`
`
`
`
`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`____________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`____________________
`
`Apple Inc.,
`Petitioner,
`v.
`
`Qualcomm Incorporated,
`Patent Owner
`____________________
`Case IPR2018-01316
`U.S. Patent No. 8,063,674
`_____________________
`
`PATENT OWNER SUR-REPLY
`
`
`
`
`
`
`
`
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`
`
`
`
`
`B.
`C.
`
`Table of Contents
`Introduction ..................................................................................................... 1
`Petitioner Does Not Disagree That AAPA Is Not Proper Prior Art In
`Inter Partes Review Proceedings ................................................................... 1
` Neither The Reply Nor Dr. Horst’s New Simulation Results Rebut
`Qualcomm’s Showing That The POSA Would Not Combine The
`Alleged AAPA And Majcherczak .................................................................. 2
`A.
`Petitioner’s Argument About The Alleged “Explicit”
`Motivation To Combine Is Erroneous .................................................. 2
`The Reply’s Criticism Of Dr. Pedram’s Testimony Is Misplaced ....... 7
`Dr. Horst’s Simulations Do Not Prove That The POSA Would
`Be Motivated To Combine The Alleged AAPA And
`Majcherczak ......................................................................................... 8
`1.
`The Simulation Results Are Unreliable And Should Be
`Disregarded ................................................................................ 9
`Dr. Horst Selected Unrealistic Parameter Values That
`The POSA Would Never Use In The Proposed
`AAPA/Majcherczak Combination ........................................... 11
`Dr. Horst Cherry Picked Parameter Values Favorable For
`Petitioner Without Providing Any Rationale For Their
`Selection ................................................................................... 16
`The Simulations Fail To Show That The
`AAPA/Majcherczak Combination Does Not Result In
`Increased Leakage Current ...................................................... 17
`The Single Simulation Result For AAPA/Majcherczak
`Reported By Dr. Horst Does Not Rebut Dr. Pedram’s
`Showing ................................................................................... 18
` Petitioner’s Extensive New Evidence And Argument Cannot Cure
`The Petition’s Failure To Show That The POSA Would Be Motivated
`To Combine Steinacker, Doyle, and Park .................................................... 19
`
`2.
`
`3.
`
`4.
`
`5.
`
`
`
`-ii-
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`Petitioner’s New Reasons For Why The POSA Would
`Allegedly Combine The References Are Deficient ........................... 21
`Dr. Horst’s New Simulation Results Do Not Rebut
`Qualcomm’s Showing That The Hypothetical
`Steinacker/Doyle/Park Circuit Has Leakage And Glitch Current
`Problems ............................................................................................. 25
`1.
`Dr. Horst’s Simulation Results Contain Significant Errors ..... 25
`2.
`The Parameter Values Selected For The Proposed
`Steinacker/Doyle/Park Combination Are Unrealistic .............. 25
`Dr. Horst’s Cherry Picked Parameter Values Show That
`His Selections Are Biased ........................................................ 26
`The Simulations Fail To Show That The Proposed
`Steinacker/Doyle/Park Circuit Does Not Result In
`Increased Leakage Current ...................................................... 26
`The Single Reported Simulation Result Does Not Rebut
`Dr. Pedram’s Showing ............................................................. 27
`Conclusion .................................................................................................... 27
`
`3.
`
`4.
`
`5.
`
`A.
`
`B.
`
`
`
`
`
`
`
`-iii-
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`
`
`
`Introduction
`Petitioner’s reply introduces unpersuasive and belated arguments and
`
`evidence that cannot salvage the petition. The Board should confirm the
`
`patentability of claims 8, 9, 12, 13, and 16-22.
`
`
`
`
`
`Petitioner Does Not Disagree That AAPA Is Not Proper Prior Art In
`Inter Partes Review Proceedings
`Qualcomm’s response showed that Grounds 2(a) and 2(b) are improper
`
`because the America Invents Act (AIA) does not permit IPR based on so-called
`
`applicants admitted prior art (AAPA). Paper 12 at 18-20.
`
`
`
`The reply does not disagree that AAPA is not proper prior art for IPR
`
`proceedings. Paper 16 at 1-2. In fact, the reply never makes the affirmative
`
`statement that AAPA should be considered prior art in IPRs. See id. Instead,
`
`Petitioner merely points out that the Institution Decision followed the logic
`
`articulated in a previous IPR where a different panel found AAPA to be prior art.1
`
`Id. But Petitioner carefully avoids endorsing the previous panel’s approach or ever
`
`stating affirmatively that AAPA is proper prior art. The reason for this is clear:
`
`
`
`1 The previous panel decision cited by Petitioner is distinguishable from the present
`case. In the cited case, AAPA was relied on as a secondary reference in an
`obviousness ground. One World Techs., Inc. v. Chamberlain Group, Inc., IPR2017-
`00126, Paper 56 at 6 (PTAB Oct. 24, 2018). By contrast, in the present case,
`Petitioner attempts to rely on AAPA as a primary reference.
`1
`
`
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`Counsel for Petitioner is currently taking the position before the Board and the
`
`Federal Circuit that AAPA is “not ‘prior art consisting of patents or printed
`
`publications’ and, thus is ineligible for inter partes review.” Ex. 2004 at 3.
`
`
`
`Qualcomm agrees with the position taken by Petitioner’s counsel in these
`
`other proceedings.2 Petitioner’s reliance on AAPA is improper, and the challenged
`
`claims should be held patentable over Grounds 2(a) and 2(b).
`
` Neither The Reply Nor Dr. Horst’s New Simulation Results Rebut
`Qualcomm’s Showing That The POSA Would Not Combine The
`Alleged AAPA And Majcherczak
`A.
`Petitioner’s Argument About The Alleged “Explicit” Motivation
`To Combine Is Erroneous
`The reply argues that a POSITA would have been motivated to integrate the
`
`
`
`feedback transistor M6 of Majcherczak’s voltage detector into the alleged AAPA to
`
`enable stabilizing of the detection device through hysteresis, as described in
`
`Majcherczak. Paper 16 at 2-6. But Petitioner is wrong because a POSA faced with
`
`
`
`2 Permitting petitioners to rely on AAPA in IPR proceedings is improper because,
`among other reasons, it dissuades patent applicants from including a background
`section in their patent applications. Further, Petitioner’s reliance on the alleged
`AAPA here is especially improper because it is being applied as a primary reference.
`Obviousness is judged by putting oneself in the mind of the POSA—and then asking
`whether that person, the POSA, would be motivated to combine the prior art to reach
`the claimed invention. Here, Petitioner puts itself not in the mind of the POSA, but
`rather in the mind of the inventor as a starting point. Applying the alleged AAPA as
`the primary reference inherently leads to hindsight bias because it starts the
`obviousness inquiry from the wrong context: the inventor’s mind, not the POSA’s.
`2
`
`
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`the problems articulated in the alleged AAPA—high leakage current and slow
`
`switching times—would never seek to add hysteresis.
`
`
`
`The primary reference of Petitioner’s Ground 2a is the alleged AAPA. See,
`
`e.g., Paper 2 at 51-53. For this ground, Petitioner proposes “integrat[ing] … the
`
`feedback transistor M6 from Majcherczak’s voltage detector into the POC system 10
`
`of the AAPA” (id. at 53):
`
`
`
`Id. at 52.
`
`
`
`The Background section of the ’674 patent describes that the alleged AAPA
`
`“ha[s] problems with leakage and switching times.” Ex. 1001 at 3:10-11.
`
`Specifically, the patent states that the circuit of Figure 1 (i.e., the alleged AAPA)
`
`suffers from undesirable leakage current that flows from VI/O to ground. Id. at 2:21-
`
`39, 2:63-3:11. The patent further describes that “to reduce this stray power
`
`
`
`3
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`consumption, one solution may be … to decrease the sizes of transistors M1-M3.
`
`By reducing the size of M1-M3, the actual amount of current that can pass through
`
`the transistors is physically limited.” Id. at 2:31-35. But this solution comes at a
`
`cost: “[B]ecause the transistors are now smaller, their switching speeds are also
`
`reduced. The reduced switching speed translates into less sensitivity in detecting
`
`power-up/down of core supply voltage 103 or longer processing time for power-
`
`up/down events.” Id. at 2:35-39; see also id. at 2:63-67. The switching speed can
`
`be increased by using lower-threshold transistors, but such transistors result in higher
`
`leakage current. See id. at 2:67-3:10. The Background section concludes that
`
`“[t]herefore, the conventional solutions still have problems with leakage and
`
`switching times.” Id. at 3:10-11.
`
`
`
`A POSA starting with the alleged AAPA would never turn to Majcherczak for
`
`its alleged “advantageous hysteresis.” Paper 16 at 5. Accordingly, the purported
`
`“explicit motivation in Majcherczak” (id. at 3) is no motivation at all because the
`
`POSA would never even get to this reference.
`
`
`
`First, the addition of Majcherczak’s hysteresis to the alleged AAPA would
`
`make the high leakage current problem identified in the AAPA worse. Qualcomm’s
`
`expert Dr. Pedram showed that Petitioner’s proposed addition of Majcherczak’s
`
`transistor M6 to Figure 1 of the ’674 Patent (the alleged AAPA) results in increased
`
`leakage current. Ex. 2002 at ¶¶69-74. The reply presents computer simulations
`
`
`
`4
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`attempting to show that the AAPA/Majcherczak circuit does not have increased
`
`leakage current, but as explained in Section III.C.2 below, the simulations use
`
`unrealistic parameter values that the POSA would never use, and therefore do not
`
`rebut Dr. Pedram’s showing.
`
`
`
`Second, in view of the switching-speed problem articulated in the alleged
`
`AAPA, the POSA would never seek to add Majcherczak’s hysteresis absent
`
`impermissible hindsight reconstruction. Hysteresis, by its very nature, slows down
`
`detection in an up/down detector because it adds a noise margin and thus delays
`
`(i) power-up detection to occur at higher voltages than normal, and (ii) power-down
`
`detection to occur at lower voltages than normal. The POSA faced with the problem
`
`of slow switching speeds described in the AAPA would thus never consider adding
`
`Majcherczak’s hysteresis absent impermissible hindsight.
`
`
`
`In fact, the addition of the feedback circuit in the ’674 patent invention is
`
`counterintuitive and further evidences the innovative nature of the invention. The
`
`’674 patent invention adds this feedback circuit in a manner that minimizes the effect
`
`on switching speed. See, e.g., Ex. 1001 at 7:4-7. In Majcherczak, by contrast, speed
`
`is not a consideration, and the major concern of the reference is to increase the
`
`stability of its detection device. See, e.g., Ex. 1008 at ¶37. Facing Majcherczak’s
`
`problem of adding stability, i.e., avoiding false detection such as by adding a noise
`
`margin, the intuitive solution would be to add hysteresis, and that is what
`
`
`
`5
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`Majcherczak describes. But the POSA faced with the switching speed problem of
`
`the alleged AAPA would never seek to add hysteresis because doing so would
`
`decrease detection speeds and thus be counterintuitive.
`
`
`
`Third, although Petitioner highlights Majcherczak’s teaching of “enabl[ing]
`
`the proper stabilizing of the detection device” (Paper 16 at 2-4), no portion of the
`
`alleged AAPA suggests that stability is a problem in the AAPA circuit. As explained
`
`above, the alleged AAPA identifies just two problems—leakage current and slow
`
`switching times—and instability of the detection device is not one of them.
`
`Accordingly, the POSA starting with the AAPA would have no reason to look to
`
`Majcherczak for its alleged teaching of proper stabilization of the detection device
`
`through hysteresis.
`
`
`
`Petitioner argues that the disadvantages resulting from the combination of the
`
`alleged AAPA and Majcherczak amount to a “teaching away” defense. Paper 16
`
`at 7. But that is wrong. As explained above, Petitioner’s one alleged motivation to
`
`combine the references—“enabl[ing] the proper stabilizing of the detection
`
`device”—is deficient, and there is no reason why a POSA would be motivated to
`
`make Petitioner’s proposed combination absent a hindsight desire to create a circuit
`
`reading on claims of the ’674 Patent. Petitioner has therefore failed to establish a
`
`prima facie case of obviousness.
`
`
`
`6
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`B.
`The Reply’s Criticism Of Dr. Pedram’s Testimony Is Misplaced
` Petitioner faults Dr. Pedram for not “cit[ing] to any references to support the
`
`
`
`list of disadvantages … arising from the Petition’s combination of AAPA with
`
`Majcherczak.” Paper 16 at 7. But the disadvantages described by Dr. Pedram—
`
`increased leakage and glitch current and the DC fighting condition—are all well-
`
`known phenomena that would be immediately apparent to the POSA upon
`
`considering the proposed combination. See Ex. 2002 at ¶¶67-85. Petitioner sets
`
`forth no argument that the disadvantages described by Dr. Pedram are obscure or not
`
`generally understood, and such an assertion would be wrong.
`
`
`
`Further, the reply cites no statute, rule, or caselaw to support the proposition
`
`that an expert’s opinion must be corroborated by independent references.
`
`Dr. Pedram has over 30 years of experience in the fields of integrated circuit devices
`
`and electronics, and his testimony on the disadvantages that would be apparent to
`
`the POSA is informed by this extensive experience. See Ex. 2002 at ¶¶4-19.
`
`Petitioner’s argument about the lack of corroborating references does not disprove
`
`Dr. Pedram’s informed testimony.
`
`
`
`The reply cites the Federal Circuit’s statement in Medichem, S.A. v. Rolabo,
`
`S.L. that “a given course of action often has simultaneous advantages and
`
`disadvantages, and this does not necessarily obviate motivation to combine.” 437
`
`F.3d 1157, 1165 (Fed. Cir. 2006). Here, however, the purported advantage of
`
`
`
`7
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`Majcherczak is no advantage at all. As explained in Section III.A above, the alleged
`
`AAPA identifies leakage current and slow switching speed as the problems of the
`
`conventional solutions, and the POSA faced with these problems would never turn
`
`to Majcherczak. The additional disadvantages identified by Dr. Pedram—increased
`
`glitch current, a DC fighting condition, etc.—further obviate any alleged motivation
`
`to combine.
`
`C. Dr. Horst’s Simulations Do Not Prove That The POSA Would Be
`Motivated To Combine The Alleged AAPA And Majcherczak
` The reply argues that each of the problems of the AAPA/Majcherczak
`
`
`
`combination articulated by Dr. Pedram can be overcome through careful selection
`
`of parameter values. Paper 16 at 7-12. To demonstrate this, Petitioner’s declarant
`
`Dr. Horst performed simulations of the proposed combination of AAPA and
`
`Majcherczak using unrealistic parameters and presents a single simulation case that
`
`allegedly avoids the problems described by Dr. Pedram. Ex. 1018 at ¶¶21-31.
`
`
`
`As an initial matter, Petitioner’s argument misunderstands the proper inquiry
`
`for obviousness. The relevant question is whether the POSA would be motivated to
`
`combine the alleged AAPA and Majcherczak to reach the claimed invention. As
`
`explained in Section III.A above, the POSA starting with the alleged AAPA would
`
`never look to Majcherczak in the first place. The reply, however, incorrectly
`
`assumes that the POSA would combine the references and works backwards to find
`
`
`
`8
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`a set of parameters that could allegedly avoid the problems articulated by
`
`Dr. Pedram. Paper 16 at 7-12. This approach is erroneous because the POSA never
`
`would have made the combination in the first place.
`
`
`
`In any case, Dr. Horst’s simulations do not prove that the POSA would be
`
`motivated to combine the alleged AAPA and Majcherczak to reach the claimed
`
`invention. This is explained below.
`
`1.
`
`The Simulation Results Are Unreliable And Should Be
`Disregarded
`Dr. Horst’s supplemental declaration filed with the reply (Ex. 1018) contains
`
`
`
`numerous errors. For instance, the declaration shows simulation results for the
`
`proposed Steinacker/Doyle/Park circuit that are identical to those for the
`
`AAPA/Majcherczak circuit, even though the respective circuits have significant
`
`differences, such as the 135-ohm resistor in Doyle. Compare results shown at
`
`Ex. 1018, page 9, to results at page 15. Given these differences, the simulation
`
`results for the two circuits should not be the same, and the results are erroneous for
`
`at least this reason.
`
`
`
`Recognizing this error, Dr. Horst prepared a corrective declaration including
`
`a new set of graphs intended to replace those on page 15 of the supplemental
`
`declaration. Ex. 2007. But even with the corrective declaration, Dr. Horst’s
`
`simulation results are still littered with errors. For example, the results for Figure 4
`
`
`
`9
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`of the ’674 patent in the corrective declaration purport to show current flowing
`
`through a transistor “M2”:
`
`Ex. 2007 at 2. But the circuit simulated by Dr. Horst for Figure 4 of the ’674 patent
`
`contains no transistor labeled “M2”:
`
`
`
`
`
`
`
`10
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`Ex. 1018 at 14. Accordingly, it is not clear what current the “Is(M2)” waveform is
`
`intended to represent or whether it is even related to this simulation at all. Dr. Horst
`
`has not corrected this error.
`
`2.
`
`Dr. Horst Selected Unrealistic Parameter Values That The
`POSA Would Never Use In The Proposed
`AAPA/Majcherczak Combination
`A POSA starting with the alleged AAPA and the problems articulated therein
`
`
`
`would never use the parameter values selected by Dr. Horst for his simulations. His
`
`parameter values are unrealistic, and the simulation results therefore prove nothing.
`
`a.
`
`The FET Threshold Voltage Of 2.3V Causes The
`Problem Of Slow Detection Speed Described In The
`Alleged AAPA
`In Petitioner’s proposed AAPA/Majcherczak combination, the primary
`
`
`
`reference is the alleged AAPA, and Majcherczak is used only for the addition of the
`
`single feedback transistor M6. Paper 2 at 43-45. As described in Section III.A
`
`above, the “conventional solutions” of the alleged AAPA “have problems with
`
`leakage and switching times.” Ex. 1001 at 3:10-11. The Background section of the
`
`’674 patent specifically indicates that higher-threshold transistors address the
`
`problem of leakage current but result in an up/down detector with slow detection
`
`speeds. See id. at 2:63-3:11. Accordingly, the POSA starting with the alleged
`
`AAPA and seeking to add Majcherczak’s feedback resistor M6 would never
`
`implement the resulting up/down detector with higher-threshold transistors because
`
`
`
`11
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`that would result in an incomplete solution which fails to remedy one of the problems
`
`explicitly called out in the patent.
`
`
`
`Yet this is exactly what Dr. Horst did in his simulations: He selected FET
`
`threshold voltages that are unrealistically high and result in the problem of slow
`
`switching speed articulated in the ’674 patent. Dr. Horst’s simulations test the ability
`
`of the AAPA/Majcherczak circuit to detect power up and down of a 3.3V supply
`
`(Vcore). Ex. 1018 at ¶6. In the simulations, Dr. Horst “set a VTO value of 2.3 volts
`
`for the NMOS devices and negative 2.3 volts for the PMOS devices,” where “‘VTO’
`
`is the turn-on threshold of the transistor” (Ex. 2006 at 88:11-18):
`
`
`
`Ex. 1018 at 14.
`
`
`
`But a FET threshold voltage of 2.3V in an up/down detector with a 3.3V
`
`supply is much too high. Modern semiconductor processes generally allow for four
`
`types of FET threshold voltages: zero-, low-, standard-, and high-threshold. Zero-
`
`threshold voltage transistors have a threshold voltage of 0V. In a system with a 3.3V
`
`supply to be detected, low-threshold and standard-threshold voltages are in the range
`
`
`
`12
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`of approximately 0.7V or less. High-threshold voltages are generally around 1.0V
`
`or less. Dr. Horst’s selected threshold voltage of 2.3V is completely unrealistic, and
`
`a transistor with such an unreasonably high threshold voltage could only possibly be
`
`fabricated—if at all—by a foundry using special, non-standard processes. These
`
`non-standard processes would be very expensive and incompatible with mainstream
`
`CMOS technologies. Also, the 2.3V threshold voltage would make the detector
`
`unsuitable for low-power applications since it would require a power supply voltage
`
`of at least 2.3V to even turn on any transistor, whereas state-of-the-art CMOS
`
`technology generally works with a supply voltage of 1V or less.
`
`
`
`Using this unrealistically high threshold voltage, Dr. Horst argues that the
`
`resulting up/down detector “does not show large ‘leakage current’ as predicted by
`
`Dr. Pedram.”3 Ex. 1018 at ¶26. Even if that is true, the detector would suffer from
`
`the precise problem articulated in the alleged AAPA for detectors employing high-
`
`threshold transistors: slow switching speeds. Dr. Horst’s declaration conveniently
`
`ignores any consideration of switching speed, and the slow switching speeds are not
`
`
`
`3 Dr. Horst’s simulation results do not support his argument. The results presented
`in the corrective declaration show that the power consumption of the proposed
`AAPA/Majcherczak circuit is approximately 12.5% higher than that of Figure 4 of
`the ’674 patent, i.e., ~450µW peak power versus ~400µW. Ex. 2007 at 2. It is thus
`clear that the proposed AAPA/Majcherczak combination addresses neither the
`leakage current problem nor the switching speed problem discussed in the alleged
`AAPA.
`
`
`
`13
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`evident from Dr. Horst’s results, which use a relatively long time scale on the order
`
`of microseconds (µs), such that any switching delays are imperceptible. But the use
`
`of such a high threshold voltage—more than double what is generally considered to
`
`be “high-threshold” voltages—would undoubtedly have this problem. In any event,
`
`Dr. Horst’s use of the 2.3V threshold voltages results in a detector that does not
`
`detect a power-up event until the core voltage has ramped up to 3.0V, and that is
`
`over 90% of the 3.3V operating voltage—unrealistically slow for an up/down
`
`detector.
`
`
`
`In sum, the FET threshold voltage of 2.3V selected by Dr. Horst is unrealistic,
`
`and the POSA starting with the AAPA would never select this high-threshold value.
`
`b.
`
`Dr. Horst Selected Channel Length And Width
`Parameters From A Reference Unrelated To The
`Technology Of The Alleged AAPA
`In performing his simulations, Dr. Horst selected channel length and width
`
`
`
`values for transistors. Ex. 1018 at ¶7. Instead of selecting these values from a
`
`reference related to the technology of the alleged AAPA—i.e., “power-up/down
`
`detectors to generate a power-on/off-control (POC) signal,” see Ex. 1001 at 1:55-
`
`57—Dr. Horst chose values from a reference entirely unrelated to this technology.
`
`The POSA starting with the alleged AAPA and seeking to add Majcherczak’s
`
`feedback transistor M6 would never select the channel length and width values used
`
`in Dr. Horst’s simulations.
`
`
`
`14
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`Dr. Horst’s declaration explains that “[w]here possible, [he] used channel
`
`
`
`length and width parameters based on the Voss reference (Ex. 1022) that was
`
`available prior to the critical date of the ’674 patent.” Ex. 1018 at ¶7. But aside
`
`from the fact that Voss was available prior to the filing date of the ’674 patent,
`
`Dr. Horst provides no reasoning or explanation as to why he selected parameter
`
`values from this reference. See generally id. Dr. Horst mentions that Voss discloses
`
`a power up/down detector (id. at ¶8), but Voss never refers to or describes any such
`
`power up/down detector. Rather, Voss is directed to a buffer for interfacing between
`
`a TTL circuit and a CMOS circuit. See, e.g., Ex. 1022 at 5:8-12.
`
`
`
`Dr. Horst provides no explanation for his decision to select parameter values
`
`from an entirely unrelated technology, and it appears that he did so merely because
`
`they provided results favorable to Petitioner.
`
`c.
`
`The Selected Parameter Values Are Divorced From
`Real-World Considerations
`Dr. Horst acknowledged at deposition that his computer simulations did not
`
`
`
`take into account various real-world considerations. For instance, it is well known
`
`that the threshold voltage of a transistor is determined by the manufacturer according
`
`to its minimum feature size (process) and supply voltage. Dr. Horst conceded,
`
`however, that his simulations enabled him to choose any desired threshold voltage,
`
`even if that voltage was inconsistent with the transistor’s minimum feature size and
`
`
`
`15
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`the supply voltage. Ex. 2006 at 93:3-13. And when asked whether his selected
`
`threshold voltage of 2.3V would be typical for transistors having a 5V supply voltage
`
`and a minimum feature size of 0.8 µm, Dr. Horst conceded that he chose the 2.3V
`
`value simply because it “gave good results.” Id. at 94:5-13
`
`
`
`This testimony shows that Dr. Horst did not consider parameter values that
`
`would be used in real-world settings, and rather, he selected values that “gave good
`
`results” in his computer simulations. This is yet another reason why Dr. Horst’s
`
`simulation parameters are unrealistic and prove nothing.
`
`3.
`
`Dr. Horst Cherry Picked Parameter Values Favorable For
`Petitioner Without Providing Any Rationale For Their
`Selection
`In a number of places, Dr. Horst selected parameter values for his simulations
`
`
`
`without providing any justification for his selections. For instance, Dr. Horst states
`
`that he used length and width values from the Voss reference “for most NMOS and
`
`[PMOS] transistors in [his] simulations.” Ex. 1018 at ¶8. Dr. Horst’s declaration
`
`also states that “[a] different value was chosen for the width for the feedback P-
`
`channel to make this transistor weaker than in the Voss circuit due to the difference
`
`in the circuits.” Id. at ¶9.
`
`
`
`But the declaration provides no explanation as to why Dr. Horst only used the
`
`Voss parameters for “most” transistors and not all of them. Likewise, the declaration
`
`does not explain why he chose “[a] different value … for the width for the feedback
`
`
`
`16
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`P-channel to make this transistor weaker,” nor does the declaration explain which
`
`alleged “difference in the circuits” necessitated this change. See id.
`
`
`
`At deposition, however, Dr. Horst explained the reason for changing the width
`
`of the feedback transistor from what is described in Voss: If Voss’s parameters were
`
`used consistently across the simulations—and not cherry picked and used for certain
`
`transistors, but not others—then the simulation would fail. Ex. 2006 at 101:24-
`
`102:17. This testimony crystallizes one of the key problems with Dr. Horst’s
`
`simulations. He selects parameter values using a given methodology when they suit
`
`his purposes—e.g., selecting parameters from Voss because it was “available prior
`
`to the critical date of the ’674 patent,” see Ex. 1018 at ¶¶7-8—but then disregards
`
`the methodology when its parameters do not yield his desired results. See also id.
`
`(Dr. Horst stating, without any reasoning, that he used LTSpice default values for
`
`some parameters and his own non-default values for other parameters).
`
`4.
`
`The Simulations Fail To Show That The
`AAPA/Majcherczak Combination Does Not Result In
`Increased Leakage Current
`Dr. Horst disputes Dr. Pedram’s showing (Ex. 2002 at ¶¶69-74) that the
`
`
`
`AAPA/Majcherczak combination has a large amount of leakage current from VI/O to
`
`VSS (ground). Ex. 1018 at ¶26. In an attempt to show that the AAPA/Majcherczak
`
`combination has low leakage current, Dr. Horst’s simulations “show the core (input)
`
`
`
`17
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`voltage ramping from 0 to 3.3V and back to zero.” Id. at ¶8. This can be seen, for
`
`instance, in the following panes from Dr. Horst’s simulation results:
`
`
`
`
`
`But ramping the Vcore up to a max value and then pulling it immediately back
`
`down is not an accurate way of measuring leakage current. As Dr. Pedram explained
`
`in his declaration, in up/down detectors, leakage power is primarily a problem when
`
`Vcore is high. Ex. 2002 at ¶¶72-73. Accordingly, one cannot accurately assess the
`
`leakage current in an up/down detector by ramping down the Vcore voltage
`
`immediately upon reaching its max value, as Dr. Horst did. To assess the leakage
`
`current, Dr. Horst should have left the Vcore voltage at the 3.3V level for at least some
`
`time. He did not.
`
`5.
`
`The Single Simulation Result For AAPA/Majcherczak
`Reported By Dr. Horst Does Not Rebut Dr. Pedram’s
`Showing
`Dr. Pedram presented analysis based on fundamental circuit analysis
`
`
`
`principles showing that the proposed AAPA/Majcherczak combination would have
`
`multiple problems that would be recognized by the POSA. Ex. 2002 at ¶¶67-85.
`
`Rather than rebut the substance of Dr. Pedram’s analysis, Dr. Horst presents results
`
`
`
`18
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`for a single computer simulation and argues that the problems “imagined” by
`
`Dr. Pedram are not present. Ex. 1018 at ¶¶21-31.
`
`
`
`But a single simulation result cannot provide an accurate indication of whether
`
`a circuit would operate correctly under realistic, real-world conditions. Typically,
`
`engineers execute thousands of computer simulations to predict the operation of
`
`circuits under various process, voltage, and temperature conditions and then do
`
`testing on the fabricated circuits. In fact, Dr. Horst himself executed many more
`
`simulations than he reported in his declaration. Ex. 2006 at 81:21-83:10, 93:23-
`
`94:4, 94:20-95:11, 102:11-19, 113:20-114:2, 116:21-117:2. Dr. Horst stated at
`
`deposition, however, that he did not save the results for any of these unreported
`
`simulations. See id. Many of these simulations likely show inoperable circuit
`
`results, and Dr. Horst conceded as much at deposition. See id. at 102:11-19 (“Q.
`
`When you ... conducted your simulations with a stronger Pchan2, the simulation
`
`failed; is that correct? A. Yes ….”).
`
`
`
`In the end, a single simulation result cannot rebut Dr. Pedram’s well-reasoned
`
`analysis.
`
` Petitioner’s Extensive New Evidence And Argument Cannot Cure The
`Petition’s Failure To Show That The POSA Would Be Motivated To
`Combine Steinacker, Doyle, and Park
`In its petition, Petitioner provided just two short paragraphs explaining why
`
`
`
`the POSA would allegedly be motivated to combine Steinacker, Doyle, and Park.
`
`
`
`19
`
`
`
`IPR2018-01316
`U.S. Patent 8,063,674
`Paper 2 at 21-22. In its Institution Decision, the Board stated that
`
`we have concerns regarding whether Petitioner’s arguments and
`evidence are based on impermissible hindsight…. Petitioner provides
`generic reasons for combining the various limitations.
`
`Paper 7 at 38.
`
`
`
`In reply, Petitioner submits new argument and over 30 paragraphs of expert
`
`testimony in an attempt to remedy the issues of its cursory motivation-to-combine
`
`argument. Ex. 1018 at ¶¶3-20, 34-49. But a reply is too late to submit this kind of
`
`extensive new evidence and argument. See Office Patent Trial Practice Guide,
`
`August 2018 Update at 14 (“Petitioner may not submit new evidence or ar