United States Court of Appeals: No. 2012-1480 (Reexamination No. 95/001,166)
United States Court of Appeals: No. 2012-1480 (Reexamination No. 95/001,166)
Document: 33
Page: 1
Filed: 01/16/2013
RAMBUS, INC. Appellant, v. DAVID J. KAPPOS, DIRECTOR, UNITED STATES PATENT AND TRADEMARK OFFICE, Appellee. Appeal from the United States Patent and Trademark Office, Board of Patent Appeals and Interferences REPLY BRIEF FOR APPELLANT RAMBUS INC. Kara F. Stoll FINNEGAN, HENDERSON, FARABOW, GARRETT & DUNNER, LLP 901 New York Avenue, N.W. Washington, D.C. 20001 (202) 408-4000 John M. Whealan 4613 Merivale Road Chevy Chase, MD 20815 (202) 994-2195 Jeffrey A. Lamken Counsel of Record Michael G. Pattillo, Jr. MOLOLAMKEN LLP The Watergate, Suite 660 600 New Hampshire Avenue, N.W. Washington, D.C. 20037 (202) 556-2010
Case: 12-1480
Document: 33
Page: 2
Filed: 01/16/2013
TABLE OF CONTENTS Page ARGUMENT..........................................................................................................2 I. The Directors Claim Constructions Cannot Be Sustained .............................2 A. B. The Claimed Signal Is Not Any Physical Representation of Data......................................................................................................2 Claim 1 Requires Its Steps To Be Performed in Sequence and Separately (Decoupled)Not Merely That the Method Exhibit Flexibility.........................................................................................10 The 109 Patent Does Not Claim the Prior Art It Distinguishes.........14 The Directors Reliance on the Delay Value Embodiment Is Misplaced ............................................................................................17
C. D.
II. The Director Cannot Salvage the Boards Anticipation Analysis.................21 A. B. Option (1)In the Request PacketFails......................................22 The Director Effectively Concedes the Failure of the Boards Options (2)-(4).....................................................................................27
CONCLUSION ....................................................................................................29
Case: 12-1480
Document: 33
Page: 3
Filed: 01/16/2013
TABLE OF AUTHORITIES Page(s) CASES Altiris, Inc. v. Symantec Corp., 318 F.3d 1363 (Fed. Cir. 2003).............................10 Andersen Corp. v. Fiber Composites, LLC, 474 F.3d 1361 (Fed. Cir. 2007) .........21 CAE Screenplates Inc. v. Heinrich Fiedler GmbH & Co. KG, 224 F.3d 1308 (Fed. Cir. 2000).....................................................................................................4 CCS Fitness, Inc. v. Brunswick Corp., 288 F.3d 1359 (Fed. Cir. 2002) .................14 Enzo Biochem, Inc. v. Applera Corp., 599 F.3d 1325 (Fed. Cir. 2010) ..................27 In re Abbott Diabetes Care Inc., 696 F.3d 1142 (Fed. Cir. 2012).........................1, 4 In re Bond, 910 F.2d 831 (Fed. Cir. 1990) ..............................................................10 In re De Blauwe, 736 F.2d 699 (Fed. Cir. 1984) .....................................................24 In re Thrift, 298 F.3d 1357 (Fed. Cir. 2002)............................................................24 Interactive Gift Express, Inc. v. Compuserve Inc., 256 F.3d 1323 (Fed. Cir. 2001)...................................................................................................21 Invitrogen Corp. v. Biocrest Mfg., L.P., 327 F.3d 1364 (Fed. Cir. 2003) ...............20 Mantech Envtl. Corp. v. Hudson Envtl. Servs., Inc., 152 F.3d 1368 (Fed. Cir. 1998)...................................................................................................11 MBO Labs., Inc. v. Becton, Dickinson & Co., 474 F.3d 1323 (Fed. Cir. 2007)...................................................................................................20 Net MoneyIN, Inc. v. VeriSign, Inc., 545 F.3d 1359 (Fed. Cir. 2008).....2, 22, 25, 28 Omega Engg, Inc., v. Raytek Corp., 334 F.3d 1314 (Fed. Cir. 2003)....................21 Phillips v. AWH Corp., 415 F.3d 1303 (Fed. Cir. 2005) (en banc) .................7, 9, 12 Renishaw PLC v. Marposs Societa per Azioni, 158 F.3d 1243, 1250 (Fed. Cir. 1998)...................................................................................................16 ii
Case: 12-1480
Document: 33
Page: 4
Filed: 01/16/2013
Rolls-Royce, PLC v. United Techs. Corp., 603 F.3d 1325 (Fed. Cir. 2010) .....17, 20 Semitool, Inc. v. Dynamic Micro Sys. Semiconductor Equip. GMBH, 444 F.3d 1337 (Fed. Cir. 2006) ....................................................................28, 29
iii
Case: 12-1480
Document: 33
Page: 5
Filed: 01/16/2013
The Directors defense of the Boards decision hinges on 5 propositions: 1. The term signal in claim 1 includes any physical representation of data, even though that claim distinguishes signal from data (and from code), using those terms to mean different things; Even though claim 1 recites 4 distinct, sequential steps (step 1 followed by 2, then 3, then 4)and steps 3 and 4 concededly must proceed in sequencethe first step (providing a first code) can come second and the second step can come first; While claim 1 and the specification support providing the write command and the signal to begin data transfer separately, separate or decoupled really means flexible; The 109 patent encompasses the prior-art 037 patent it distinguishes as inflexible; and [P]roviding a signal to the memory device encompasses moving a signal within the memory device.
2.
3.
4. 5.
The Director thus reads claim 1 as satisfied whenever any data (including the code specifying the write operation or the bits to be written to memory), obtained from any location (whether sent to the device or already stored within the device), provided in any sequence with other claim steps (including before the write has been specified or after the bits have been written), indicates to the memory device when data transfer should initiateso long as the result can be considered flexible. Although the Board is entitled to give claim terms their broadest reasonable construction . . . consistent with the specification, In re Abbott Diabetes Care Inc., 696 F.3d 1142, 1149 (Fed. Cir. 2012), its construction here is neither reasonable nor consistent with the specification.
Case: 12-1480
Document: 33
Page: 6
Filed: 01/16/2013
Even apart from claim construction, the finding of anticipation is erroneous. The prior-art 037 patent does not disclose[ ] within the four corners of the document . . . all of [claim 1s] limitations, much less all of the limitations arranged or combined in the same way as recited in the claim. Net MoneyIN, Inc. v. VeriSign, Inc., 545 F.3d 1359, 1371 (Fed. Cir. 2008). The Director focuses on only one of the Boards four theories for how the 037 patent anticipates the 109 patentthat the 037 patent discloses placing a delay value in the request packet. But that theory lacks support in the 037 patent and overlooks critical differences between the particular method the 037 patent discloses and the invention claimed in the 109 patent. ARGUMENT I. The Directors Claim Constructions Cannot Be Sustained A. The Claimed Signal Is Not Any Physical Representation of Data
The prior-art 037 patent cannot anticipate claim 1 unless, as the Board posited, claim 1s recited signal (which is sent to the memory device to indicate when to begin sampling data, i.e., transferring data to memory) is the same as the 037 patents delay value (a value previously stored in memory that tells the memory device how long to wait before sampling). Rather than analyze claim 1s terms in context, the Board invoked the IEEE dictionary to assert that signal includes any physical representation of data. See Rambus Br. 38-41. But the
Case: 12-1480
Document: 33
Page: 7
Filed: 01/16/2013
109 patent uses signal differently to mean a transmission that incites the memory device to act: Like the starting gun is the signal that tells sprinters when to begin running, claim 1s signal tells the memory device when to begin sampling data. The signal is not data (to be stored, written, or otherwise), much less a value telling the memory device how long to wait before writing data. 1. The Director nowhere denies that, if signal means any physical
representation of data, it would encompass anything transmitted across a bus. Rambus Br. 39. Claim 1 defies that boundless construction. It carefully distinguishes transmissions, differentiating signal from data and from code. Claim 1 first recites in paragraph [1] providing a code (the write command) that indicates the operation to be performed (the write operation); it then recites in paragraph [2] providing a signal that indicates when sampling is to begin; and finally it recites in paragraphs [3] and [4] providing the data (the bits to be sampled). Id. at 39-40; A74-75, 41:61-43:2. The Directors construction, however, would equate everything in claim 1 with the signal recited in paragraph [2] including the code specifying the write operation in paragraph [1], and the data to be written to memory in paragraphs [3] and [4], A74-75, 41:61-43:2every one of which is a physical representation of data. That cannot be correct. The Court must presume that the use of these different terms in the claims connotes different
Case: 12-1480
Document: 33
Page: 8
Filed: 01/16/2013
meanings. CAE Screenplates Inc. v. Heinrich Fiedler GmbH & Co. KG, 224 F.3d 1308, 1317 (Fed. Cir. 2000). 2. The Director nowhere disputes that a dictionary-first approach contra-
venes this Courts directive that claim terms be interpreted within the context of the patent, and the specification in particular. Abbott, 696 F.3d at 1149. Instead, the Director downplays the Boards reliance on the dictionary. The Board began, the Director declares, by looking to the language of the claims, then by making detailed findings regarding the use of the term signal in the 109 patents specification, and then only referenced the dictionary. PTO Br. 25 (citing JA10, 14, 21). That is mistaken. As to the language of claim 1, the Board nowhere acknowledged that it distinguishes signal from data and from code, distinctions the Boards dictionary definition (any physical representation of data) would obliterate. Far from beginning with claim 1s text, the Board passed over itso does the Director. Indeed, the pages of the decision cited by the Director highlight the Boards erroneous dictionary-directed approach. For example, the first cited page of the Boards decision (JA10) moves directly from stating that the 109 patent does not define the term signal to the supposedly plain meaning of signal, point[ing] to an IEEE Standard Dictionary. The discussion thus goes straight from claim term
Case: 12-1480
Document: 33
Page: 9
Filed: 01/16/2013
to dictionary, without anchoring the definition of signal in the context of the entire claim. As for the Boards supposedly detailed findings regarding the use of signal in the . . . specification, PTO Br. 25, the Director avoids describing them, and with reason. As mentioned, the first cited page (JA10) begins with the dictionary and, in any event, recites the parties arguments, not the Boards findings. The next cited page (JA14) again underscores the Boards dictionary-first approach: It begins by citing the IEEE Dictionary (at the top of the page), while its remaining analysis addresses decoupling, not signal. And the last cited page (JA21)
invokes a supposedly plain meaning of signal, again drawn from the IEEE Dictionary definition.1 The one Board finding the Director proffers is the assertion that the patent uses the term signal broadly, using signal and information interchangeably. Br. 12 (JA21). But that putative finding (actually construing phrases recited in claims 7, 14, and 25, JA21) is indefensible. The cited sentence from the 109 patents specification states that the data transfer start information (a strobe signal) [is] sent from the controller. A58, 9:4-6. That nowhere suggests that ____________________________ The Director states that Rambus has cited this definition in other litigation. PTO Br. 12-13. But that reflected specific points of construction for different claim terms in different patentsnot the view that signal broadly means the physical representation of data in all contexts. 5
1
Case: 12-1480
Document: 33
Page: 10
Filed: 01/16/2013
signal means any transmitted information; it indicates only that a strobe signal can convey one type of informationthe data transfer start information that triggers sampling. The Boards contrary view is like reading the phrase Italian sports cars (Ferraris) and declaring that the word Ferraris encompasses all cars. The equation fails, logically and grammatically. In any event, it cannot overcome claim 1s clear indication that signal is distinct from data and code, and thus not a synonym for any information. The Director also urges that the Board did not equate signal with code, data, and value, but instead equated it with another term of art in the specification, delay value. PTO Br. 26. According to the Director, the Board concluded that signal means not only a transmission that incites the memory device to act, but also a transmitted value that tells the memory device how many clock cycles to wait before acting. See id.2 That misses the point: No one has cited any accepted definition of signal that means a transmission that incites action, like the strobe the patent concededly covers, plus the delay value the Director identifies, but excludes any other values, code, or data that might control when sampling begins. For that reason, the Board could construe signal to encompass delay value only by stretching that term to encompass any physical representa____________________________ Any effort to equate[ ] signal with delay value is self-evidently wrong: The term signal concededly encompasses strobe signal, which is not a delay value. 6
2
Case: 12-1480
Document: 33
Page: 11
Filed: 01/16/2013
tion of data, as the Boards repeated invocation of that definition makes clear. See A10, 13-14, 21. The consequence is to sweep in everything that is transmitted in claim 1 as the recited signal as well. That approach defies the inventors disparate use of terms like signal, data, and code: Where a transmission in claim 1 represents a value (as opposed to something that triggers a response), the claim uses data or code. It also defies practical application. No one would have thought claim 1 covers every method in which any transmitted data (whether the write command or the bits of write data) controls when sampling occurs, a construction so expansive as to cover anything with a write command and data. When the Board starts with the broad dictionary definition without appreciating how the patent itself implicitly limits that definition, the error will systematically cause the construction of the claim to be unduly expansive. Phillips v. AWH Corp., 415 F.3d 1303, 1321 (Fed. Cir. 2005) (en banc). That is what happened here. 3. The Boards construction and the Directors arguments, moreover, are
inconsistent with the specification, which is the single best guide to the meaning of a disputed term. Phillips, 415 F.3d at 1315. Claim 1s signal indicates when the memory device is to begin sampling write data, A74, 42:61-63, telling the memory device to begin sampling in response to the signals arrival; it is not data telling the memory device to wait a specified period following the receipt of the
Case: 12-1480
Document: 33
Page: 12
Filed: 01/16/2013
operation code before beginning sampling. See Rambus Br. 29-31. The specification repeatedly makes clear that [t]he timing of the strobe signal itself indicates when the memory device is to begin sampling. A64, 21:31-32; see also A64, 21:33-36 (sampling begins [i]n response to signal); Rambus Br. 30-31 (terminate signal). The Directors failure to identify any use of signal in the 109 patent to support the Boards physical representation of data definition, A10, 13-14, is telling. The Director nowhere denies that the specification states that the memory device begins sampling [i]n response to the claimed signal. The Director instead disputes that construction by insisting it would require the signal to incite immediate action. PTO Br. 13. The signal, the Director asserts, does not elicit an immediate response. Id. at 14. Respectfully, the Director misunderstands Rambuss position and the invention. The memory device need not begin sampling immediately on receiving the signal. The patent makes clear, for example, that some strobe [signal]-to-data delay is intrinsic to the circuit, as Rambus noted. Rambus Br. 31 (quoting A58, 9:46). The point is that the memory device begins sampling [i]n response to the signal. A64, 21:31-36; see Rambus Br. 29-31. A runner may briefly hesitate before leaving the starting blocks in response to the starting gun; but that latency does not change the fact that the starting gun is a signal when to begin the race, with runners beginning in response to it. Claim
Case: 12-1480
Document: 33
Page: 13
Filed: 01/16/2013
1s signal is no different.
immediate response; the write data arrives on the clock cycle immediately following the strobe signal.3 The Director errs both in requiring immediacy and in denying that it is met. Ultimately, the Director urges that, if Rambus wanted to exclude a construction of signal that encompasses delay values, the specification would have had to limit the claim language explicitly. PTO Br. 14. That is incorrect. A patentee need not use words or expressions of manifest exclusion or restriction to limit a claim against all conceivable readings. Phillips, 415 F.3d at 1320-21. Here, the patent itself shows that claim 1s signal indicat[es] when sampling is to begin by arrivingthat the memory device acts [i]n response to the signal. See Rambus Br. 29-30 (citing A58, 9:4-6, 9:35-37, 10:48-49; A59, 11:38-44; A64, 21:31-36, 21:57-60, 22:48-53; A65, 23:66-24:5). By contrast, the Director has no such support in the patent for the over-expansive signal-encompasses-any-data construction adopted below. Because the Board gave the term signal its broadest conceivable interpretation, rather than the broadest reasonable interpretation
____________________________ See, e.g., A67 (write on cycle 40, strobe on cycle 47, data on cycles 48-63; write on cycle 72, strobe on cycle 79, data on cycles 80-95); A70 (write on cycle 45, strobe on cycle 47, data on cycles 48-63); A71 (write on cycle 113, strobe on cycle 115, data on cycles 116-131). 9
3
Case: 12-1480
Document: 33
Page: 14
Filed: 01/16/2013
consistent with the specification, In re Bond, 910 F.2d 831, 833 (Fed. Cir. 1990) (emphasis added), its ruling cannot be sustained. B. Claim 1 Requires Its Steps To Be Performed in Sequence and Separately (Decoupled)Not Merely That the Method Exhibit Flexibility
The Director acknowledges that provid[ing] a mechanism to decouple control timing from data timing is one object of the 109 patent. PTO Br. 14 (quoting A55, 3:33-34). Indeed, it is the essence of the 109 patents claimed advance over the prior art. See Rambus Br. 26-29, 35-37. Claim 1 requires, as a matter of logic [and] grammar, Altiris, Inc. v. Symantec Corp., 318 F.3d 1363, 1369 (Fed. Cir. 2003), that its steps be performed separately and in the order written. See Rambus Br. 26-28. 1. [1] [2] [3] [4] No one disputes that, in four separate paragraphs, claim 1 recites: a first code specifying the write operation, followed by a signal to the memory device that indicates when the memory device is to begin sampling write data, followed by providing a first bit of the write data to the memory device, and then providing a second bit of the write data to the memory device.
A74-75, 41:61-43:2; Rambus Br. 10. Nor does anyone deny that claim paragraph [1] designates the code specifying the write operation as first. And the Director concedes that the steps in paragraphs [3] and [4] (providing the first bit of data and then providing the second, respectively) must proceed in order.
10
Case: 12-1480
Document: 33
Page: 15
Filed: 01/16/2013
The Director instead debates whether steps one and two must be performed in sequence. Claims may implicitly require[ ] a specific order. PTO Br. 17. That was true in Mantech Environmental Corp. v. Hudson Environmental Services, Inc., 152 F.3d 1368 (Fed. Cir. 1998) (Rambus Br. 27), the Director asserts, because the order was required for the claim to make sense; each step simply could not be performed before the prior step had occurred. PTO Br. 18. The same is true here: Claim 1s method would not function if its steps were performed out of order. For example, it would make no sense to send the signal in paragraph [2], telling the memory device when to begin sampling (transferring data into memory) for a write operation, before the memory device was told (in paragraph [1]) that the operation would be a write (a read transfers data out of memory). The signal thus would be meaningless to the memory device at that time. And unlike the prior-art 037 patent, JA17614, 27:32-36, the 109 patent does not provide for storing a signal (or delay values) for later reference (assuming dubitante the signal can be a value). With no means of storage, the signal in paragraph [2] would have no effect if sent first. When the first code in paragraph [1] specifying the write operation arrived later, the memory device would have no means of recalling when to begin sampling. Similarly, the writing of the first and second bits of data to the memory device in paragraphs [3] and [4]which the Director concedes must be performed in order, PTO Br. 19could not occur before the
11
Case: 12-1480
Document: 33
Page: 16
Filed: 01/16/2013
steps in paragraph [1] specifying the write operation and paragraph [2] telling the memory device when to begin sampling. The Directors assertion that the claim language does not require a specific order, PTO Br. 17, is thus incorrect. The invention could not function with paragraph [2] being executed before paragraph [1], or paragraphs [3] and [4] ahead of the others (as the specification makes clear, see Rambus Br. 28-31). Because the method could not be implemented in the Directors paragraph [2]-[1]-[3]-[4] sequence, the Boards ruling cannot stand.4 2. The Director also urges that decouplingsending the signal to begin
sampling separate from the write commandis not required because [c]laim 1 does not recite the term decouple, . . . separate or subsequent to[,] or otherwise require that the signal and write command be performed separately. PTO Br. 1415. As explained above, however, claim 1 presents a series of separate and
sequential steps that grammatically and logically follow one after another. Claim 1 recites providing separately for each stepproviding the write command, then providing the signalnot one step of providing the code and signal together. It is no more proper to combine those steps than to scramble them. The
specificationthe single best guide to the meaning of a disputed term, Phillips, ____________________________ The Director asserts that claim 1 contain[s] a write operation step that is clearly out of order, PTO Br. 18-19, but fails to identify which step or why. 12
4
Case: 12-1480
Document: 33
Page: 17
Filed: 01/16/2013
415 F.3d at 1315repeatedly specifies that Claim 1s signal is sent by the controller to the DRAM separate from and subsequent to the write command. A57-58, 8:65-9:2; pp. 13-14, infra; Rambus Br. 28-29. The Director re-interprets the specifications use of separate from or decoupled. Invoking a section titled Decoupled Data Transfer Control Information, the Director urges that decoupling of timing information does not require that various types of control information be sent separately; it merely requires that the timing of the data control information be flexible so that the timing between the request to begin a write operation and the actual data transmission vary. PTO Br. 15 (quoting JA58, 10:25) (emphasis added). But decoupled is used in the specification to require that the information be sent separately, A58, 10:40-43; it does not, as the Director would have it, mean coupled but flexible. That cannot be dismissed as an effort to import a limitation from the specification into the claim. PTO Br. 16. It is the specifications description of the decoupled method in claim 1 itself. See pp. 13-14, infra; Rambus Br. 26-29. The Directors focus on varying the time between the write command and the actual data transmission also looks to the wrong events. The invention does not require separating the memory devices receipt of the write command from its initiation of data transfer. It specifies decoupling the controllers provision of the write command from its later provision of a signal to begin sampling; the
13
Case: 12-1480
Document: 33
Page: 18
Filed: 01/16/2013
controllers write command and signal are decoupled, not just the memorys response thereto. Thus, the DRAM is configured to start . . . transmission of data based on data transfer control information sent by the controller to the DRAM separate from and subsequent to the transmission of the command control information. A57-58, 8:65-9:2 (emphasis added); A58, 9:26-27 (transmitting data transfer control information separate[ly] from the command control information (emphasis added)); A58, 10:41-43 (signal which controls the timing of the data transfer associated with a request packet is sent separately from the command control information to which it corresponds (emphasis added)). C. The 109 Patent Does Not Claim the Prior Art It Distinguishes
By reading signal expansively and overlooking claim 1s separate and sequential steps, the Board and the Director concluded that the 109 patent reads on the prior-art Farmwald 037 patent Rambus had filed 5 years earlier. But that result makes their construction more unreasonable: The inventors expressly distinguished that prior art; it is implausible that they were simultaneously claiming it. See CCS Fitness, Inc. v. Brunswick Corp., 288 F.3d 1359, 1366-67 (Fed. Cir. 2002); Rambus Br. 31-37. And if Rambus had meant to claim its own prior-art patent, it would have claimed priority to it; instead, Rambus distinguished it. The Director denies that the 037 patent was the prior art being distinguished. See PTO Br. 30-33. But the prior art the 109 patent distinguishes
14
Case: 12-1480
Document: 33
Page: 19
Filed: 01/16/2013
utilizes delay value[s] stored in a register within the DRAM. A58, 10:35-36. That describes the 037 patent precisely, as evidenced by the portions of the 037 patent the Director cites. See, e.g., PTO Br. 27-28 (retrieving previously stored delay values). Neither the Board nor the Director identified any other art to which it might refer.5 Citing the 109 patent specifications statement that the fixed timing between requests and data transmissions render[ed] the prior art systems inflexible, PTO Br. 16 (quoting JA59, 11:1-2), the Director urges that the specification could not have been distinguishing the 037 patent, which (according to the Director) was flexible and variable, id. at 30. But flexible is a relative term. The 037 patent may have been flexible compared to the art that preceded it; but it is inflexible compared to the 109 patent. Before the invention disclosed in the 037 patent, asynchronous memory devices largely began writing whenever possible following receipt of the write command; timing was largely uncontrolled. Rambus Br. 4-8. The 037 patents synchronous, register-based, delay-value method represented an advance in flexibility. By selecting among op codes when sending the write command, the con____________________________ The Director urges that Farmwald cannot be the prior art distinguished in the 109 patent because Farmwald mirrors the alternate embodiment, which Rambus has not contended is prior art. PTO Br. 31. That is incorrect: Unlike Farmwald, that alternate embodiment does not store delay values in registers for later usea feature identified with the prior art. See A58, 10:52-67; Rambus Br. 31-37. 15
5
Case: 12-1480
Document: 33
Page: 20
Filed: 01/16/2013
troller could defer data transfer for any of the periods previously stored as delay values in the memory devices access-time registers. Id. at 7-8. Timing choices, however, were constrained to values that had been previously stored. A58, 10:3339. Choicesand flexibilitywere limited. Moreover, because the timing of data transfer could be determined only when the operation was specified, timing became fixed the moment the write command was sent. A58, 10:27-28, 10:57-60. That too limited flexibility: Once the memory controller sent an op code indicating a write command and selecting an access-time register with a stored delay value, the timing of data transfer was locked in; the controller could not adjust timing later. Rambus Br. 49-50. That inflexibility (timing becoming fixed when the write command is sent) was overcome only by sending the signal indicating when to sample separate from and subsequent to the write command. Separating the timing signal from the write command, and thereby overcoming the identified inflexible aspects of Farmwald, is what the inventors actually invented and intended to envelop with the claim. Renishaw PLC v. Marposs Societa per Azioni, 158 F.3d 1243, 1250 (Fed. Cir. 1998). In any event, the Directors focus on flexibility is mistaken. See Rambus Br. 50-53. There may be multiple ways in which succeeding generations of technology have been structured to be flexible over what came before. PTO Br.
16
Case: 12-1480
Document: 33
Page: 21
Filed: 01/16/2013
15. But that does not mean the 109 patent should be construed to encompass all of them, and certainly not the specific prior-art method it distinguishes. D. The Directors Reliance on the Delay Value Embodiment Is Misplaced
Ultimately, the Director invokes an alternate (the so-called delay value) embodiment to justify the Boards decision. But the mention of such an embodiment cannot justify rewriting claim 1s terms to encompass it. The embodiment was claimed in a parent patent, and that patents terms belie the Boards overexpansive constructions. 1. The alternate embodiment lies outside claim 1 because it does not use
a signal (which initiates action) but instead sends a value specifying how long to defer action. And it lacks the critical feature of decoupling because it sends the write command together with the value used to control when sampling begins. Rambus Br. 42-46; pp. 12-14, supra. The Director errs in rewriting claim 1 to encompass the alternate embodiment. See Rolls-Royce, PLC v. United Techs. Corp., 603 F.3d 1325, 1335 (Fed. Cir. 2010). The PTO urges that the alleged delay value embodiment does decouple[ ]. That is true only if one accepts the Directors sui generis definition of decoupling, i.e., varying the timing between the write command and the actual transfer of data. PTO Br. 20. As explained above (pp. 12-14), the 109 patent distinguishes between (a) decoupling the signal from the write command and 17
Case: 12-1480
Document: 33
Page: 22
Filed: 01/16/2013
It
explains that decoupling means disconnecting, such that the signal that controls the timing of the data transfer associated with a request packet is sent separately from the command control information (i.e., the operation code). A58, 10:40-43. And it distinguishes the ability to vary the timing of data transfer from transaction to transaction, describing it as another aspect of the invention. Id. The delayvalue embodiment may also allow the time between the write command and data transfer to vary for different write operations, but the delay value and write command are provided in the same request packet for each operation; they are not decoupled. A58, 10:52-60. The absence of that decoupling in the alternate embodiment, moreover, eliminates the flexibility the 109 patent sought to achieve: It prevents the controller from determining when data transfer should occur separately and after having sent the write command in light of changing conditions, precisely the advance the inventors claimed. Rambus Br. 12-13; pp. 15-17, supra.6 Further, while claim 1 requires a signal, A74, 42:61-63, the alternate embodiment is described as operating without the use of strobe or terminate ____________________________
6
Contrary to the Directors assertion, PTO Br. 21, Rambus does not contend that the delay-value embodiment is the prior art that the 109 patent distinguishes. It simply shares characteristics with the prior artuse of a delay value and a failure to decouple the timing signal from the write commandthat preclude it and the prior art from falling within claim 1. See Rambus Br. 50. 18
Case: 12-1480
Document: 33
Page: 23
Filed: 01/16/2013
signals, using a delay value instead, A58, 10:55-58. The Director urges that the specification describes the embodiment as operating without the use of strobe or terminate signals, not without signals. PTO Br. 20-21. But those are the only signals the specification identifies. Besides, the 109 patent repeatedly distinguishes between signals and delay values, see Rambus Br. 45, and the Director can point to nothing in the patent equating the two. The fact that the alternate embodiment is described in terms of values, not signals, undermines any suggestion it was claimed. The Director responds that a delay value satisfies claim 1s limitation of a signal that indicates when the memory device is to begin sampling, A74, 42:61-63, because the delay value indicates when the data will be sent, PTO Br. 21. But that assumes the Boards overbroad definition of signal. Claim 1 recites a signalas opposed to a code or databecause the memory device begins sampling [i]n response to the signal itself, A64, 21:31-36; the signal is not a value but a signal to the memory device to begin sampling. In the alternate embodiment, by contrast, the memory device does not begin sampling in response to transmission of the delay value; the delay value included in the request packet indicates that sampling should be deferred until a designated time relative to the time at which the request packet is sent without need for a separate signal.
19
Case: 12-1480
Document: 33
Page: 24
Filed: 01/16/2013
A58, 10:57-60. The delay value embodiment thus operates differently from the signal described in the patent. Rambus Br. 29-31. Rambus is mindful of this Courts caution regarding constructions that would exclude a preferred embodiment. MBO Labs., Inc. v. Becton, Dickinson & Co., 474 F.3d 1323, 1333 (Fed. Cir. 2007) (emphasis added); Invitrogen Corp. v. Biocrest Mfg., L.P., 327 F.3d 1364, 1369 (Fed. Cir. 2003). But there is no such concern here in excluding the alternate delay-value embodiment, A58, 10:52, because the 109 patents primary, signal-based embodiments are still claimed. [T]his [C]ourt, moreover, must not allow the disclosed embodiment to outweigh the language of the claim, especially when the [exclusionary] construction is supported by the intrinsic evidence. Rolls-Royce, 603 F.3d at 1334. Yet the PTO urges that result here. 2. The alternate embodiment, moreover, was claimed by a parent patent,
Patent No. 5,748,914. Rambus Br. 46-48. Even if that does not by itself preclude the alternate embodiment from being claimed in the 109 patent, PTO Br. 22-24, the terms of the 914 patents claims do. The 914 patent too distinguishes between the signal (recited in the 109 patent) and the delay values the alternate embodiment and prior art employ. Rambus Br. 46-48.7 ____________________________ The Director contends that reliance on the 914 patent is waived. PTO Br. 23. Where the patentee has asserted a particular construction of its claim terms below, 20
7
Case: 12-1480
Document: 33
Page: 25
Filed: 01/16/2013
Claims 7 and 8 of the 914 patent recite two different methods for controlling when sampling begins. A17658, 43:53-54, 43:58-59. Claim 7 recites transmitting a delay value in the control information. A17658, 43:53-57 (emphasis added). Claim 8, by contrast, recites transmitting a strobe signal a selected number of clock cycles after transmitting the control information. A17658, 43:58-64 (emphasis added). The 914 patents separate use of delay values and signals as distinct types of transmissions in separate claims that are alternatives to each other makes it clearer still that those different words . . . have different meanings and scope. Andersen Corp. v. Fiber Composites, LLC, 474 F.3d 1361, 1369 (Fed. Cir. 2007) (quotation marks omitted); see also Omega Engg, Inc., v. Raytek Corp., 334 F.3d 1314, 1334 (Fed. Cir. 2003) (rule applicable to terms in the same patent or related patents). II. The Director Cannot Salvage the Boards Anticipation Analysis Even accepting the PTOs over-expansive view that signal encompasses delay value and disregarding the separateness of claim 1s sequential steps, the Boards anticipation finding cannot be sustained. The 037 patent anticipates claim 1 only if it discloses within the four corners of the document . . . all of the _________________________________________________________________ however, only arguments that change the scope of [the appellants] claim construction positions . . . will be deemed to be waived. Interactive Gift Express, Inc. v. Compuserve Inc., 256 F.3d 1323, 1347-48 (Fed. Cir. 2001). Rambuss arguments concerning the 914 patent are additional support for its position before the Board that the alternate embodiment is not claimed. 21
Case: 12-1480
Document: 33
Page: 26
Filed: 01/16/2013
limitations in claim 1, with all of the limitations arranged or combined in the same way as recited in the claim. Net MoneyIN, 545 F.3d at 1371. The Board invoked four options that supposedly catalogue the different ways the delay value in Farmwald . . . corresponds to the claimed signal in the 109 patent. A17528. Farmwalds delay value, the Board stated, is either: (1) (2) (3) (4) in the request packet; stored in an access-register; provided for comparison to a clock; and/or implicitly used to generate another implicit signal after the comparison to signify a match.
A13. But the Board did not explain, much less cite substantial evidence showing, where the 037 patent discloses those four options or that they are arranged or combined in the same way as claim 1s signal. The Director unsuccessfully attempts to recharacterize the Boards first option (placing the delay value in the request packet). And he all but abandons the Boards other three optionsall of which defy claim 1s text. A. Option (1)In the Request PacketFails
The Board found anticipation first by asserting that Farmwald anticipates claim 1s signal by placing a delay value in the request packet. A13. But that misreads Farmwald. Even if claim 1 did provide for sending a delay value in the request packet with the write commandas opposed to sending a signal
22
Case: 12-1480
Document: 33
Page: 27
Filed: 01/16/2013
separate from and subsequent to the write command (decoupling them)the Farmwald 037 patent never discloses placing its delay value in the request packet. See Rambus Br. 52. The Director does not suggest otherwise. 1. Instead, the Director revises the Boards opinion, claiming that the
Board found that Farmwald discloses a request packet that controls the timing of write data either by directly select[ing] a certain register in the slave DRAM memory device which stores the (delay value) timing information or indirectly by indicat[ing] pre-selected (delay value) access times. PTO Br. 28 (quoting JA8); see also id. at 34. That was not the Boards basis for anticipation. The Board found anticipation based on Farmwalds delay value being placed in the request packet. A13 (emphasis added). The portion of the opinion that the Director partially quotes (from the Boards background discussion, see A8) does not refer to the delay value being in the request packet. It states that data transfer timing information arrives in the request packet in the form of an op code, which is then used to select a delay value that has been previously stored in a programmable access-time register, or a delay value in a fixed register previously associated with that op code. Id. (emphasis added); see JA17605, 9:31-10:5, JA17606, 11:31-39 (chart); Rambus Br. 52. That distinction is critical. The Directors theorythat claim 1s signal is satisfied by the 037 patents inclusion of an op code in a request packet, which in turn
23
Case: 12-1480
Document: 33
Page: 28
Filed: 01/16/2013
selects a location to find a delay value that is not in the request packetdoes not appear in the Boards decision, and thus may not be the basis for affirmance. In re Thrift, 298 F.3d 1357, 1367 (Fed. Cir. 2002); In re De Blauwe, 736 F.2d 699, 705 n.7 (Fed. Cir. 1984). The new theory is also incorrect. Claim 1 recites providing a signal to the memory device, wherein the signal indicates when the memory device is to begin sampling write data. A74, 42:61-63 (emphasis added). Nothing in the 037 patents request packet constitutes a signal that itself indicates when the memory device is to begin sampling. The Director may be claiming that the op code in the request packet is the signal, but the op code specifies the operation and refers the memory device to an access-time register; the memory device then looks to that register to find a previously stored delay value. JA17605, 9:31-10:5. The Director asserts that the data block transfer occurs at a later time specified in the request packet control information. PTO Br. 34 (quoting JA17605, 9:18-19). As the 037 patents next sentence makes clear, however, the request packet does not itself specify when to the memory device. Instead, [t]he time after which a data block is driven onto the bus lines is selected from values stored in slave access-time registers that are part of the memory device. A17605, 9:23-25. That difference is fatal. To anticipate, the 037 patent must disclose the 109 patents claimed elements, arranged or combined . . . in the same way as recited
24
Case: 12-1480
Document: 33
Page: 29
Filed: 01/16/2013
in the 109 patent itself. Net MoneyIN, 545 F.3d at 1371. The difference between the arrangement and use of the 109 patents signal, and the 037 patents use of an op code to select a register that contains a previously stored delay value, could not be more stark. The 109 patents signal is akin to telling someone to board their train because it is leaving the station. The 037 patents op code is like telling someone to consult a posted schedule, in a specific room of the station, to determine when the train is leaving. The two are not the same, and the Director erred in equating them. 2. The Boards effort to equate Farmwalds delay value with the 109
patents signal independently fails because the two inventions require a different order. Farmwald discloses transmitting delay values to, and storing the values in, the DRAMs access-time registers before the write request is issued (at initialization). See Rambus Br. 32-33. Claim 1 of the 109 patent reverses the order: It requires that the first code the controller provides to the memory device for a transaction be a code specifying the write operation; the signal comes later. See pp. 10-12, supra; Rambus Br. 26-28. The Director asserts that substantial evidence supports a finding that Farmwald provides a signal at least when the write operation is requested, not before by statically storing a predetermined delay value. PTO Br. 29. That contradicts the Boards finding that Farmwalds step of storing the delay value in
25
Case: 12-1480
Document: 33
Page: 30
Filed: 01/16/2013
an access-time register is one of the options that satisfies claim 1s step of providing a signal, A14; that storage occurs at initialization before any op code is transmitted, JA17603, 6:46-48. And the Director fails to identify any supporting evidence. The 037 patent could not be clearer that the delay value is provided to the memory device and stored in an access-time register before the controller transmits the op code that indicates a write command and identifies the register that should be consulted for timing. Id.; JA17605, 9:31-10:5. The Director urges that the fact that certain delay values are stored within the memory device, before an instruction, does not mean that the signal indicating when to write the instruction is provided before the operation. PTO Br. 30. The 037 patent does provide for the controller, when sending a request packet, to determine when the slave memory device acts. Id. But that does not help the Director. The Director still not does identify what in Farmwald corresponds to the signal, or how that signal indicates when the memory device is to begin sampling in the same form and order as the 109 patents signal. To the extent the Director means to repeat that the op code is the signal, that fails for the reasons above: The Board made no such finding, and the op codewhich directs the memory device to find a pre-stored delay value in an access-time register operates in a fundamentally different way than the 109 patents signal. See pp. 23-25, supra.
26
Case: 12-1480
Document: 33
Page: 31
Filed: 01/16/2013
B.
The Director Effectively Concedes the Failure of the Boards Options (2)-(4)
The Boards remaining alternative[ ] theories of anticipationthat claim 1s signal is satisfied when Farmwalds delay value is (2) stored in an accessregister; (3) provided for comparison to a clock; and/or (4) implicitly used to generate another implicit signal after the comparison to signify a matchalso fail. A13-14. 1. The Board did not trace any of those options within Farmwald
itself, much less compare the processes side-by-side with the method recited in claim 1 of the 109 patent. Nor does the Director. That should be dispositive. See Enzo Biochem, Inc. v. Applera Corp., 599 F.3d 1325, 1332 (Fed. Cir. 2010) (anticipation requires a comparison of the properly construed claim to the prior art). The Director asserts that [t]he Boards enumerated options were not unsupported hypotheticals, but were discussed and substantiated throughout the Boards decision. PTO Br. 34. But those options were not substantiated by the Board; they were merely repeated without support. Rambus Br. 53-60. And the Director essentially does the same. While he asserts that each of the Boards options in Farmwald is a signal that may be provided separately, he neither explains why they are arranged as in the 109 patent nor provides supporting citations to the Farmwald patent. PTO Br. 35. Rambuss explanation (Br. 53-60) of why the Boards alternative options in Farmwald constitute entirely different methods, 27
Case: 12-1480
Document: 33
Page: 32
Filed: 01/16/2013
with parts distinct from, and that operat[e] in a different way from, claim 1 of the 109 patent, Net MoneyIN, 545 F.3d at 1370, remains undisputed. 2. Besides, options (2)-(4)storing the delay value in the access-
register; retrieving it; and comparing it to a clock value and generating another (implicit) signal, A14do not meet claim 1s limitation of providing a signal to the memory device, A74, 42:61 (emphasis added). See Rambus Br. 54, 56-58. The Director does not dispute that storage, retrieval, and comparison of any delay value in Farmwald all occur within the memory device; the access-time register is part of the memory device. See PTO Br. 35-36. The Director urges, however, that the limitation of providing a signal to the memory device encompasses a signal provided from one component of the memory device to another, id. at 35i.e., that providing to can mean finding within. That is incorrect. Rambus Br. 54. The specification is clear that, for
information to be provided to the memory device, it must come from outside the memory device; for example, by being sent by the controller to the DRAM. A57-58, 8:67-9:1. Similarly, one would not say that moving files within an office is the same as moving files to an office. This Court has ruled that a limitation of supplying drying gas to the process chamber could be satisfied if the condenser providing that gas is outside the processing chamber, but not if the condenser is inside the processing chamber. Semitool, Inc. v. Dynamic Micro
28
Case: 12-1480
Document: 33
Page: 33
Filed: 01/16/2013
Sys. Semiconductor Equip. GMBH, 444 F.3d 1337, 1348 (Fed. Cir. 2006) (emphasis added). The Director urges that Semitool is distinguishable based on specific facts, PTO Br. 35 n.9, but never explains why the identified facts make a difference. Semitool recognized that supplying gas from a component already within a vessel is not supplying gas to the vessel. The same is true here. If something comes from within the memory device, it cannot be said to have been provided to the memory device as claim 1 requires. For each of claim 1s four steps, the thing provid[ed] to the memory device originates outside it. Paragraph [1]s control information comes from outside the memory device; so does the data provid[ed] to it in paragraphs [3] and [4]. See p. 10, supra. The Director offers no reason providing a signal to in paragraph [2] should be different. Yet again, the Director and the Board adopt a construction that defies consistent usage in the specification and the claim. CONCLUSION The Court should reverse the Boards rejection of the 109 patents claims.
29
Case: 12-1480
Document: 33
Page: 34
Filed: 01/16/2013
January 16, 2013 Kara F. Stoll FINNEGAN, HENDERSON, FARABOW, GARRETT & DUNNER, LLP 901 New York Avenue, N.W. Washington, D.C. 20001 (202) 408-4000
Respectfully submitted, John M. Whealan 4613 Merivale Road Chevy Chase, MD 20815 (202) 994-2195 /s/ Jeffrey A. Lamken Jeffrey A. Lamken Counsel of Record Michael G. Pattillo, Jr. MOLOLAMKEN LLP The Watergate, Suite 660 600 New Hampshire Avenue, N.W. Washington, D.C. 20037 (202) 556-2010
30
Case: 12-1480
Document: 33
Page: 35
Filed: 01/16/2013
CERTIFICATE OF SERVICE I certify that on January 16, 2013, the foregoing Reply Brief for Appellant Rambus Inc. was filed electronically with the Clerk of the Court for the U.S. Court of Appeals for the Federal Circuit using the CM/ECF system and served electronically by ECF on the following counsel: Raymond T. Chen Nathan K. Kelley Coke Stewart William LaMarca United States Patent and Trademark Office Office of the Solicitor P.O. Box 1450 Mail Stop 8 Alexandria, Virginia 22213
Case: 12-1480
Document: 33
Page: 36
Filed: 01/16/2013
CERTIFICATE OF COMPLIANCE 1. X This brief complies with the type-volume limitation of Fed. R. App. P. 32(a)(7) because: This brief contains 7,000 words, excluding the parts of the brief exempted by Fed. R. App. P. 32(a)(7)(B)(iii) and Local Rule 32(b). This brief uses a monospaced typeface and contains [state the number of ] lines of text, excluding the parts of the brief exempted by Fed. R. App. P. 32(a)(7)(B)(iii). 2. X This brief complies with the typeface requirements of Fed. R. App. P. 32(a)(5) and the type style requirements of Fed. R. App. P. 32(a)(6) because: this brief has been prepared in a proportionally spaced typeface using Microsoft Word in Times New Roman 14 point font, or this brief has been prepared in a monospaced typeface using [state name and version of word processing program] with [state number of characters per inch and name of type style]. /s/ Jeffrey A. Lamken Jeffrey A. Lamken