|Publication number||US20070111606 A1|
|Application number||US 11/564,199|
|Publication date||May 17, 2007|
|Filing date||Nov 28, 2006|
|Priority date||Sep 3, 2004|
|Also published as||US7511968, US20060049512, US20060050497|
|Publication number||11564199, 564199, US 2007/0111606 A1, US 2007/111606 A1, US 20070111606 A1, US 20070111606A1, US 2007111606 A1, US 2007111606A1, US-A1-20070111606, US-A1-2007111606, US2007/0111606A1, US2007/111606A1, US20070111606 A1, US20070111606A1, US2007111606 A1, US2007111606A1|
|Original Assignee||Staktek Group L.P., A Texas Limited Partnership|
|Export Citation||BiBTeX, EndNote, RefMan|
|Referenced by (31), Classifications (21), Legal Events (1)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This application is a continuation of U.S. patent application Ser. No. 11/007,551 filed Dec. 8, 2004, which is a continuation-in-part of U.S. patent application Ser. No. 10/934,027 filed Sep. 3, ?9004. U.S. patent application Ser. Nos. 11/007,551 and 11/934,027 are incorporated herein by reference.
The present invention relates to systems and methods for creating high density circuit modules.
The well-known DIMM (Dual In-line Memory Module) board has been used for years, in various forms, to provide memory expansion. A typical DIMM includes a conventional PCB (printed circuit board) with memory devices and supporting digital logic devices mounted on both sides. The DIMM is typically mounted in the host computer system by inserting a contact-bearing edge of the DIMM into a card edge connector. Systems that employ DIMMs provide, however, very limited profile space for such devices and conventional DIMM-based solutions have typically provided only a moderate amount of memory expansion.
As bus speeds have increased, fewer devices per channel can be reliably addressed with a DIMM-based solution. For example, 288 ICs or devices per channel may be addressed using the SDRAM-100 bus protocol with an unbuffered DIMM. Using the DDR-200 bus protocol, approximately 144 devices may be address per channels with the DDR2-400 bus protocol only 72 devices per channel may be addressed. This constraint has led to the development of the fully-buffered DIMM (FB-DIMM) with buffered C/A and data in which 288 devices per channel may be addressed. With the FB-DIMM, not only has capacity increased, pin count has declined to approximately 69 from the approximately 240 pins previously required.
The FB-DIMM circuit solution is expected to offer practical motherboard memory capacities of up to about 192 gigabytes with six channels and eight DIMMs per channel and two ranks per DIMM using one gigabyte DRAMs. This solution should also be adaptable to next generation technologies and should exhibit significant downward compatibility.
This great has, however, come with some cost and will eventually be self-limiting. The basic principle of systems that employ FB-DIMM relies upon a point-to-point or serial addressing scheme rather than the parallel multi-drop interface that dictates non-buffered DIMM addressing. That is, one DIMM is in point-to-point relationship with the memory controller and each DIMM is in point-to-point relationship with adjacent DIMMs. Consequently, as bus speeds increase, the number of DIMMs on a bus will decline as the discontinuities caused by the chain of point to point connections from the controller to the “last” DIMM become magnified in effect as speeds increase. Consequently, methods to increase the capacity of a single DIMM find value in contemporary memory and computing systems.
There are several known methods to improve the limited capacity of a DIMM or other circuit board. In one strategy, for example, small circuit boards (daughter cards) are connected to the DIMM to provide extra mounting space. The additional connection may cause, however, flawed signal integrity for the data signals passing from the DIMM to the daughter card and the additional thickness of the daughter card(s) increases the profile of the DIMM.
Multiple die packages (MDP) are also used to increase DIMM capacity while preserving profile conformity. This scheme increases the capacity of the memory devices on the DIMM by including multiple semiconductor die in a single device package. The additional heat generated by the multiple die typically requires, however, additional cooling capabilities to operate at maximum operating speed. Further, the MDP scheme may exhibit increased costs because of increased yield loss from packaging together multiple die that are not fully pre-tested.
Stacked packages are yet another strategy used to increase circuit board capacity. This scheme increases capacity by stacking packaged integrated circuits to create a high-density circuit module for mounting on the circuit board. In some techniques, flexible conductors are used to selectively interconnect packaged integrated circuits. Staktek Group L.P. has developed numerous systems for aggregating CSP (chipscale packaged) devices in space saving topologies. The increased component height of some stacking techniques may alter, however, system requirements such as, for example, required cooling airflow or the minimum spacing around a circuit board on its host system.
What is needed, however, are methods and structures for increasing the flexibility of the FB-DIMM solution.
Multiple fully buffered DIMM circuits or instantiations are combined in a single module to provide on a single module circuitry that is substantially the functional equivalent of two or more FB-DIMMs but avoids some of the drawbacks associated with having two discrete FB-DIMMs In a preferred embodiment, integrated circuits (preferably memory CSPs) and accompanying AMBs are arranged in two ranks in two fields on each side of a flexible circuit. The flexible circuit has expansion contacts disposed along one side. The flexible circuit is disposed about a supporting substrate or board to place at least one FB-DIMM instantiation on each side of the constructed module. In alternative, but also preferred embodiments, the ICs on the side of the flexible circuit closest to the substrate are disposed, at least partially, in what are, in a preferred embodiment, windows, pockets, or cutaway areas in the substrate. Other embodiments may only populate one side of the flexible circuit or may only remove enough substrate material to reduce but not eliminate the entire substrate contribution to overall profile. Other embodiments may connect the constituent devices in a way that creates a FB-DIMM circuit or instantiation with the devices on the upper half of the module while another FB-DIMM instantiation is created with the devices on the lower half of the module. Other embodiments may, for example, combine selected circuitry from one side of the module (memory CSPs for example) with circuitry on the other side of the module (an AMB, for example) in creating one of plural FB-DIMM instantiations on a single module. Other embodiments employ stacks to provide multiple FB-DIMM circuits or instantiations On a low profile module. The flexible circuit may exhibit one or two or more conductive layers, and may have changes in the layered structure or have split layers. Other embodiments may stagger or offset the ICs or include greater numbers of ICs.
ICs 18 on flexible circuit 12 are, in this embodiment, chip-scale packaged memory devices of small scale. For purposes of this disclosure, the term chip-scale or “CSP” shall refer to integrated circuitry of any function with an array package providing connection to one or more die through contacts (often embodied as “bumps” or “balls” for example) distributed across a major, surface of the package or die. CSP does not refer to leaded devices that provide connection to an integrated circuit within the package through leads emergent from at least one side of the periphery of the package such as, for example, a TSOP.
Embodiments of the present invention may be employed with leaded or CSP devices or other devices in both packaged and unpackaged forms but where the term CSP is used, the above definition for CSP should be adopted. Consequently, although CSP excludes leaded devices, references to CSP are to be broadly construed to include the large variety of array devices (and not to be limited to memory only) and whether die-sized or other size such as BGA and micro BGA as well as flip-chip. As those of skill will understand after appreciating this disclosure, some embodiments of the present invention may be devised to employ stacks of ICs each disposed where an IC 18 is indicated in the exemplar Figs.
Multiple integrated circuit die may be included in a package depicted as a single IC 18. While in this embodiment memory ICs are used to provide a memory expansion board or module, and various embodiments may include a variety of integrated circuits and other components. Such variety may include microprocessors, FPGA's, RF transceiver circuitry, digital logic, as a list of non-limiting examples, or other circuits or systems which may benefit from a high-density circuit board or module capability. Circuit 19 depicted between ICs 18 may be a memory buffer or controller but in a preferred embodiment is the well known advanced memory buffer or “AMB”.
The depiction of
Field F1 of side 8 of flex circuit 12 is shown populated with first plurality of CSPs ICR1 and second plurality of CSPs ICR2 while second field F2 of side 8 of flex circuit 12 is shown populated with first plurality of CSPs ICR1 and second plurality of CSPs ICR2. Those of skill will recognize that the identified pluralities of CSPs are, when disposed in the configurations depicted, typically described as “ranks”. Between the ranks ICR2 of field F1 and ICR2 of field F2, flex circuit 12 bears a plurality of module contacts allocated in this embodiment into two rows (CR1 and CR2) of module contacts 20. When flex circuit 12 is folded as later depicted, side 8 depicted in
Various discrete components such as termination resistors, bypass capacitors and bias resistors, in addition to the buffers 19 shown on side 8 of flex circuit 12, may be mounted on either or both of sides 8 and 9 of flex 12. Such discrete components are not shown to simplify the drawing. Flex circuit 12 may also depicted with reference to its perimeter edges, two of which are typically long (PElong1 and PElong2) and two of which are typically shorter (PEshort1 and PEshort2). Other embodiments may employ flex circuits 12 that are not rectangular in shape and may be square in which case the perimeter edges would be of equal size or other convenient shape to adapt to manufacturing particulars. Other embodiments may also have fewer or greater numbers of ranks or pluralities of ICs in each field or on a side of a flex circuit.
In this embodiment, flex circuit 12 has module contacts 20 positioned in a manner devised to fit in a circuit board card edge connector or socket and connect to corresponding contacts in the connector (not shown). While module contacts 20 are shown protruding from the surface of flex circuit 12, other embodiments may have flush contacts or contacts below the surface level of flex 12. Substrate 14 supports module contacts 20 from behind flex circuit 12 in a manner devised to provide the mechanical form required for insertion into a socket. In other embodiments, the thickness or shape of substrate 14 in the vicinity of perimeter edge 16A may differ from that in the vicinity of perimeter edge 16B. Substrate 14 in the depicted embodiment is preferably made of a metal such as aluminum or copper, as non-limiting examples, or where thermal management is less of an issue, materials such as FR4 (flame retardant type 4) epoxy laminate, PTFE (poly-tetra-fluoro-ethylene) or plastic. In another embodiment, advantageous features from multiple technologies may be combined with use of FR4 having a layer of copper on both sides to provide a substrate 14 devised from familiar materials which may provide heat conduction or a ground plane.
In this embodiment, there are three layers of flex circuit 12 between the two depicted ICs 18 1 and 18 2. Conductive layers 64 and 66 express conductive traces that connect to the IC's and may further connect to other discrete components (not shown). Preferably, the conductive layers arc metal stick as, for example, copper or alloy 110. Vias such as the exemplar vias 23 connect the two conductive layers 64 and 66 and thereby enable connection between conductive layer 64 and module contacts 20. In this preferred embodiment having a three-layer portion of flex circuit 12, the two conductive layers 64 and 66 may be devised in a manner so that one of them has substantial area employed as a ground plane. The other layer may employ substantial area as a voltage reference plane. The use of plural conductive layers provides advantages and the creation of a distributed capacitance intended to reduce noise or bounce effects that can, particularly at higher frequencies, degrade signal integrity, as those of skill in the art will recognize. If more than two conductive layers are employed, additional conductive layers may be added with insulating layers separating conductive layers. Portions of flex circuit 12 may in some embodiments be rigid portions (rigid-flex). Construction of rigid-flex circuitry is known in the art.
With the construction of an embodiment such as that shown in
The principles of the present invention may, however, be employed where only one IC 18 is resident on a side of a flex circuit 12 or where multiple ranks or pluralities of ICS are resident on a side of flex circuit 12, or, as will be later shown, where multiple ICs 18 are disposed one atop the other to give a single module 10 materially greater.
As shown in
For purposes herein, the term window may refer to an opening all the way through substrate 14 across span “S” which corresponds to the width or height dimension of packaged IC 18 or, it may also refer to that opening where cutaway areas on each of the two sides of substrate 14 overlap.
Where cutaway areas 250B3 and 250B4 overlap, there are, as depicted, windows all the way through substrate 14. In some embodiments, cutaway areas 250B3 and 2501B4 may not overlap or in other embodiments, there may be pockets or cutaway areas only on one side of substrate 14. Those of skill will recognize that cutaway areas such as those identified with references 250B3 and 250B4 may be formed in a variety of ways depending on the material of substrate 14 and need not literally be “cut” away but may be formed by a variety of molding, milling and cutting processes as is understood by those in the field.
Four flex circuits are employed in module 10 as depicted in
In a typical FB-DIMM system employing multiple FB-DIMM circuits, the respective AMB's from one FB-DIMM circuit to another FB-DIMM circuit are separated by what can be conceived of as three impedance discontinuities as represented in the system depicted in
In contrast to the system represented by
The present invention may be employed to advantage in a variety of applications and environment such as, for example, in computers such as servers and notebook computers by being placed in motherboard expansion slots to provide enhanced memory capacity while utilizing fewer sockets. The two high rank embodiments or the single rank high embodiments may both be employed to such advantage as those of skill will recognize after appreciating this specification.
One advantageous methodology for efficiently assembling a circuit module 10 such as described and depicted herein is as follows. In a preferred method of assembling a preferred module assembly 10, flex circuit 12 is placed flat and both sides populated according to circuit board assembly techniques known in the art. Flex circuit 12 is then folded about end 16A of substrate 14. Flex 12 may be laminated or otherwise attached to substrate 14.
Although the present invention has been described in detail, it will be apparent to those skilled in the alt that many embodiments taking a variety of specific forms and reflecting changes, substitutions and alterations can be made without departing from the spirit and scope of the invention. Therefore. the described embodiments illustrate but do not restrict the scope of the claims.
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7307863 *||Aug 2, 2005||Dec 11, 2007||Inphi Corporation||Programmable strength output buffer for RDIMM address register|
|US7393226 *||Mar 7, 2007||Jul 1, 2008||Microelectronics Assembly Technologies, Inc.||Thin multichip flex-module|
|US7394149 *||Mar 7, 2007||Jul 1, 2008||Microelectronics Assembly Technologies, Inc.||Thin multichip flex-module|
|US7429788 *||Mar 7, 2007||Sep 30, 2008||Microelectronics Assembly Technologies, Inc.||Thin multichip flex-module|
|US7442050||Aug 28, 2006||Oct 28, 2008||Netlist, Inc.||Circuit card with flexible connection for memory module with heat spreader|
|US7480152 *||Dec 7, 2004||Jan 20, 2009||Entorian Technologies, Lp||Thin module system and method|
|US7511968 *||Dec 8, 2004||Mar 31, 2009||Entorian Technologies, Lp||Buffered thin module system and method|
|US7520781 *||Mar 7, 2007||Apr 21, 2009||Microelectronics Assembly Technologies||Thin multichip flex-module|
|US7558887||Sep 5, 2007||Jul 7, 2009||International Business Machines Corporation||Method for supporting partial cache line read and write operations to a memory module to reduce read and write data traffic on a memory channel|
|US7584308||Aug 31, 2007||Sep 1, 2009||International Business Machines Corporation||System for supporting partial cache line write operations to a memory module to reduce write data traffic on a memory channel|
|US7618874||May 2, 2008||Nov 17, 2009||Micron Technology, Inc.||Methods of forming capacitors|
|US7696056||May 2, 2008||Apr 13, 2010||Micron Technology, Inc.||Methods of forming capacitors|
|US7770077||Jan 24, 2008||Aug 3, 2010||International Business Machines Corporation||Using cache that is embedded in a memory hub to replace failed memory cells in a memory subsystem|
|US7811097||Sep 24, 2008||Oct 12, 2010||Netlist, Inc.||Circuit with flexible portion|
|US7818497||Aug 31, 2007||Oct 19, 2010||International Business Machines Corporation||Buffered memory module supporting two independent memory channels|
|US7839643||Nov 12, 2009||Nov 23, 2010||Netlist, Inc.||Heat spreader for memory modules|
|US7839645||Oct 26, 2009||Nov 23, 2010||Netlist, Inc.||Module having at least two surfaces and at least one thermally conductive layer therebetween|
|US7840748||Aug 31, 2007||Nov 23, 2010||International Business Machines Corporation||Buffered memory module with multiple memory device data interface ports supporting double the memory capacity|
|US7861014||Aug 31, 2007||Dec 28, 2010||International Business Machines Corporation||System for supporting partial cache line read operations to a memory module to reduce read data traffic on a memory channel|
|US7865674||Aug 31, 2007||Jan 4, 2011||International Business Machines Corporation||System for enhancing the memory bandwidth available through a memory module|
|US7899983||Aug 31, 2007||Mar 1, 2011||International Business Machines Corporation||Buffered memory module supporting double the memory device data width in the same physical space as a conventional memory module|
|US7925824||Jan 24, 2008||Apr 12, 2011||International Business Machines Corporation||System to reduce latency by running a memory channel frequency fully asynchronous from a memory device frequency|
|US7925825||Jan 24, 2008||Apr 12, 2011||International Business Machines Corporation||System to support a full asynchronous interface within a memory hub device|
|US7925826||Jan 24, 2008||Apr 12, 2011||International Business Machines Corporation||System to increase the overall bandwidth of a memory channel by allowing the memory channel to operate at a frequency independent from a memory device frequency|
|US7930469||Jan 24, 2008||Apr 19, 2011||International Business Machines Corporation||System to provide memory system power reduction without reducing overall memory system performance|
|US7930470||Jan 24, 2008||Apr 19, 2011||International Business Machines Corporation||System to enable a memory hub device to manage thermal conditions at a memory device level transparent to a memory controller|
|US7964471||Mar 3, 2010||Jun 21, 2011||Micron Technology, Inc.||Methods of forming capacitors|
|US8241987||May 17, 2011||Aug 14, 2012||Micron Technology, Inc.||Methods of forming capacitors|
|US8318578||Oct 7, 2009||Nov 27, 2012||Micron Technology, Inc.||Method of forming capacitors|
|US8623725||Jul 23, 2012||Jan 7, 2014||Micron Technology, Inc.||Methods of forming capacitors|
|DE102007044200A1 *||Sep 17, 2007||Mar 19, 2009||Qimonda Ag||Memory module has two advanced memory buffer chips electrically coupled over point-to-point connections and electrical connection between two advanced memory buffer chips and memory controller over one of advanced memory buffer chip|
|U.S. Classification||439/633, 257/E25.023|
|Cooperative Classification||H01L23/36, H05K2203/1572, H05K2201/2018, H05K1/189, H05K2201/1056, H05K1/0203, G11C5/04, H01L25/105, H05K2201/10734, H05K1/118, H05K3/0061, H05K2201/056, H05K1/181, H01L23/34, H01L25/0652, H01L2224/16225|
|European Classification||G11C5/04, H05K1/18F|
|Mar 27, 2009||AS||Assignment|
Owner name: ENTORIAN TECHNOLOGIES, LP,TEXAS
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:GOODWIN, PAUL;REEL/FRAME:022461/0404
Effective date: 20041207