3

is a block diagram of a 64-bit CPU PCB of the embodiment of

FIG. 2

;

FIG. 4

is a block diagram of a memory PCB of the embodiment of

FIG. 2

configured with 128M bytes of memory;

FIG. 5

is a block diagram of an alternate memory PCB of the embodiment of

FIG. 2

which may be configured with 16M byte or 64M bytes of memory;

FIG. 6

is a backplane wiring diagram for the memory cards of

FIG. 2

;

FIG. 7

is a block diagram of a controller PCB of the embodiment of

FIG. 2

;

FIG. 8

is a block diagram of a 32-bit CPU PCB of the embodiment of

FIG. 2

;

FIG. 9

is a block diagram of a 128-bit CPU PCB of the embodiment of

FIG. 2

;

FIG. 10

is a block diagram showing an arrangement of memory boards connected to a 128-bit PCB;

FIG. 11

shows an embodiment of the present invention incorporating a single backplane using surface mount technology and through-hole connectors;

FIG. 12

shows another view of the backplane of

FIG. 11

with the PCBs removed;

FIG. 13

shows an embodiment of the present invention incorporating multiple backplanes connected by a right angle high-density connector;

FIG. 14

shows an embodiment of the present invention incorporating multiple parallel backplanes connected by a right angle high-density connector;

FIG. 15

shows an embodiment of the present invention incorporating multiple parallel backplanes connected by a right angle high-density connector, where one backplane uses surface mount connectors on both sides;

FIG. 16

shows an embodiment of the present invention incorporating multiple backplanes connected by a vertical high-density connector;

FIG. 17

shows an embodiment of the present invention incorporating multiple backplanes connected by a vertical high-density connector;

FIG. 18

shows an embodiment of the present invention incorporating multiple backplanes connected by a vertical high-density connector;

FIG. 19

shows a side view of an embodiment of the present invention wherein surface mounted connectors for PCBs are placed on a double sided backplane in a way that balances impedances of the paths to the PCBs;

FIG. 20

shows an impedance loading of the embodiment of

FIG. 19

;

FIG. 21

is a diagram showing an arrangement of backplanes to maximize cooling, yet allow large components to be a part of the system;

FIG. 22

is a diagram of a projection component of a high-density connector connecting the PCBs of the present invention;

FIG. 23

is a diagram of a side view of two projection components of a high-density connector;

FIG. 24

is a diagram of a high-density connector for a right-angle connection;

FIG. 25

is a diagram of a group of receiving type component for a high-density connector;

FIG. 26

is a diagram of the projection component of

FIG. 22

mated with the receiving type component of

FIG. 25

;

FIG. 27

is a perspective view of an electrical interconnect system showing insulative electrical carriers functioning as the substrates for the system for a vertical connection;

FIG. 28

is a perspective view of another electrical interconnect system showing insulative electrical carriers functioning as the substrates for the system for a right angle connection;

FIG. 29

is a grid showing clusters of high-density connectors arranged in a nested fashion;

FIG. 30

is a detail of the nested clusters of

FIG. 29

;

FIG. 31

is a grid showing clusters of high-density connectors arranged in a modified version of the arrangement of

FIG. 29

;

FIG. 32

is a detail of the clusters of

FIG. 31

;

FIG. 33

is a detail of the clusters of

FIG. 31

mated;

FIG. 34

is a detail of the clusters of

FIG. 31

;

FIG. 35

is a diagram of a high-density connector used in an embodiment of the present invention;

FIG. 36

is a diagram of another high-density connector used in an embodiment of the present invention;

FIG. 37

is a diagram of a high-density connector system used in an embodiment of the present invention; and

FIG. 38

is a diagram of a high-density connector system used in an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

Reference will now be made in detail to the preferred embodiments of the invention, examples of which are illustrated in the accompanying drawings. Wherever possible, the same reference numbers will be used throughout the drawings to refer to the same or like parts.

The present invention is a computer system architecture in which functionally compatible electronic components are located on modular printed circuit boards. Thus, for example, a type of processor used by the system can be changed by replacing the printed circuit board incorporating the processor. Similarly a type of peripheral bus used can be changed simply by replacing the printed circuit board containing the peripheral controller. High-density connectors connect the circuit boards. Some embodiments of the invention use a single backplane. Other embodiments place peripheral slots on a second, passive backplane.

A. Description of the Architecture

FIG. 1

is a block diagram of an embodiment of a computer system

100

in accordance with the present invention. Computer system

100

includes at least a CPU printed circuit board (PCB)

102

, at least one memory PCB

104

, and at least one controller PCB

108

. In

FIG. 1

, dotted lines are used to show elements that may not be present in certain embodiments of the invention (e.g., Some embodiments may have only one memory PCB

104

; some embodiments may have only one controller PCB

108

; and/or some embodiments may have various numbers of local bus slots and zero or more peripheral bus slots).

CPU PCB

102

includes a central processing unit (also called a “CPU” or a “processor”) and various circuitry relating to the CPU, e.g., bus interface logic and/or cache memory. The CPU and related circuitry are described in detail below. Memory PCB(s)

104

includes a plurality of memory chips, as described below. The details of exemplary memory PCB(s)

104

are also described in detail below. Controller PCB(s)

108

include a peripheral controller and also contain electronic components to handle input and output (I/O) to/from the system

100

. A controller PCB

108

also are described in detail below.

Computer system

100

also includes a local bus

110

connecting CPU PCB

102

to controller PCB(s)

108

and to slots

107

. Local bus

110

includes address lines, data lines, and various control lines. Computer system

100

also includes a memory bus

112

connecting CPU PCB

102

and memory PCB(s)

104

. Memory bus

112

is discussed in detail below. Computer system

100

also includes one or more peripheral busses

114

connecting the controller PCB(s)

108

and peripheral slots

109

. The protocol of peripheral bus

114

is determined by a type of peripheral controller contained on controller PCB(s)

108

.

The present invention may include various numbers of slots

107

. Examples of PCBs for slots

107

are discussed below. Slots

109

are designed so that PCBs designed to connect to peripheral bus

114

may be inserted therein. The present invention may include various numbers of slots

109

. Examples of PCBs for slots

109

are discussed below in connection with, e.g.,

FIGS. 2 and 7

.

Computer system

100

also includes a first backplane

124

and a second backplane

126

. A high-density connector

120

connects local bus

110

and peripheral bus

114

across the backplanes. Additional high-density connectors

122

connect CPU PCB

102

to local bus

110

and to memory bus

112

; connect memory PCB(s)

104

to memory bus

112

; and connect controller PCB(s)

108

to local bus

110

and to peripheral bus

114

. Each connector

122

is a mated pair of high-density connectors or high-density connector systems. One connector is attached to the PCB and its mate is attached to the backplane. Connector

120

is a mated pair of high-density connectors or high-density connector systems. (See FIGS.

35

-

38

).

Other embodiments of the invention may occupy a single backplane instead of a plurality of backplanes. In this case, connector

120

is not needed and is not used. Still other embodiments may occupy more than two backplanes and may, for example, use more than one high-density connector

120

to connect the backplanes. Other embodiments may use high-density connectors to connect two PCBs located on opposite sides of a backplane or to connect two backplanes placed “back-to-back.”

In other embodiments of the invention, some subset of connectors

122

shown in

FIG. 1

may be used or additional connectors

122

may be used to connect peripheral slots to local bus

110

or to peripheral bus

114

.

FIG. 1

shows a single connector

122

attached to each of PCBs

102

,

104

, and

108

. Other embodiments use multiple high-density connectors

122

to connect PCBs to busses. Still other embodiments use multiple connectors at locations where

FIG. 1

shows single connectors.

FIG. 2

is a detailed block diagram of a preferred embodiment of the present invention. Alternately, various other types and sizes of processors may be included on a CPU PCB

102

, as is discussed below in connection with, e.g.,

FIGS. 3

,

8

and

9

. In

FIG. 2

, local bus

110

preferably is a Peripheral Component Interconnect (PCI) bus, which is described in “Peripheral Component Interconnect (PCI), Revision 1.0 Specification,” Jun. 22, 1992, which can be obtained from Intel Corp. and which is herein incorporated by reference. (“PCI” and “Peripheral Component Interconnect” are both trademarks of Intel Corp.) Other embodiments may use other busses as local bus

110

. Computer system

100

also includes lines IRQ and ICC, which connect CPU PCB

102

and controller PCBs

108

. These lines are used to manage the interrupts of the system and the peripherals.

FIG. 2

shows two memory PCBs

104

a

and

104

b

. The memory chips in memory PCBs

104

a

preferably are MT4C16M1A 16M×1 DRAMs manufactured by Micron Technology, Inc., although PCBs

104

a

and

104

b

can include other types of memory, as described below. The signals input to and output from memory PCBs

104

a

and

104

b

are described below in connection with

FIGS. 4-6

. Memory bus

112

is also described in connection with

FIGS. 4-6

.

In

FIG. 2

, controller PCB

108

preferably controls a plurality of (EISA) (Enhanced Industry Standard Architecture) peripherals inserted in slots

109

. Thus, in the described embodiment, peripheral bus

114

preferably includes an EISA standard bus. The EISA standard is described in “EISA Specification, V 3.10,” which can be obtained from BCPR Services, Inc. in Washington, D.C., and which is incorporated by reference. Other parts of the bus

114

control SCSI devices, IDE devices, etc., in a manner known to persons of ordinary skill in the art. The signals being transferred between the various components of

FIG. 2

are shown in Table 1.

TABLE 1

PCI BUS BACKPLANE SIGNALS (input/output to/from the PCB in slot

107)

PCI SIGNALS

As defined in the PCI specification.

AD[0:63]

Address and Data: Bi-directional data

lines for the PCI bus. The signals

sample or drive the address and data

on the PCI bus.

GNT-[1:7]

Grant: When asserted, indicates that

access to the PCI bus has been granted

to a bus master by the PCI bus arbi-

ter. Each PCI bus master device has a

GNT signal.

REQ-[1:7]

Request: A PCI bus master device as-

serts this signal to indicate to the

PCI bus arbiter that it is requesting

use of the PCI bus. Each PCI bus

master device has a REQ signal.

CONTROLLER BOARD INTERFACE SIGNALS

(for controller PCB 108)

IRQ[1:15]

Interrupt Request Lines: These 15

interrupt inputs accept interrupt re-

quests from I/O or other devices.

ICC[0:3]

Interrupt Controller Communications

Bus: The ICC bus is used to pass in-

terrupt messages among I/O units and

multiple processors.

MEMORY INTERFACE SIGNALS

(output from CPU 102 to memory PCBs 104)

MD[0:31]

Memory Data Bus, low order bits: Bi-

directional data lines for the memory

data bus. The signals drive data

destined for either the Host data bus

or the PCI bus. They also input data

that originated from either the Host

data bus or the PCI bus.

MD[32:63]

Memory Data Bus, high order bits: Bi-

directional data lines for the memory

data bus. The signals drive data des-

tined for either the Host data bus or

the PCI bus. They also input data

that originated from either the Host

data bus or the PCI bus.

DP[0:7]

Memory Parity: Bi-directional byte

enable parity signals for the memory

data bus. The low order parity bit

DP[0] corresponds to D[0:7], while the

high order parity bit DP[3] cor-

responds to D[24:31]. The DP[0:7] are

parity outputs during write cycles to

memory and parity inputs during read

cycles from memory.

MA[1:11]

DRAM Multiplexed Address: Provide the

row and column address to the DRAM

array.

CAS[0:7]

Column Address Strobes: These signals

are used to latch the column address

on the MA[1:11] lines into the DRAMs.

Each signal corresponds to one byte of

the eight byte wide array.

RAS[0:5]

Row Address Strobes: These signals

are used to latch the row address on

the MA[1:11] lines into the DRAMs.

Each signal corresponds to one eight

byte wide DRAM row.

MRW-

DRAM Write Enable: This signal is

asserted during both CPU and PCI mas-

ter writes to main memory.

BANKSEL

Memory Bank Select

EMAO

Even Memory Address, bit 0

ROMAO

Odd Memory Address, bit 0

ADDITIONAL I/O PORTS (connected to controller PCB 108)

EISA BUS

Enhanced Industry Standard Architec-

ture Bus. There are five EISA expan-

sion slots.

SCSI BUS

Small Computer Systems Interface Bus.

The SCSI host adapter supports up to

eight peripherals, operating under

SCSI-2 protocols at 10Mbytes/s. Refer

to NCR 53C810 Data Manual.

IDE BUS

Integrated Drive Electronics Bus. As

defined in the Chips and Technologies,

Inc. 82C711 Data Book (San Jose,

California 1993), which is

incorporated by reference.

FLOPPY DRIVE

Control signals for floppy disk drives

BUS

as per CHIPs 82C711 specification.

KEYBOARD

Standard 101-key enhanced keyboard

INTERFACE

interface.

MOUSE PORT

Microsoft mouse compatible 9-pin

interface.

PARALLEL PORT

IBM XT/AT compatible parallel port,

with bi-directional support.

SERIAL PORT #1

NS16450 compatible UART for serial

data transmission.

SERIAL PORT #2

Same as above.

The signal definitions in Table 1 serve to completely define the interface between CPU PCB

102

, memory PCBs

104

a

and

104

b

, controller PCB

108

, slots

107

and

109

, local bus

110

, memory bus

112

; and peripheral bus

114

. Each signal input or output over any of these busses is defined within Table 1. Thus, in the described embodiment, the signals output from at least CPU PCB

102

and memory PCBs

104

a

and

104

b

are standardized. The signals between CPU PCB

102

and controller

108

and between CPU PCB

102

and slots

107

are also standardized. Other embodiments of the invention may standardize the inputs and outputs of the PCBs in different ways. It is, however, this standardization of the signals between the PCBs that allows the PCBs of the invention to be modular.

FIG. 3

is a block diagram of a 64-bit CPU PCB

102

that may be used in the embodiment of FIG.

2

. The CPU PCB

102

of

FIG. 3

preferably includes a 66 megahertz Pentium microprocessor chip, which is manufactured by Intel Corp. and is described in Pentium Processor User's Manual, Vol. 1-3, 1993, and which is incorporated herein by reference. Other embodiments may use other processors.

FIG. 3

includes the CPU

302

, an address latch

303

, a 512K byte cache RAM

304

, a PCI bridge interface (PCMC)

306

, a Local Bus Accelerator (LBX) low order bytes

308

, an LBX high order bytes

310

, a programmable interrupt control (APIC)

314

, additional buffering and control logic (not shown for clarity of explanation), and a high-density connector

316

. Connector

316

has signals corresponding to the signals shown on CPU PCB

102

of FIG.

2

.

The cache memory is configured as four 64K×18 SRAMs, including parity and is accessed by the 64-bit host data bus. The PCMC chip integrates the cache control and main memory DRAM control functions and provides bus control for transfers between the CPU, cache, memory

104

, and local bus

110

(the PCI bus). The cache controller supports up to 512K byte of cache RAM. The PCMC also integrates a high-performance Tag RAM. In some embodiments, the cache is removable from the PCB, thereby allowing caches of various sizes to be used with the microprocessor.

There are two LBXs—one for low order and one for high order addresses, respectively. The two devices

308

and

310

provide a 64-bit path between the CPU/cache and main memory

104

, a 32-bit data path between the CPU and the PCI bus (local bus

110

) and memory

104

. A dual-port architecture allows concurrent operations on the host and PCI busses. The LBXs support byte parity for the host and main memory busses. During bus operations between the CPU

302

, memory

104

, and PCI bus

10

, the PCMC commands the LBX to performs functions such as latching addresses and data, merging data, and enabling output buffers. The LBXs also contain write buffers and read pre-fetch buffers.

PCMC

306

is preferably an 82434LX chip from Intel. LBX

308

and

310

are preferably 82433LX chips from Intel. Address latch

303

is preferably 74AS373 latches from Toshiba. Cache RAM

304

is preferably MCM620520 chips from Motorola. APIC

314

is preferably an 82498DX APIC from Intel. High-density connector

316

is described below. Signals sent between the elements of

FIG. 3

are shown in Table 2 below.

TABLE 2

Pentium CPU Card Signal Description

HOST BUS SIGNALS

HA[0:31]

Host Address Lines: These are the

address lines of the Host bus.

Together with the byte enable signals,

they define the physical area of

memory or input/output space accessed.

Address lines HA3-HA31 are connected

to the CPU, and are used for address

decoding and L2 cache management.

HAO-2 are only connected to the local

bus accelerators (LBX).

BE[0:7]

Byte Enable: These signals indicate

which byte lanes on the CPU data bus

carry valid data during the current

bus cycle.

HP[0:7]

Host Data Parity: There is one bit

for every byte on the data bus. They

are driven by the CPU with even parity

information on writes in the same

clock as write data. The LBX transfer

the parity information to and from

main memory DP[0:7].

HD[0:63]

Host Data Lines: They are the data

lines from the microprocessor. The

LBX transfers this data to and from

main memory. The HD interfaces

directly to the cache memory

PCI/CACHE AND DRAM CONTROLLER (PCMC) AND LOCAL

BUS ACCELERATORS (LBX)

AD[0:31]

Address and Data: Low order bi-

directional data lines for the PCI

bus. The signals sample or drive the

address and data on the PCI bus.

AD[32:63]

Address and Data: High order bi-

directional data lines for the PCI

bus. The signals sample or drive the

address and data on the PCI bus.

These signals are not used with a 32-

bit PCI bus.

PCI SIGNALS

As defined in the PCI specification.

(Referenced elsewhere in this

specification).

MA[0:11]

DRAM Multiplexed Address: Provide the

row and column address to the DRAM

array.

CAS[0:7]

Column Address Strobes: These signals

are used to latch the column address

on the MA[0:11] lines into the DRAMs.

Each signal corresponds to one byte of

the eight byte wide array.

RAS[0:5]

Row Address Strobes: These signals

are used to latch the row address on

the MA[0:10] lines into the DRAMs.

Each signal corresponds to one eight

byte wide DRAM row.

MRW-

DRAM Write Enable: This signal is

asserted during both CPU and PCI

master writes to main memory.

CAA[3:6]

Second Level Cache Address: Addresses

for SRAM.

CWE[0:7]

Cache Write Enables: Asserted to

write data to the second level cache

SRAMs on a byte-by-byte basis.

CA[7:18]

Latched Cache Address: Generated from

the processor address lines, active

when accessing the cache memory.

MD[0:63]

Memory Data Bus: Bi-directional data

lines for the memory data bus. The

signals drive data destined for either

the Host data bus or the PCI bus. It

also inputs data that originated from

either the Host data bus or the PCI

bus.

DP[0:7]

Memory Parity: Bi-directional byte

enable parity signals for the memory

data bus. The low order parity bit

DP[0] corresponds to MD[0:7], while

the high order parity bit DP[3]

corresponds to MD[24:31]. The DP[0:7]

are parity outputs during write cycles

to memory and parity inputs during

read cycles from memory.

INTERRUPT CONTROLLER SIGNALS

IRQ[1:15]

Interrupt Request Lines: These 15

interrupt inputs accept interrupt

requests from I/O or other devices.

ICC[0:3]

Interrupt Controller Communications

Bus: The ICC bus is used to pass

interrupt messages among I/O units and

multiple processors.

Electrical characteristics and timing information for the Intel chip sets that can be used to implement the elements shown in

FIG. 3

are described in “82430 PCIset Cache/Memory Subsystem,” which is available from Intel Corp., which is herein incorporated by reference. Timing requirements for the PCI bus are described in “The PCI Specification,” published

1992

, available from Intel, specifically in chapter 4.42 “System Parameters,” which is herein incorporated by reference.

When the local bus

110

is implemented as a PCI bus, some of the signals that make up the bus need to meet stringent timing requirements. These requirements are specified in 1) “Intel 82430 PCI Set Cache/Memory Subsystem, Section 9.4—AC Characteristics,” pages 171-178, which is herein incorporated by reference, and in 2) “Intel PCI Rev. 1 Specification,” pages 69-74, which is incorporated by reference. When high-density connector

316

is used, these timing requirements are met.

FIG. 4 and 5

are respective block diagrams of the memory PCB

104

a

or

104

b

of the embodiment of FIG.

2

. (PCB

104

a

receives MD[0:31]; PCB

104

b

receives MD[32:63]).

FIG. 4

shows a PCB using 16M×1 DRAMs and

FIG. 5

shows a PCB using 1M×4DRAMs or 4M×4 DRAMs. Connectors

416

and

516

have signals corresponding to the signals shown on memory PCB

104

a

of FIG.

2

. The PCB of

FIG. 4

may have up to 128 Mbytes of memory, organized as a single row by eight columns of eight bits, plus parity. Other embodiments may use fewer chips or have less memory. The PCB of

FIG. 5

may have up to 16 Mbytes when populated with 1M×4 DRAMs and up to 64 Mbytes when populated with 4M×4 DRAMs. The 1(4)M×4 PCB of

FIG. 5

is organized as two banks of two rows by eight columns of eight bits, plus parity.

In

FIG. 4

, 16M×1 DRAMs

404

are MT4C16M1A1 DRAMs from Micron Technology Corp. Buffers 402 are 74AS244 buffers from Toshiba. High-density Connector

416

is described below. Other embodiments may use other similar components. In

FIG. 5

, the 1M×4 DRAMs are MT4C4001 DRAMs from Micron Technology. The 4M×4 DRAMs are MT4C4M4A1 DRAMs from Micron Technology. Buffers

502

are 74AS244 from Toshiba. High-density connector

516

is described below.

The memory arrangements shown support interleaved and non-interleaved memory configuration. Whether or not a memory interleaving scheme is used is determined by which type of CPU PCB used and the signals it provides to the memory cards. The memory cards are the same for either interleaved or non-interleaved memory configurations. Whether interleaving is used depends entirely on the microprocessor. Thus, the same memory PCBs may be used as interleaving and non-interleaving memory.

The 486-based CPU PCB uses a memory interleaving scheme described below. When, e.g., a 486-based CPU PCB is used (described below in connection with FIG.

8

), the system supports the 80486 processor family burst cycles to/from memory for the CPU and PCI memory access cycles. In this case, the EMA

0

and OMA

0

lines are generated by the CDC (

FIG. 8

) and are decoded from the HA

2

or A_D

2

lines. The EMA

0

and OMA

0

set up burst accesses for the odd and even memory banks, by changing state after the first and second access of a burst and setting up the correct column address for the third and fourth accesses of the burst. The memory PCBs receive these signals and use them as MA[0] for the respective memory banks.

A Pentium-based CPU, for example (see FIG.

3

), uses non-interleaved memory. In that case the Pentium CPU PCB connects EMA

0

and OMA

0

together (on the CPU PCB) and connects them to MA[0] on the connector interface.

Signals sent between the elements of

FIGS. 4 and 5

are shown in Table 3 below. The signals are essentially the same for both PCBs. Other embodiments may include any mixture of memory chips, with a combination of two 16M×1 RAMs yielding a largest memory for the embodiment shown.

TABLE 3

BACKPLANE INTERFACE

MD[0:31]

Memory Data Bus, low order bits: Bi-

(or MD 32:63

directional data lines for the memory

for PCB 104b

data bus. The signals drive data

connected to

destined for either the Host data bus

the Pentium)

or the PCI bus. It also inputs data

that originated from either the Host

data bus or the PCI bus.

DP[0:3]

Memory Parity: Bi-directional byte

enable parity signals for the memory

data bus. The low order parity bit

DP[0] corresponds to MD[0:7] (of the

data bus), while the high order parity

bit DP[3] corresponds to MD[24:31].

The DP[0:7] are parity outputs during

write cycles to memory and parity

inputs during read cycles from memory.

MA[1:11]

DRAM Multiplexed Address: Provide the

row and column address to the DRAM

array.

CAS[0:7]

Column Address Strobes: These signals

are used to latch the column address

on the MA[1:11] lines into the DRAMs.

Each signal corresponds to one byte of

the eight byte wide array.

RAS[0:3]

Row Address Strobes: These signals

are used to latch the row address on

the MA[1:11] lines into the DRAMs.

Each signal corresponds to one eight

byte wide DRAM row.

MRW-

DRAM Write Enable: This signal is

asserted during both CPU and PCI

master writes to main memory.

EMAO

Even Memory Address, bit 0.

ROMAO

Odd Memory Address, bit 0.

BANKSEL

Memory Bank Select.

BUFFERED/SELECT SIGNALS

RAS[0:1], A&B

Row Address Strobe: Buffered RAS[0:1]

to drive multiple memory devices.

MAO, A&B

Memory Address: Buffered EMAO and

OMAO or MA[0].

CASA[0:3]

Column Address Strobe: Buffered

CAS[0:3].

CASB[0:3]

Column Address Strobe: Buffered

CAS[4:7].

WEA&B

DRAM Write Enable: Buffered MRW-

MBA[1:11]

Multiplexed DRAM Address: Buffered

MA[1:11].

FIG. 6

is a backplane wiring diagram for the embodiment of

FIG. 2

showing how wiring differs for the respective memory PCBs of

FIGS. 4 and 5

. Specifically, in the described embodiment, pin

62

(RAS[0]M16) is not connected in the memory PCB when 1(4)M×4 memories are used. In contrast, pins

61

(RAS[0]M4) and

63

(RAS[1]M4) are not connected on the memory PCB when 16M×1 memories are used. Other embodiments may use other wiring schemes and pin assignments.

FIG. 7

is a block diagram of an example of peripheral controller PCB

108

of the embodiment of FIG.

2

. Controller PCB

108

contains all the necessary logic to interface the CPU PCB

102

with multiple peripheral busses. These peripheral busses provide interfaces to I/O devices such as printers, modems, mass storage devices, and video, etc. Some of these I/O devices may use one of various industry standard busses. Others may use various proprietary busses without departing from the spirit of the invention, as long as the peripheral controller includes means to convert local bus signals to control a desired peripheral or I/O device.

Controller PCB

108

acts as a bridge between local bus

110

and other peripheral busses. As shown in

FIG. 2

, controller PCB

108

acts as an interface between local bus

110

and EISA, SCSI, IDE, and floppy drive busses, as well as driving a parallel port, two serial ports, the keyboard, and mouse I/O. A similar controller PCB could be plugged into the same slot on the backplane instead of the PCB of

FIG. 7

to provide a bridge between local bus

110

and, e.g., the ISA or Microchannel busses, as well as other I/O busses. As shown in

FIG. 2

, in the described embodiment, controller PCB

108

controls two PCI cards and five EISA slots. Other controllers could use other numbers of slots.

FIG. 7

includes an EISA controller

702

; an X-bus decoder/buffer

704

; miscellaneous logic

706

, which includes a flash memory, BIOS store, and a real time clock; a PCI/EISA bridge

708

; a SCSI controller

710

; a port controller

712

; electronics to drive indicators for a keyboard, mouse, etc.

714

; a connector

716

; a PCI bus

724

; an EISA bus

722

; and miscellaneous busses

720

. Busses

720

,

722

, and

724

all connect through connector

716

, which has signals corresponding to the signals shown on controller PCB

108

of FIG.

2

.

EISA controller

702

is preferably an 82374EB EISA System Component (ESC) chip manufactured by Intel Corp. and described in “84420/82430 PCIset ISA and EISA Bridges,” Intel Corp.

1993

. PCI/EISA Bridge

708

is preferably an 82375EB PCI-EISA Bridge (PCEB) manufactured by Intel Corp. and described in “82420/82430 PCIset ISA and EISA Bridges,” Intel Corp. 1993, the entirety of which is herein incorporated by reference.

In the described embodiment, ESC and PCEB

702

and

708

also perform other functions, such as PCI parity and system error reporting, buffer coherency, management protocol, PCI and EISA memory and I/O address space mapping and decoding. PCEB

708

provides master/slave functions on both the PCI and EISA busses. Functioning as a bridge between the PCI and EISA busses, it provides address and data paths, bus controls, and bus protocol translations between the busses. PCEB

708

performs the functions of the PCI interface/arbiter, data swap logic, and the BIOS timer.

In the described embodiment, ESC

702

serves primarily as the EISA bus controller. It also integrates EISA compatible DMA controller, interrupt controller, timer/counter, and EISA arbitration logic. ESC

702

also provides control signals for the X-bus decode logic. The X-bus is decoded from the SD bus and serves as the interface to the BIOS store, keyboard, and additional indicators. Port controller

712

preferably is an 82C711 Port Controller chip, manufactured by Chips and Technologies, Inc., San Jose, Calif., and provides an interface between the system bus (SD) and the IDE, floppy, parallel, and serial ports.

The SCSI bus controller preferably is an NCR53C810 chip, manufactured by NCR. It is connected directly to the PCI bus and drives internal and external SCSI compatible peripherals.

In other embodiments, controller PCB

108

acts as a bridge between local bus

110

(e.g., the PCI bus) and an ISA bus. In this case, a bridge between PCI and ISA is implemented through use of an 82378IB System I/O chip, available from Intel Corp., described in 82420/82430 PCI set is a EISA Bridge herein incorporated by reference.

In

FIG. 7

, X-bus decoder

708

is preferably a 74F543 decoder from Toshiba. Flash memory is preferably a 28F512 from Intel. Keyboard mouse control is preferably an 87C42 from Chip and Technologies, Inc. High-density connector

716

is described below.

Signals sent between the elements of

FIG. 7

are shown in Table 4 below.

TABLE 4

PCI BUS BACKPLANE SIGNALS

PCI SIGNALS

As defined in the PCI specification.

AD[0:31]

Address and Data: Low order bi-

directional data lines for the PCI

bus. The signals sample or drive the

address and data on the PCI bus.

AD[32:63]

Address and Data: High order bi-

directional data lines for the PCI

bus. The signals sample or drive the

address and data on the PCI bus. These

signals cannot be used with a 32-bit

PCI bus.

GNT-[1:7]

Grant: When asserted, indicates that

access to the PCI bus has been granted

to a bus master by the PCI bus

arbiter. Each PCI bus master device

has a GNT signal.

REQ-[1:7]

Request: A PCI bus master device

asserts this signal to indicate to the

PCI bus arbiter that it is requesting

use of the PCI bus. Each PCI bus

master device has a REQ signal.

CONTROLLER BOARD INTERFACE SIGNALS

IRQ[1:15]

Interrupt Request Lines: These 15

interrupt inputs accept interrupt

requests from I/O or other devices.

ICC[0:3]

Interrupt Controller Communications

Bus: The ICC bus is used to pass

interrupt messages among I/O units and

multiple processors.

ADDITIONAL I/O PORTS

EISA BUS

Enhanced Industry Standard

Architecture Bus. There are five EISA

expansion slots.

SCSI BUS

Small Computer Systems Interface Bus.

The SCSI host adapter supports up to

eight peripherals, operating under

SCSI-2 protocols at 10 M bytes/s.

Refer to NCR 53C810 Data Manual.

IDE BUS

Integrated Drive Electronics Bus. As

defined in the Chips and Technologies,

Inc. 82C711 Data Book (San Jose,

California, 1993).

KEYBOARD

Standard 101-key enhanced keyboard

INTERFACE

interface.

MOUSE PORT

Microsoft mouse compatible 9-pin

interface.

PARALLEL PORT

IBM XT/AT compatible parallel port,

with bi-directional support.

SERIAL PORT #1

NS16450 compatible UART for serial

data transmission.

SERIAL PORT #2

NS16450 compatible UART for serial

data transmission.

FLOPPY DRIVE

Control signals for floppy disk drives

BUS

as per CHIPs 82C711 specification.

PCI/EISA Bridge and EISA Controller

PCEB/ESC Com.

These signals perform the following

functions:

-Arbitration and Interrupt

Acknowledge Control

-EISA/PCI Buffer Coherency

Control

-Data Swap Logic between the EISA

and the System Data bus

LA[2:23]

EISA Address bus: These signals are

connected directly to the EISA address

bus. These signals are used to decode

accesses to the EISA Controller's

internal resources. They also address

memory devices, such as a BIOS SRAM &

flash memory.

LA-[24:31]

EISA Address Bus/Configuration RAM

Page address: LA[27:31] are directly

connected to the EISA bus. During I/O

access to 0800h-08FFh, These signals

contain the configuration page address

of the BIOS SRAM.

SD[0:7]

System Data: These signals are

directly connected to the System Data

Bus.

SA[0:19]

Latched Address Bus.

XD[0:7]

Peripheral Data Bus: connect to SD

bus via transceivers, interface to

BIOS ROM, keyboard interface and mouse

interface.

Embodiments of the system of

FIG. 2

include two peripheral controllers PCB

108

. For example, a first peripheral controller PCB might control a plurality of a first type of peripheral slots, e.g., EISA slots, while a second type of peripheral controller PCB might control a second type of peripheral slots, e.g., ISA slots. Any combination of modular peripheral controller PCBs may be used that are compatible with the purpose of the computer system. Both PCBs

108

would preferably have identical connectors

716

, although not all signals on connector

716

would be used.

In

FIG. 2

, all peripherals slots use through-hole mounted connectors that are not high-density connectors. Other embodiments of the invention may use high-density connectors to connect the peripherals to the backplane. The signals to/from industry standard peripheral busses would remain the same if high-density connectors are used.

In general, the high-density connectors used in the present invention are surface mounted connectors. Other embodiments may, however, use through-hole mounted high-density connectors or high-density connectors mounted using other mounting techniques. Various mounting techniques are described below in connection with

FIGS. 22-36

.

FIG. 8

is a block diagram of an alternate embodiment of CPU PCB

102

of an alternate embodiment of FIG.

2

. CPU PCB

102

contains a 32-bit microprocessor

802

, which is an 80486DX2 device from Intel Corp.; an upgrade socket

810

for a microprocessor upgrade device, such as an accelerator chip (e.g., a 486 Overdrive chip, manufactured by Intel); a cache/DRAM controller (CDC)

804

; 512K bytes of cache RAM

808

; a data path unit (DPU)

812

; a Tag RAM

806

; a programmable interrupt controller (APIC)

814

; and additional logic and buffering devices that are not shown in the Figure for clarity of explanation.

In

FIG. 8

, processor

802

is a 80486DX2-66 processor, upgrade Processor socket

810

is a P24T socket, CDC

804

is a 82424TX CDC, DPU

812

is a 82423TX DPU, and APIC

814

is a 82489DX APIC, all of which are manufactured by Intel. Tag RAM

806

is a MCM670510 RAM from Motorola. Cache RAM

808

is a MCM620520 RAM from Motorola. High-density connector

816

is described below. Other embodiments may use other components.

As shown in

FIG. 8

, the CPU PCB

102

interfaces to the backplane via one or more high-density connectors

816

. CDC

804

and DPU

812

provide memory and second level cache control, as well as providing a bridge between an internal CPU bus

820

and local bus

110

(not shown). CDC

804

integrates cache

808

and memory

104

a

and

104

b

DRAM control functions and provides address paths and bus control for transfers between processor

102

, memory

104

a

,

104

b

and local bus

110

. CDC

804

has a dual-ported architecture that permits concurrent operations on both processor

102

and local bus

110

. CDC

804

also provides control signals for cache RAM

808

, Tag RAM

806

, and a “dirty-bit” SRAM (used by the cache) located in CDC

804

. Finally, CDC

804

provides support for a two-way interleaved DRAM organization.

DPU

312

provides a 32-bit data path connection between the host, memory

104

a

and

104

b

and local bus

110

. DPU

812

also has a dual-ported architecture to support concurrent host and local bus operations. DPU

812

supports byte parity for the host, memory

104

a

and

104

b

and local bus

110

. Signals sent between the elements of

FIG. 8

are shown in Table 5 below.

TABLE 5

486 CPU PCB

HOST BUS SIGNALS

HA[2:31]

Host Microprocessor Address Lines:

HA2-HA31, together with the byte en-

able signals, define the physical area

of memory or input/output space ac-

cessed. Address lines HA4-A21 are

used to drive addresses into the

microprocessor to perform cache line

invalidations.

HBE-[0:3]

Host Byte Enable: These signals indi-

cate active bytes during read and

write cycles.

HDP[0:3]

Host Data Parity: There is one bit

for every byte on the data bus. The

processor generates the data parity on

all write data cycles with the same

timing as the data driven by the mi-

croprocessor. The Data Path Unit

reads the parity bit from main memory

back into the processor.

HD[0:31]

Host Data Lines: They are the data

lines from the microprocessor.

COMM. SIG.

Communication Signals: These are ad-

ditional signals that handle bus con-

trol, bus cycle and arbitration, in-

terrupts and cache invalidations, etc.

CACHE/DRAM CONTROLLER

AD[0:31]

Address and Data: Low order bi-

directional data lines for the PCI

bus. The signals sample or drive the

address and data on the PCI bus.

AD[32:63]

Address and Data: High order bi-

directional data lines for the PCI

bus. The signals sample or drive the

address and data on the PCI bus.

These signals are not used with a 32-

bit PCI bus.

MA[1:11]

DRAM Multiplexed Address: Provide the

row and column address to the DRAM

array.

CAS[0:7]

Column Address Strobes: These signals

are issued to latch the column address

on the MA[1:11] lines into the DRAMs.

Each signal corresponds to one byte of

the eight byte wide array.

RAS[0:3]

Row Address Strobes: These signals

are used to latch the row address on

the MA[1:11] lines into the DRAMs.

Each signal corresponds to one eight

byte wide DRAM row.

MRW-

DRAM Write Enable: This signal is

asserted during both CPU and PCI mas-

ter writes to main memory.

EMAO

Even Memory Address: Generated by the

CDC and decoded from HA[2] for burst

cycle support.

ROMAO

Odd Memory Address: Generated by the

CDC and decoded from HA[2] for burst

cycle support.

BANKSEL

Memory Bank Select:

TA[0:7]

Tag RAM Addresses: For a cache size

of 512K byte these correspond to

HA(19:26). They are used by the CDC

to determine a cache hit/miss.

CA[4:18]

Latched Cache Address: Generated from

the processor address lines, active

when accessing the cache memory.

DATA PATH UNIT/APIC/INTERRUPT CONTROL SIGNALS

PCI SIGNALS

As defined in the PCI specification.

MD[0:31]

Memory Data Bus, low order bits: Bi-

directional data lines for the memory

data bus. The signals drive data des-

tined for either the Host data bus or

the PCI bus. It also inputs data that

originated from either the Host data

bus or the PCI bus.

DP[0:3]

Memory Parity: Bi-directional byte

enable parity signals for the memory

data bus. The low order parity bit

DP[0] corresponds to MD[0:7], while

the high order parity bit DP[3] cor-

responds to MD[24:31]. The DP[0:7]

are parity outputs during write cycles

to memory and parity inputs during

read cycles from memory.

IRQ[1:15]

Interrupt Request Lines: These 15

interrupt inputs accept interrupt re-

quests from I/O or other devices.

ICC[0:3]

Interrupt Controller Communications

Bus: The ICC bus is used to pass in-

terrupt messages among I/O units and

multiple processors.

Connector

316

has signals corresponding to the signals shown on CPU PCB

102

of FIG.

2

.

FIG. 9

is a functional block diagram of a 128-bit CPU PCB

102

that may also be placed in the embodiment of FIG.

2

.

FIG. 9

includes a CPU

902

, which is preferably an R4400 128-bit microprocessor, which is a RISC chip described in the R4400 User's Manual available from Toshiba, which is herein incorporated by reference. Processor

902

of

FIG. 9

has integrated cache control.

FIG. 9

also includes a secondary cache RAM

904

, an address path controller

906

, one or more data path controllers

908

, an address/local bus interface, and an interrupt controller

914

.

Processor

902

includes two separate data busses: a 128-bit bus

960

that connects directly to cache

904

and a 64-bit multiplexed System Add/Data bus

954

for interface to memory

104

and local bus

110

. The 64-bit System Add/Data bus

954

is translated by Address Path Controller

906

and Data Path Controller(s)

908

into X86, Intel compatible, separate address, data busses and control signals. Address/Local Bus Interface

910

provides the memory

104

control functions and bus control for the transfers between the CPU, memory

104

and the local bus. In

FIG. 9

, local bus

910

is preferably a PCI bus.

Data Path Controller(s)

908

and

912

each support a 64-bit path to memory

104

. In addition, Data Path Controller(s)

908

generate a multiplexed local bus address and data. If only one Data Path Controller

908

is used the system and Data Path Controller

912

is not used, the system has a 64-bit memory bus is compatible with the embodiment of FIG.

2

. When two Data Path Controllers

908

and

912

are used, memory is accessed via a 128-bit bus and can be connected to the 128-bit memory shown in FIG.

10

.

B. Examples of PCBs and Backplane Arrangements

FIG. 10

is a block diagram showing an arrangement of memory boards connected to a CPU slot supporting a 128-bit processor. For example, the 128-bit CPU of

FIG. 9

connects to memory PCBs as shown in FIG.

10

. Thus,

FIG. 10

shows memory PCBs used in an embodiment other than that shown in FIG.

2

. In

FIG. 10

, a high-density connector (not shown) connects CPU PCB

1002

to a backplane. Similarly a high-density connector (not shown) connects the memory PCBs

1004

to a backplane. The connector for

1002

differs from connector

316

of

FIG. 3

because it contains a wider data bus.

FIG. 11

shows an embodiment of the present invention incorporating a single backplane

1102

. In the Figure, four PCBs

102

,

104

a

,

104

b

, and

108

are mounted on the backplane

1102

using surface mount high-density connectors

1104

-

1114

, while seven PCBs

107

and

109

are mounted on the backplane

1002

using through-hole connectors

1108

. In other embodiments, one or more of boards

107

and

109

may also be mounted using high-density connectors. In this example, the PCBs correspond to the PCBs of

FIG. 2

that have similar reference numbers. Other embodiments may have different numbers and/or types of PCBs mounted on backplane

1102

. As a general rule, mixing surface mount and through-hole technology on a single backplane makes the manufacturing process more complex than if only one technique is used. In

FIG. 11

, the PCBs (EISA/ISA cards and PCI cards) are mounted using conventional through-hole connectors

1108

. The surface mount connectors

1104

,

1106

,

1110

,

1112

, and

1114

(plus other connectors not visible in the Figure), however, are a high-density connector described in detail later in this application.

FIG. 12

shows another view of the backplane of

FIG. 11

with the PCBs removed so that all the connectors are visible. All of the connectors

1104

,

1106

,

1110

,

1112

, and

1114

in

FIG. 12

are high-density connectors. Some of the high-density connectors are arranged in pairs to form contacts with both sides of a PCB. For example connectors

1106

and

1107

contact both sides of PCB

102

when PCB

102

is mounted on backplane

1102

. In contrast, connector

1120

uses a card interface of 40 contacts per linear inch.

FIG. 13

shows an embodiment of the present invention incorporating multiple backplanes

1302

and

1304

connected by right angle high-density connectors

1306

. In this arrangement, the components mounted via different techniques can be assembled separately, thus making the manufacturing process more efficient. In addition, the arrangement of

FIG. 13

makes the backplanes more modular. For example, it is easier to remove and replace all the EISA card slots with ISA, Microchannel, PCI card slots or some other desired card slots by simply switching backplane

1304

.

FIG. 14

shows a different view of the embodiment of FIG.

13

. Specifically,

FIG. 14

makes it clear that all the PCBs on backplane

1302

are connected to right angle high-density connectors

1410

that are attached to the backplane using surface mount technology. In contrast, all PCBs on backplane

1304

are mounted using conventional through-hole connectors

1412

. Backplanes

1302

and

1304

are connected by two or more high-density connectors

1306

and

1406

—at least one high-density connector for each side of the backplanes.

FIG. 15

shows an embodiment of the present invention incorporating multiple backplanes

1502

and

1504

connected by a right angle high-density connector

1506

, where backplane

1502

is double sided. Again, the PCBs are mounted on backplane

1504

via through-hole connectors and are mounted on backplane

1502

via surface mount high-density connectors. The PCBs can be connected one to another by use of a high-density connector through backplane

1502

.

FIG. 16

shows an embodiment of the present invention incorporating multiple backplanes

1602

and

1604

connected by a vertical high-density connector

1606

. Surface mounted connectors and through-hole mounted connectors can be located on one side or on both sides of the backplanes, and PCBs can plug into both types of connectors. Each backplane has both types of mounting technology. Such an arrangement makes it easier to replace an entire backplane while making field upgrades, for example.

FIG. 17

shows an embodiment of the present invention incorporating multiple backplanes

1702

and

1704

connected by a vertical high-density connector

1706

. In all the described embodiments, a high-density connector between backplanes or for mounting a PCB can also be configured as several smaller high-density connectors.

FIG. 18

shows an embodiment of the present invention incorporating multiple backplanes

1802

and

1804

connected by a vertical high-density connector

1806

. In the embodiment of the Figure, all surface mounted connectors are on backplane

1804

while all through-hole mounts are on backplane

1802

.

FIG. 19

shows a side view of an embodiment of the present invention wherein PCBs

1908

are connected to a double sided backplane

1902

using high-density surface mount connectors

1906

. Each PCB

1908

has many traces

1904

connecting it to other PCBs on both sides of the backplane. Use of surface mounted connectors makes it easier for a designer to position the connectors so as to balance the impedances of the traces between boards. The trace lengths (stubs) can be shorter than in, e.g., a double sided backplane using through-hole connectors.

FIG. 20

shows an impedance loading of the embodiment of FIG.

19

. If a signal source is to see a balanced load, the impedance loading of the traces and boards should be made to match (Zt1=Zb1=Zt2).

FIG. 21

is a diagram showing an arrangement of circular backplanes

2102

having surface mounted PCBs, e.g., PCBs

2106

,

2108

,

2110

, thereon. This arrangement maximizes cooling, yet allows large components to be placed on the outer edges of the PCBs. In addition, signals at the center of the circular backplane can traverse the circle in less time than signals at the outer part of the circle. Thus, more time-critical elements are placed towards the inside of the circle.

FIG. 21

shows vertically placed cleated PCBs

2110

and horizontally placed cleated PCBs

2111

. The orientation of PCBs

2111

makes PCBs

2111

especially easy to cool.

C. Description of a High-density Connector/Connector System

FIGS. 22-38

show a high-density connector (or a connector system) used to connect PCBs and backplanes of the present invention. Similar connectors may be used to connect components to the PCBs, and indeed may be used anywhere a connector is required. The high-density connector is described in co-pending U.S. patent application Ser. No. 07/983,083 to S. Crane filed an Dec. 1, 1992, which is herein expressly incorporated by reference. The high-density connector is also described in U.S. patent application Ser. No. 08/381,142 to S. Crane filed on Jan. 31, 1995, entitled “High-density Electrical Interconnect System,” and U.S. patent application Ser. No. 08/208,519 to S. Crane filed on Mar. 11, 1994, entitled “Apparatus Having Inner Layers Supporting Surface-Mount Components,” each of which are herein expressly incorporated by reference.

1. The Projection Type Interconnect Component

FIGS. 22 and 23

are a diagram of a projection component

2210

of a high-density connector that connects the PCBs and/or backplanes of the present invention.

FIGS. 35-38

, for example, discuss a “high-density connector system” where two high-density connectors are mounted on respective sides of a PCB using surface mount technology. The high-density connector system used in one preferred embodiment of the present invention has a density of at least 100 contacts per linear inch, as discussed below in connection with

FIGS. 35 and 37

. The high-density connector system used in another preferred embodiment of the present invention has a density of at least 128 contacts per linear inch, as discussed below in connection with FIG.

36

. The high-density connector system used in another preferred embodiment of the present invention has a density of at least 300 contacts per linear inch, as discussed below in connection with FIG.

38

. Other embodiments of the present invention may have high-density connectors of different dimensions and/or densities. In general, pairs of connectors mounted on opposite sides of a PCB are referred to as “connector systems.” When the meaning seems clear from context, connector systems are sometimes referred to as simply “connectors.” The terms “connector” and/or “connector system” generally refer to an unmated male portion (projecting) or a female portion (receiving). The term “connector” may also be used as a shorthand way to refer to mated male and female connector pairs.

The connector comprises a projection-type interconnect component and includes at least one conductive post

2211

. In

FIGS. 22 and 23

, reference numeral

2217

designates the contact portion of each conductive post

2211

; reference numeral

2318

designates a stabilizing portion of each conductive post; and reference numeral

2319

designates a foot portion of each conductive post. When the projection-type interconnect component

2210

is received within a corresponding receiving-type interconnect component, electrical signals may be transferred from the foot portion

2319

of each conductive post

2211

through the stabilizing and contact portions of that post to the receiving-type interconnect component, and vice versa.

Each conductive post

2211

may be formed of beryllium copper, phosphor bronze, brass, a copper alloy, tin, gold, palladium, or any other suitable metal or conductive material. In a preferred embodiment, each conductive post

2211

is formed of beryllium copper, phosphor bronze, brass, or a copper alloy, and plated with tin, gold, palladium, nickel, or a combination including at least two of tin, gold, nickel, and palladium. The entire surface of each post may be plated, or just a selected portion

2216

corresponding to the portion of conductive post

2211

that will contact a conductive beam when the projection-type interconnect component is received within the corresponding receiving-type interconnect component. The projection-type interconnect component

2210

may also include an insulative buttress

2212

, although use of a buttress is not required. The conductive posts and the buttress (when used) are attached to an insulative substrate

2213

. The conductive posts are electrically isolated from one another by the substrate

2213

and the buttress

2212

(when used).

FIG. 23

is a side view of two conductive projection-type interconnect components. The buttress

2212

and the substrate

2213

may be integrally molded from a single unit of insulative material. Preferably, the material of the buttress and the substrate is an insulative material that does not shrink when molded (for example, a liquid crystal polymer such as VECTRA, which is a trademark of Hoescht Celanese). The conductive posts

2211

are inserted into the substrate

2213

through holes in the substrate represented by the dotted lines in FIG.

23

.

As seen from

FIG. 23

, the buttress

2212

includes an elongated portion

2314

having a rectangular (e.g., square) cross-section, and a tip portion

2315

located at the top of the elongated portion. The buttress dimensions shown in

FIG. 23

are exemplary and, accordingly, various dimensions for buttress

2212

may be used. For example, the cross-section of the buttress

12

may be 0.5 mm×0.5 mm rather than the illustrated dimensions of 0.9 mm×0.9 mm.

Each conductive post

2211

includes three sections: a contact portion, a stabilizing portion, and a foot portion. In

FIG. 23

, the contact portion of each conductive post is shown in a position adjacent the buttress

2212

. The stabilizing portion

2318

(not shown in

FIG. 22

) is the portion of each post that is secured to the substrate

2213

. The foot portion

2319

(not shown in

FIG. 22

) extends from the side of the substrate opposite the contact portion. The conductive posts may have a rectangular (e.g., square) cross-section, or a cross-section that is triangular, semicircular, or some other shape.

The different portions of each conductive post

2211

each perform a different function. The contact portion

2317

establishes contact with a conductive beam of a receiving-type interconnect component when projection-type and receiving-type interconnect components are mated. The stabilizing portion

2318

secures the conductive post to the substrate

2213

during handling, mating, and manufacturing. The stabilizing portion

2318

is of a dimension that locks the post into the substrate

2213

while allowing an adequate portion of the insulative substrate to exist between adjacent conductive posts. The foot portion

2319

connects to an interface device (e.g., a semiconductor chip, a PCB, a backplane, a wire, or a round, flat, or flex cable) using the electrical interconnect system as an interface. The contact and foot portions may be aligned or offset with respect to the stabilizing portion to provide advantages that will be discussed below.

The configuration of the foot portion

2319

of each conductive post

2211

depends on the type of device with which that foot portion is interfacing. For example, the foot portion

2319

will have a cylindrical configuration if interfacing with a through-hole of a printed wiring board. The foot portion

2319

will be configured as in

FIG. 23

if interfacing with a printed wiring board through a surface mount process. If interfacing with a round cable or wire, the foot portion

2319

may be configured to encircle the cable or wire. Other configurations may be used depending on the type of device with which the foot portion

2319

is interfacing.

FIG. 24

shows a foot portion

2319

of a conductive post configured for surface mounting on a PCB or backplane. As shown in

FIG. 24

, the substrate

2213

may be positioned at a right-angle with respect to the PCB

2420

. This positioning increases space efficiency and can facilitate cooling of the components on the PCB and/or shorten various signal paths. Although not explicitly shown in

FIG. 24

, the substrate

2213

may be positioned at a right-angle with respect to the device with which the foot portion is interfacing (e.g., a PCB or a cable) regardless of the nature of the device. As seen from

FIG. 24

, such positioning necessitates the bending of the foot portion

2319

at a right-angle at a point

2421

of the foot portion. The bend at point

2421

and/or the bend of the foot portion

2319

near the PCB

2420

may be sharp, as depicted in

FIG. 24

, or one or both of each bends could be gradual or curved.

With reference to

FIG. 24

, each foot portion

2319

extends cut from a horizontal surface of substrate

2213

, and then bends toward the surface of the interface device at a point

2421

of that foot portion. The foot portions

2319

in the example are bent such that the foot portions contact the interface device in three separate rows (i.e., rows C, D, and E of FIG.

24

).

In various embodiments of the present invention, either the post or the beam portion of the connector can be located on a connector placed at right angles to the PCB of the backplane. Other embodiments are contemplated and described in the original application.

2. The Receiving-Type Interconnect Component

The receiving-type electrical interconnect component of the present invention used to connect a PCB to a backplane includes several electrically conductive beams attached to an insulative substrate. The receiving-type electrical interconnect component is configured to receive a corresponding projection-type electrical interconnect component within a space between the conductive beams. The substrate insulates the conductive beams from one another so that a different electrical signal may be transmitted on each beam.

FIG. 25

illustrates a portion of a receiving-type interconnect component

2530

in accordance with an embodiment of the present invention. The receiving-type component

2530

comprises several electrically conductive, flexible beams

2531

attached to an electrically insulated substrate (not shown in FIG.

25

). Preferably, the material of the substrate is an insulative material that does not shrink when molded (for example, a liquid crystal polymer such as VECTRA, which is a trademark of Hoescht Celanese). Portions of the conductive beams

2531

bend away from each other to receive the projection-type interconnect component within the space between the conductive beams.

Each conductive beam

2531

may be formed from the same materials used to make the conductive posts

2211

of the projection-type electrical interconnect component. For example, each conductive beam

2531

may be formed of beryllium copper, phosphor bronze, brass, or a copper alloy, and plated with tin, gold, or palladium at a selected portion of the conductive beam which will contact a conductive post of the projection-type interconnect component when the projection-type interconnect component is received within the receiving-type interconnect component

2530

.

A conductive beam

2531

that may be used in the electrical interconnect system of the present invention includes three sections: a contact portion

2532

; a stabilizing portion

2533

; and a foot portion

2534

.

The contact portion

2532

of each conductive beam

2531

contacts a conductive post of a corresponding projection-type receiving component when the projection-type receiving component is received within the corresponding receiving-type interconnect component. The contact portion

2532

of each conductive beam includes an interface portion

2535

and a lead-in portion

2536

. The interface portion

2535

is the portion of the conductive portion

2532

which contacts a conductive post when the projection-type and receiving-type interconnect components are mated. The lead-in portion

2536

comprises a sloped surface which initiates separation of the conductive beams during mating upon coming into contact with the tip portion of the buttress of the projection-type interconnect component (or, when a buttress is not used, upon coming into contact with one or more posts of the projection-type interconnect component).

The stabilizing portion

2533

is secured to the substrate that supports the conductive beam

2531

. The stabilizing portion

2533

of each conductive beam prevents that beam from twisting or being dislodged during handling, mating, and manufacturing. The stabilizing portion

2533

is of a dimension that locks the beam into the substrate while allowing an adequate portion of the insulative substrate to exist between adjacent conductive beams.

The foot portion

2534

is very similar to the foot portion

2319

of the conductive post

2211

described above in connection with the projection-type interconnect component

2210

. Like foot portion

2319

, the foot portion

2534

connects to an interface device (e.g., a semiconductor chip, a printed wiring board, a wire, or a round, flat, or flex cable) which uses the electrical interconnect system as an interface.

In the same manner as foot portion

2319

, the configuration of the foot portion

2534

depends on the type of device with which it is interfacing. Possible configurations of the foot portion

2534

are the same as the possible configurations discussed above in connection with the foot portion

2319

above.

Like foot portion

2319

, the foot portion

2534

will be bent at a right-angle in situations where the substrate of the receiving-type interconnect component is located at a right-angle with respect to the interface device with which the foot portion

2534

is interfacing. The contact and foot portions of each conductive beam may be aligned or offset with respect to the stabilizing portion to provide advantages that will be discussed in detail below.

3. Mating of the Interconnect Components

FIG. 26

shows the receiving-type interconnect component

2530

in the mated condition. When the projection-type and receiving-type interconnect components are mated, the contact portions

2532

of the conductive beams bend or spread apart to receive the projection-type interconnect component within the space between the contact portions of the conductive beams. In other embodiments, the contact portion

2532

for two of the beams is longer than the contact portion for the other two beams.

It should be noted that the configuration of the receiving-type component depends on the configuration of the projection-type interconnect component, or vice versa. For example, if the projection-type interconnect component comprises a cross-shaped buttress surrounded by conductive posts, then the receiving-type component should be configured to receive that type of projection-type interconnect component. Other embodiments are described in U.S. application Ser. No. 07/983,083 to S. Crane filed on Dec. 1, 1992.

FIG. 26

shows a projection-type interconnect component

2210

received within the conductive beams of a receiving-type interconnect component

2530

. When the projection-type interconnect component is received within the receiving-type interconnect component in this fashion, such interconnect components are said to be mated or plugged together.

The mated position shown in

FIG. 26

is achieved by moving the projection-type interconnect component

2210

and the receiving-type interconnect component

2530

toward one another in the direction or arrow Y shown in FIG.

26

. In the mated position, the contact portion of each conductive beam exerts a normal force against a contact portion of a corresponding one of the conductive posts in a direction within plane XZ. In

FIG. 26

, arrow Y is perpendicular with respect to plane XZ.

The process of mating a projection-type interconnect component

2210

with a corresponding receiving-type interconnect component

2530

will now be discussed.

FIGS. 22-25

show the state of the projection-type interconnect component

2210

and the corresponding receiving-type interconnect component

2530

prior to mating. The contact portions

2532

of the beams of the receiving-type interconnect component are clustered together before mating with the projection-type interconnect component. Such clustering may involve contact between two or more of the beams.

Next, the projection-type and receiving-type interconnect components are moved toward one another in the direction of the arrow Y shown in FIG.

26

. Eventually, the lead-in portions

2536

of each conductive beam

2531

contact the tip portion of the buttress

2212

(when used). Upon further relative movement of the interconnect components toward one another, the sloped configuration of the tip portion causes the contact portions

2532

of the conductive beams to start to spread apart. Further spreading of the contact portions

2532

occurs with additional relative movement between the interconnect components due to the sloped upper surfaces of the conductive posts

2211

of the projection-type component. Such spreading causes the conductive beams

2531

to exert a normal force against the conductive posts

2211

in the fully mated position, thereby ensuring reliable electrical contact between the beams and posts. It should be noted that when a buttress is not used, the initial spreading of the contact portions

2532

is caused by one or more posts

2211

of the projection-type interconnect component rather than a buttress tip portion.

The insertion force required to mate the projection-type interconnect

2210

within the receiving-type interconnect component

2530

is highest at the point corresponding to the early phases of spreading of the conductive beams

2531

. The subsequent insertion force is less as it relates to frictional forces rather than spreading forces. The insertion force required to mate the projection-type and receiving-type interconnect components can be reduced (and programmed mating, wherein one or more interconnections are completed before one or more other interconnections, may be provided) using a projection-type interconnect component having conductive posts which vary in height.

In another embodiment, conductive posts

2211

can be arranged so that one pair of opposing posts has a first height, and the other pair of opposing posts has a second height. In essence, this configuration breaks the peak of the initial insertion force into separate components occurring at different times so that the required insertion force is spread out incrementally over time as the mating process is carried out.

In another embodiment, the required insertion force can be spread out over time as mating occurs (and in which programmed mating can be provided). Different rows of projection-type interconnect components

2210

can have different heights so that mating is initiated for different rows of the interconnect components at different times. The rows may can be alternately high and low in height, for example, or the height of the rows can increase progressively with each row. Also, the components within a given row may have different heights. Further, various embodiments may be combined to achieve an embodiment wherein different rows of interconnect components vary in height, and the conductive posts of each interconnect component within the different rows also vary in height. Also, the conductive beams

2531

or the contact portions

2532

of each receiving-type interconnect component could vary in length to similarly reduce the insertion force or provide programmed mating.

The action of sliding down the side of the post of the conductive beams

2531

during mating performs a wiping function to wipe away debris and other contaminants that may be present on the surfaces of the posts

2211

, the buttress

2212

(if used), and the beams

2531

. Such wiping allows for more reliable electrical interconnection and the provision of a greater contact area between mated conductive elements.

The insertion force can essentially be entirely eliminated or reduced using a zero-insertion-force receiving-type or a low-insertion-force type interconnect component as described in U.S. application Ser. No. 07/983,083 to S. Crane filed on Dec. 1, 1992.

4. The Insulative Substrates

As explained above, the conductive posts of the projection-type interconnect component are attached to an insulative substrate

2213

. Likewise, the conductive beams of the receiving-type component are attached to an insulative substrate

2537

.

FIGS. 27 and 28

show an example of an insulative electrical carrier functioning as the substrate

2213

for the projection-type interconnect component

2210

and an insulative electrical carrier functioning as the substrate

2537

for the receiving-type interconnect component

2530

. The carrier

2213

in

FIG. 28

is arranged so that a right-angle connection may be made using the foot portions of the projection-type interconnect component

2210

. The carrier

2537

in

FIG. 28

, as well as the carriers in

FIG. 27

, flare arranged for straight rather than right-angle connections. Any carrier in

FIG. 27

or

FIG. 28

could be a right-angle or a straight carrier. The connector

2213

of

FIG. 28

are called a “right angle connector” and the connectors of

FIG. 27

is called “vertical connectors.” Either or both of the insertion or the projection component can have a right angle connector.

When used for surface mounting to a PCB or backplane, for example, the foot portion of each post and/or beam that is being surface mounted should extend beyond the furthest extending portion of the substrate by approximately 0.15 mm. This compensates for inconsistencies on the PCB or backplane, and makes the electrical interconnect system more flexible and compliant.

In some embodiments, the connectors of

FIGS. 27 and 28

are polarized so that the chance of backward mating is eliminated. Other embodiments use keying to differentiate two connectors having the same contact count.

5. The Interconnect Arrangement

The present invention holds a distinct advantage over conventional electrical interconnect systems because the interconnect components of the present invention can be arranged in a nested or other modified configurations far more dense than typical grid arrays or connector arrangements. Such configurations are not contemplated by existing conventional electrical interconnect systems.

The present invention is capable of providing much higher densities than conventional connectors. Instead of using a grid or rows of individual posts for connecting to respective individual sockets, the electrical interconnect system of the present invention arranges a plurality of conductive posts into groups (or “clusters”), with the groups being interleaved among one another for receipt of each group within a respective receiving-type interconnect component. Like the conductive posts, the conductive beams are also arranged into groups (or “clusters”), with the groups being interleaved among one another each for receiving a respective projection-type interconnect component. Thus, while conventional interconnect systems function by interconnecting individual pins with individual sockets, the present invention increases density and flexibility by interconnecting individual projection-type interconnect components including groups of posts with individual receiving-type interconnect components including groups of beams, in the most efficient manner possible.

FIG. 29

depicts an arrangement of groups of holes or passages in accordance with the present invention. In accordance with the arrangement of

FIG. 29

, the groups of holes or passages are formed in an insulated substrate

2213

. A conductive post

2211

is fitted within each of the passages to form an array of projection-type interconnect components or, alternatively, a conductive beam is fitted into each of the passages to form an array of receiving-type interconnect components. This arrangement can be used for either vertical or horizontal connections.

Herein, reference numeral

2982

will be used to refer to each group of contacts forming an interconnect component or, more generically, to the interconnect component including the group of contacts. Thus, each nested interconnect component

2982

referred to herein may be a projection-type interconnect component

2210

including a plurality of conductive posts

2211

or, alternatively, receiving-type interconnect component

2530

including a plurality of conductive beams

2531

or, alternatively, a hybrid interconnect component including a plurality of conductive posts

2211

and a plurality of conductive beams

2531

.

If the electrical interconnect components

2982

are projection-type interconnect components, each of the interconnect components

2982

is configured for receipt within a corresponding receiving-type interconnect component. Furthermore, the conductive contacts of each interconnect component are arranged such that the contacts of each interconnect component may be interleaved or nested within the contacts of other ones of the interconnect components. In other words, the conductive contacts of the array are arranged so that portions of each group

2982

overlap into columns and rows of adjacent groups of contacts to achieve the highest possible density while providing adequate clearance for the mating beams of the receiving-type interconnect components used. It should be noted that while each group of contacts or electrical interconnect component

2982

of

FIG. 29

, when such components are projection-type interconnect components or hybrid interconnect components, may have a buttress

2212

located at a central portion of that interconnect component, either in contact with the conductive contact or not in contact with the conductive contacts, one or more (e.g., all) of the interconnect components may be without a buttress. When the electrical interconnect components are receiving-type interconnect components, such components do not include a buttress.

As shown in

FIG. 29

, each group of contacts

2982

forming an interconnect component may be arranged in the shape of a cross. An arrangement such as that shown in

FIG. 29

(or

FIG. 31

) may be used to connect, e.g., backplanes. Connectors used to connect PCBs use a nested or modified configuration similar to that taught in

FIG. 29

or

31

, but will usually contain fewer rows and/or columns of clusters. For example, a connector connecting two PCBs or a connector connecting a PCB to a backplane may have only two rows of clusters.

Any other shapes that can easily be nested may also be used. The grouping of contacts into the shape of a cross (as in

FIG. 29

) aids in balancing beam stresses to keep the conductive beams of each receiving-type interconnect component or hybrid interconnect component, i.e., a component having both posts and beams, from being overly stressed. Further, the use of cross-shaped groups results in alignment advantages not found in conventional systems. For example, the cross-shaped interconnect components shown in

FIG. 29

, when the electrical interconnect components

2982

are projection-type interconnect components each align with the beams of a corresponding receiving-type interconnect component, causing the whole arrangement of

FIG. 29

to be similarly aligned.

The nesting of groups (e.g., cross-shaped groups) of holes or contacts (i.e., the nesting of projection-type, receiving-type, or hybrid interconnect components) allows adequate clearance between the contacts for mating with corresponding interconnect components, while decreasing to a minimum the space between the contacts. No conventional system known to the inventor utilizes space in this manner. Furthermore, as explained above, when the electrical interconnect components

2982

are projection-type interconnect components or hybrid interconnect components, the inclusion of a buttress between the contacts of each electrical interconnect component

2982

is optional. In the absence of a buttress, each group of posts

2211

for each projection-type interconnect component or hybrid interconnect component is capable of spreading corresponding conductive beams of corresponding interconnect components during mating due to the sloped upper surfaces of the posts.

It should be noted that the nested configuration of

FIG. 29

eliminates the need for providing insulative walls between the contacts, although such insulative walls may be used if desired. It should also be noted that although the nested configuration of

FIG. 29

may be an arrangement for the posts

2211

of projection-type interconnect components in an electrical interconnect system, the nested configuration of

FIG. 29

could also be the arrangement for the beams

2531

of the receiving-type interconnect components for that system. For example, for both the projection-type and receiving-type interconnect components within a given electrical interconnect system, the contacts of such components could be arranged so that portions of each group of contacts associated with an electrical interconnect component overlap into columns and rows of adjacent groups of contacts associated with other electrical interconnect components. In other words, both the projection-type and receiving-type components within a given electrical interconnect system may be arranged in a nested configuration. This also applies to electrical interconnect systems incorporating hybrid electrical interconnect components. Furthermore, by arranging the contacts into groups (e.g., the cross-shaped groups

2982

of FIG.

29

), the foot portions of the interconnect components for each group may be arranged to enhance the layout and trace routing of the interface devices (e.g., PCBs or backplanes) being interconnected.

The density of the interconnect arrangement of

FIG. 29

, when the electrical interconnect components

2982

are projection-type interconnect components or interconnect components each including buttress, depends on the configuration of the posts and beams, the spacing between buttresses, and the size of the buttresses used.

An arrangement wherein each buttress is 0.5 mm×0.5 mm is shown in FIG.

30

. Even higher densities may be achieved when a buttress is not used. For the arrangement of

FIG. 29

, when a 0.9 mm×0.9 mm buttress is used, a center-line to center-line distance X between columns of electrical interconnect components may be 1.5 mm; a center-line to center-line distance Y between rows of electrical interconnect components may be 1.25 mm; and the overall density for the arrangement may be 680 contacts per square inch. When a 0.5 mm×0.5 mm buttress is used, a center-line to center-line distance X between columns of electrical interconnect components may be 1.0 mm; a center-line to center-line distance Y between rows of electrical interconnect components may be 1.5 mm; and the overall density for the arrangement may be 828 contacts per square inch. When no buttress is used, a center-line to center-line distance X between columns of electrical interconnect components in a row may be 0.9 mm; a center-line to center-line distance Y between rows of electrical interconnect components may be 1.25 mm; and the overall density for the arrangement may be 1,028 contacts per square inch.

In the nested arrangement depicted in

FIG. 29

, the electrical interconnect components

2982

, whether of the projection-type, the receiving-type, or some other type, are arranged in rows and columns on the insulative substrate

2913

(the dotted lines in

FIG. 29

designate a row and a column, respectively); the electrical interconnect components of adjacent rows of the arrangement are staggered as are the electrical interconnect components from adjacent columns of the arrangement; and the electrical interconnect components are interleaved among one another in a nested configuration such that a portion of each electrical interconnect component overlaps into an adjacent row of the electrical interconnect components or an adjacent column of the electrical interconnect components. The projection-type, receiving-type, and/or components within a given electrical interconnect system may all be arranged in accordance with the nested arrangement depicted in FIG.

29

.

The nested configuration of

FIG. 29

can be modified to provide even greater densities. An example of one contemplated modification is depicted in FIG.

31

. In the arrangement of

FIG. 31

, the electrical interconnect components

2982

, whether of the projection-type, the receiving-type, or the hybrid-type, are arranged in rows and columns on the insulative substrate

2913

; and at least one contact (e.g., a post

2911

in

FIG. 31

) of each electrical interconnect component

2982

includes a front surface

2983

facing outwardly and away from that interconnect component along a line initially intersected by a side surface

2984

of a contact from another electrical interconnect component of the arrangement. It should be noted that, as with the nested arrangement depicted in

FIG. 29

, the arrangement in

FIG. 31

uses cross-shaped groups of contacts for the electrical interconnect components, although other shapes are contemplated. Also, all electrical interconnect components within a given electrical interconnect system (e.g., both the projection-type and receiving-type interconnect components in a pluggable system) may be arranged in accordance with the arrangement depicted in FIG.

31

.

FIG. 32

shows a portion of the arrangement in accordance with

FIG. 31

using buttresses that have a cross-section of 0.5 mm×0.5 mm. As seen from

FIG. 33

, when the projection-type electrical interconnect components

2982

from

FIG. 31

are each received within a corresponding receiving-type interconnect component

2530

, the conductive contacts or beams

2531

of the receiving-type interconnect components are separated by a distance of 0.2 mm, for example.

FIG. 34

is a view of projection-type electrical interconnect components

2210

arranged in accordance with the arrangement of FIG.

31

and received within corresponding receiving-type interconnect components

2530

. In

FIG. 34

, the buttresses

2212

for the projection-type interconnect components

2210

have a cross-section of 0.9 mm×0.9 mm. The distance between each conductive contact or beam

2531

and the contact which it faces is 0.4 mm, for example.

It should be noted that for the arrangement of

FIG. 31

, when a 0.9 mm×0.9 mm buttress is used, the distance d between like surfaces of the contacts may be 2.19 mm; and the overall density for the arrangement may be 460 contacts per square inch. When a 0.5 mm×0.5 mm buttress is used, the distance d may be 1.60 mm; and the overall density for the arrangement may be 900 contacts per square inch. When no buttress is used, the distance d may be 1.5 mm; and the overall density for the arrangement may be 1,155 contacts per square inch.

FIG. 35

is a diagram of a high-density projection-type connector used in an embodiment of the present invention. Receiving type connectors and/or connector systems can also use the arrangement shown or any similar arrangement. The orientation of clusters in the rows of

FIG. 35

is similar to that of

FIG. 30

, but also could be similar to that of FIG.

32

.

When a connector such as the connector in

FIG. 35

is surface mounted on each side of a PCB, the resulting connector system has a density of approximately 100 contacts per linear inch ((25.4 Tm in one inch/4 mm between center of clusters)×2 rows×4 posts×2 sides of the PCB=approximately 100). If, for example, a third row was added to each of the two connectors in the described connector system, the density would increase to approximately 152 contacts per linear inch. In general, several factors affect how closely the clusters can be placed on the connector. For instance, cluster spacing is affected by how closely holes can be formed in the insulator for the post and beam portions (i.e., the projection and insertion portions). The cluster spacing must also allow room for the beams to spread upon mating and yet not touch each other. In addition, the cluster spacing must allow room to mount traces on the PCB.

FIG. 36

is a diagram of another projection type high-density connector used in an embodiment of the present invention. Receiving type connectors and/or connector systems can also use the nested or modified arrangement shown, or any similar arrangement. The orientation of clusters in the rows of

FIG. 36

is similar to that of

FIG. 30

, but also could be similar to that of FIG.

32

. When a connector such as the connector in

FIG. 36

is surface mounted on each side of a PCB (see FIG.

37

), the resulting connector system has a density of approximately 128 contacts per linear inch ((25.4 mm per inch/3 mm between center of clusters)×2 rows×4 posts×2 sides of the PCB=approximately 128). If, for example, a third row was added to each of the two connectors in the described connector system, the density would increase to approximately 208 contacts per linear inch.

FIG. 37

shows a projection-type high-density connector system of used in an embodiment of the present invention. Receiving type high-density connector systems can also have the arrangement shown. In

FIG. 37

, two high-density connectors

3712

and

3714

are surface mounted on a PCB

3716

.

FIG. 37

is not shown to scale. The arrangement in

FIG. 37

is a right-angle connector but could easily be adapted for use in a vertical connector. The nesting arrangement of, e.g.,

FIG. 35

could also be used, which would have a density of approximately 100 contacts per linear inch.

FIG. 38

shows a projection type high-density connector system of used in an embodiment of the present invention. Receiving type high-density connector systems can also have the arrangement shown. In

FIG. 38

, two high-density connectors

3812

and

3814

are surface mounted on a PCB

3816

.

FIG. 38

is not shown to scale. The arrangement in

FIG. 38

is a right-angle connector but could easily be adapted for use in a vertical connector. Other nesting or modified arrangements could also be used in this connector. When a connector such as the connector in

FIG. 35

is surface mounted on each side of a PCB, the resulting connector system has a density of approximately 300 contacts per linear inch ((25.4 mm in one inch/2 mm between center of clusters)×3 rows×4 posts×2 sides of the PCB=approximately 300).

In the arrangements of, e.g.,

FIGS. 29

,

31

,

35

,

36

,

37

, and

38

, the rows and columns of the arrangement are continuous. In other words, aside from the regular spacing between the electrical interconnect components in each row and column, there are no breaks or interruptions in the rows or columns of the electrical interconnect components. Such continuous rows and columns are particularly useful in connection with semiconductor chip bonding technologies wherein bonding occurs not only around the periphery of the semiconductor chip, but also directly beneath the chip. This spacing is valuable in high pin count interconnects as well.

Instead of being arranged in continuous rows and columns, the electrical interconnect components

2982

(regardless of whether such components are of the projection-type, the receiving-type, or the hybrid-type) can be arranged in groups or clusters of four or more components separated by channels. This type of arrangement, utilizing the channels for routing traces, allows PCBs (or backplanes) and other interface surface traces to be routed easily to vias and the like on the interface surface. To promote such routing, the channels between the groups of clusters of electrical interconnect components

2982

are wider than the spacings between the electrical interconnect components

2982

within each group or cluster. The use of the channels is applicable to all of the arrangements disclosed in the present application.

Like the contact portion, the foot portion of a conductive post

2211

or conductive beam

2531

may be aligned with or offset from its corresponding stabilizing portion.

Other advantages result from the use of a post

2211

and/or beam

2531

including separate contact, stabilizing, and foot portions, and configurations of such portions other than those discussed above are contemplated. For example, the contact portion of a post or beam may be the same size as the stabilizing portion of that post or beam for ease of manufacturing, or the contact portion may be smaller (i.e., narrower) than the stabilizing portion to increase the density of the interconnect system.

In the situation where the contact portion is made narrower than its corresponding stabilizing portion, the hole or passage in which the post or beam is secured may be configured to have a different width or diameter at different levels. For example, the width or diameter near the portion of the hole through which the contact portion protrudes may be narrower than the width or diameter at the other side of the substrate through which the foot portion protrudes. In this type of configuration, the post or beam is inserted into the hole with the contact portion entering first, and then pushed further into the hole until the shoulder of the stabilizing portion abuts the section of the hole having the narrower width or diameter. By configuring the hole in this manner, over-insertion (i.e., insertion of the post or beam to the extent that the stabilizing portion extends through the hole), as well as push-out due to high mating forces, may be prevented.

Like the contact portion, the foot portion of each post or beam may be the same size as the stabilizing portion of that post or beam, or the foot portion may be smaller (i.e., narrower) than the stabilizing portion to interface with high-density interface devices and/or provide circuit design and routing flexibility. In the situation where the foot portion is made narrower than its corresponding stabilizing portion, the hole or passage in which the post or beam is secured may be configured to have a different width or diameter at different levels. For example, the width or diameter near the portion of the hole through which the foot portion protrudes may be narrower than the width or diameter at the other side of the substrate through which the contact portion protrudes. In this type of configuration, the post or beam is inserted into the hole with the foot portion entering first, and then pushed further into the hole until the shoulder of the stabilizing portion abuts the section of the hole having the narrower width or diameter. By configuring the hole in this manner, over-insertion (i.e., insertion of the post or beam to the extent that the stabilizing portion extends through the hole), as well as push-out due to high mating forces, may be prevented.

It should be noted that when the contact portion of a post or beam is offset from the stabilizing portion, the post or beam must be inserted into the corresponding hole with the foot portion entering first. Similarly, when the foot portion of a post or beam is offset from the stabilizing portion, the post or beam must be inserted into the corresponding hole with the contact portion entering first.

The foot portion of each post or beam may be arranged in many different configurations. For example, the foot portion may have its central axis aligned with the central axis of the stabilizing portion. Alternatively, the foot portion may be offset from the stabilizing portion so that a side of the foot portion is coplanar with a side of the stabilizing portion.

Also, the foot portion of each post or beam may be attached to different portions of the stabilizing portion. For example, the foot portion may be attached to the middle, corner, or side of a stabilizing portion to allow trace routing and circuit design flexibility, and increased interface device density.

Further variations of the foot portion of each post or beam are contemplated. Within a given projection-type or receiving-type interconnect component, the foot portions of that component can be configured to face toward or away from one another, or certain foot portions may face toward one another while other ones of the foot portions face away from one another. Likewise, the foot portions of a given interconnect component may be arranged so that each foot portion faces the foot portion to its immediate left, or so that each foot portion faces the foot portion to its immediate right.

Also, a secondary molding operation could be used to bind the foot portions of one or more interconnect components together. In this type of configuration, an insulative yoke or substrate could be formed around the foot portions just above the point at which the foot portions connect to the interface device to hold the foot The use of posts and beams which include separate contact, stabilizing, and foot portions formed from a single piece maximizes the efficiency and effectiveness of the interconnect arrangement of the present invention. Further, the selective structure of the conductive posts and beams allows flexibility in circuit design and signal routing not possible through the use of existing interconnect systems.

6. Manufacturing

The conductive posts and conductive beams of the electrical interconnect components may be stamped from strips or from drawn wire, and are designed to ensure that the contact and interface portions face in the proper direction in accordance with the description of the posts and beams above. Both methods allow for selective plating and automated insertion. The foot portions in the right-angle embodiments protrude from the center of the stabilizing section, thereby allowing one pin die with different tail lengths to supply contacts for all sides and levels of the electrical interconnect system of the present invention. However, for maximum density, the foot portions may be moved away from the center of the stabilizing portion to allow maximum density while avoiding interference between adjacent foot portions.

The stamped contacts can be either loose or on a strip since the asymmetrical shape lends itself to consistent orientation in automated assembly equipment. Strips can either be between stabilizing areas, at the tips, or as part of a bandolier which retains individual contacts. The different length tails on the right-angle versions assist with orientation and vibratory bowl feeding during automated assembly.

The present invention is compatible with both stitching and gang insertion assembly equipment. The insulative connector bodies and packaging have been designed to facilitate automatic and robotic insertion onto PCB or backplanes or in termination of wire to connector. As an alternative to forming an insulative substrate and then inserting the contacts into the substrate, the insulative substrate may be formed around the contacts in an insert molding process.

D. Summary

The PCBs of the present invention input and output high bandwidth data and are connected using high-density connectors, examples of which are given herein.

Other embodiments will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. It is intended that the specification and examples be considered as exemplary only, with a true scope of the invention being indicated by the following claims.

Number	Name	Date	Kind
3337838	Damiano et al.	Aug 1967	A
3366915	Miller	Jan 1968	A
3444506	Wedekind	May 1969	A
3848221	Lee, Jr.	Nov 1974	A
4274700	Keglewitsch et al.	Jun 1981	A
4487463	Tillotson	Dec 1984	A
4572604	Ammon et al.	Feb 1986	A
4616406	Brown	Oct 1986	A
4654472	Goldfarb	Mar 1987	A
4655526	Shaffer	Apr 1987	A
4663729	Matick et al.	May 1987	A
4697858	Balakrishnan	Oct 1987	A
4698663	Sugimoto et al.	Oct 1987	A
4700274	Laut	Oct 1987	A
4734042	Martens et al.	Mar 1988	A
4871324	Brune et al.	Oct 1989	A
4876664	Bittorf et al.	Oct 1989	A
4881901	Mendenhall et al.	Nov 1989	A
4897055	Jurista et al.	Jan 1990	A
4932902	Crane, Jr.	Jun 1990	A
4943846	Shirling	Jul 1990	A
4959750	Cnyrim et al.	Sep 1990	A
4975066	Sucheski et al.	Dec 1990	A
4979075	Murphy	Dec 1990	A
4997376	Buck et al.	Mar 1991	A
5015207	Koepke	May 1991	A
5037311	Frankeny et al.	Aug 1991	A
5052936	Biechler et al.	Oct 1991	A
5065141	Whitsitt	Nov 1991	A
5071363	Reylek et al.	Dec 1991	A
5073864	Methvin et al.	Dec 1991	A
5081563	Feng et al.	Jan 1992	A
5102342	Marian	Apr 1992	A
5117069	Higgins, III	May 1992	A
5119498	McNeill et al.	Jun 1992	A
5123164	Shaheen et al.	Jun 1992	A
5137456	Desai et al.	Aug 1992	A
5162979	Anzelone et al.	Nov 1992	A
5167511	Krajewski et al.	Dec 1992	A
5192220	Billman et al.	Mar 1993	A
5205739	Malo et al.	Apr 1993	A
5227957	Deters	Jul 1993	A
5281151	Arima et al.	Jan 1994	A
5301281	Kennedy	Apr 1994	A
5309024	Hirano	May 1994	A
5317697	Husak et al.	May 1994	A
5326936	Taniuchi et al.	Jul 1994	A
5330372	Pope et al.	Jul 1994	A
5334279	Gregoire	Aug 1994	A
5342999	Frei et al.	Aug 1994	A
5349343	Oliver	Sep 1994	A
5351393	Gregoire	Oct 1994	A
5355453	Row et al.	Oct 1994	A
5357624	Lavan	Oct 1994	A
5371404	Juskey et al.	Dec 1994	A
5371866	Cady	Dec 1994	A
5376825	Tukamoto et al.	Dec 1994	A
5383787	Switky et al.	Jan 1995	A
5384692	Jaff	Jan 1995	A
5390412	Gregoire	Feb 1995	A
5428750	Hsieh et al.	Jun 1995	A
5432682	Giehl et al.	Jul 1995	A
5461723	Shah et al.	Oct 1995	A
5499341	Wilson et al.	Mar 1996	A
5572685	Fisher et al.	Nov 1996	A
5575688	Crane, Jr.	Nov 1996	A
5617546	Shih et al.	Apr 1997	A
5625780	Hsieh et al.	Apr 1997	A
5625802	Cho et al.	Apr 1997	A
5630163	Fung et al.	May 1997	A
5649100	Ertel et al.	Jul 1997	A
5667884	Bolger	Sep 1997	A
5812797	Crane, Jr. et al.	Sep 1998	A
5822551	Crane, Jr. et al.	Oct 1998	A
6073229	Crane, Jr. et al.	Jun 2000	A

Number	Date	Country
3737819	May 1988	DE
0 237 193	Nov 1987	EP
0 321 212	Jun 1989	EP
0 400 930	Dec 1990	EP
0 405 454	Jan 1991	EP
0 467 698	Jan 1992	EP
0 509 765	Oct 1992	EP
0 526 726	Feb 1993	EP
0 574 133	Dec 1993	EP
2 513 477	Mar 1983	FR
1129608	Oct 1968	GB
1991-0008536	May 1991	KR
1993-10729	Jun 1993	KR
WO 94-13034	Jun 1994	WO
WO 9427345	Nov 1994	WO

	Number	Date	Country
Parent	08/921463	Sep 1997	US
Child	09/536628		US
Parent	08/464388	Jun 1995	US
Child	08/921463		US

Modular architecture for high bandwidth computers

Information

Patent Number

Date Filed

Date Issued

Inventors

Original Assignees

Examiners

Agents

CPC

US Classifications

Field of Search

US

International Classifications

Abstract

Description

Claims

Parent Case Info

US Referenced Citations (75)

Foreign Referenced Citations (15)

Non-Patent Literature Citations (24)

Continuations (2)

Entry
M.J. Doughterty, “A Sem-E Module Avionics Computer with PI-Bus Backplane Communication”, Institute of Electrical and Electronics Engineers, May 21-25, 1990, pp. 169-173.
D. Eidsmore, “MC68000 and Z80A Share Multiprocesor Bus in S-100 SBC Line”, Computer Design, Jun. 1982, Winchester, Massachusetts, vol. 21, No. 6, pp. 38-40.
Y. Belopolsky, “Interaction of Multichip Module Substrates with High-density Connectors” of IEEE Micro, Apr. 1993, vol. 13, No. 2, ISSN 0272-1732, pp. 36-44.
C. Van Veen, “LRU Servicing Practice Impacts Interconnection”, Electronic Product Design, Jun. 1992, UK., vol. 13, No. 6, ISSN 0263-1474, pp. 45-47.
E. Baker and J. Shakib, “Input/Output Cards for Device Attachments”, IBM Technical Disclosure Bulletin, vol. 27, No. 4b, Sep. 1984, New York, pp. 2406-2407.
M. Ernstberger et al., “The High Density Backplane Interconnect System”, Connection Technology, Aug. 1991, vol. 7, No. 8, ISSN 8756-4076, pp. 19-21.
Readout, “Board Level Options for RISC processors”, Electronic Engineering 61, Mar. 1989, No. 747, Woolwich, London, GB, pp. 71-72.
C Van Veen, “Making the Right Packaging Connections with Backplane Interconnections,” Electronic Packaging & Production, Newton, MA, May 1991., pp. 76-79.
“AMP Product Guide”, AMP Inc., 1991, pp. 3008, 3067, 3068, 3102, 3103 and 3123.
R.R. Tummala et al., “Microelectronics Packaging Handbook,” Van Nostrand Reinhold, 1989, pp. 38-43, 398,403, 779-791, 853-859, and 900-905.
“Packing,” Intel Corporation, 1993, pp. 2-36; 2-96, 2-97, 2-100, 3-2, 3-24, and 3-25.
George D. Gregoire, “3-Dimensional Circuitry Solves fine Pitch SMT Device Assembly Problem,” Connection Technology.
Dimensional Circuits Corporation, “Dimensional Circuits Corp. Awarded Two U.S. Patents,” D.D.C. News, Apr. 5, 1994.
George D. Gregoire, “Very Fine Line Recessed Circuitry—A New PCB Fabrication Process”.
“AMP-ASC Interconnection Systems,” AMP Incorporated, Product Information Bulletin, pp. 1-4, copyright 1991.
Barnhouse, “Bifurcated Through-Hole Technology—An Innovative Solution to Circuit Density,” Connection Technology, pp. 33-35, Feb. 1992.
“Micro-Strip Interconnection System,” AMP Product Guide, pp. 3413-3414, Jun., 1991.
“Rib-Cage II Through-Mount Shrouded Headers” and “Micropax Board-to-Board Interconnect System,” Du Pont Connector Systems, Product Catalog A, pp. 2-6, 3.0, 3-1, Feb., 1992.
“82430 PCIset Cache/Memory Subsystem,” Intel Corporation, pp. 1-888, Mar. 1993.
“The PCI Revision 1.0 Specification,” published by Intel Corporation, specifically Chapter 4.42“Systems, Parameters,” Jun. 22, 1992.
“82420/82430 PCIset ISA and EISA Bridges,” Intel Corporation, pp. 1-350, Dec. 1993.
“R4000 Microprocessor User's Manual” with inserts by Toshiba.
“Pentium Processor User's Manual,” Intel corporation, vols. 1-3, 1993.
“Short Form Catalog,” 82C711 Data Book, Chips adn Technologies, Inc., Fall, 1992.