Page 78
`
` UNITED STATES PATENT AND TRADEMARK OFFICE
` ____________________________
` BEFORE THE PATENT TRIAL AND APPEAL BOARD
` ____________________________
` DR. MICHAEL FARMWALD and RPX CORPORATION
` Petitioner
`
` v.
` PARKERVISION, INC.,
` Patent Owner
`
` ____________________________
`
` Case IPR2014-00946
` Patent 6,266,518
` Case IPR2014-00947
` Patent 6,061,551
`
` Case IPR2014-00948
` Patent 6,370,371
` VOLUME 2 OF 3
`
` Thursday, June 4, 2015 - 9:25 a.m.
`
` Oral deposition of BRUCE A. FETTE,
`Ph.D., a witness, taken by Petitioner, pursuant
`to Notice, held at the Offices of Sterne,
`Kessler, Goldstein & Fox, P.L.L.C., 1100 New
`York Avenue NW, Washington, DC, before RYAN K.
`BLACK, a Registered Professional Reporter,
`Certified Livenote Reporter and Notary Public
`for the District of Columbia.
`
`Veritext Legal Solutions
`215-241-1000 ~ 610-434-8588 ~ 302-571-0510 ~ 202-803-8830
`
`1
`2
`3
`4
`5
`
`6 7
`
`8
`
`9
`
`10
`
`11
`12
`
`13
`
`14
`15
`16
`
`17
`18
`19
`20
`21
`22
`23
`24
`25
`
`Farmwald and RPX Exhibit 1066, pg. 1
`Farmwald and RPX v. ParkerVision
`IPR2014-00947
`
`

`

`Page 79
`
`Page 81
`
`1 I N D E X
`2 TESTIMONY OF: BRUCE A. FETTE, Ph.D. PAGE
`3 By Mr. Bailey.................................82
`
`4 5 6
`
` E X H I B I T S
`7 EXHIBIT DESCRIPTION PAGE
`8 No. 1054 a graph done by Mr. Bailey........97
`9 No. 1055 a graph done by Mr. Bailey........99
`10 No. 1056 a graph done by Mr. Bailey.......101
`11 No. 1057 demonstrative produced by
`12 ParkerVision.....................108
`13 No. 1058 a hand-drawn graph...............207
`14 No. 1059 the patent owner's preliminary
`15 response regarding the '371
`16 patent...........................229
`17 No. 1059 a document titled Supplemental
`18 Declaration of Bruce A. Fette,
`19 Ph.D., in support to Patent Owner's
`20 Response to Petition.............133
`21 No. 1060 a document titled In Re.:
`22 Sorrells, et al., Appl. No.
`23 09/293,342, Amendment and Reply
`24 under 37 C.F.R. Subsection
`25 1.111............................285
`
`1 A P P E A R A N C E S:
`
`23
`
` OBLON SPIVAK
`4 BY: W. TODD BAKER, ESQ.
`5 1940 Duke Street
`6 Alexandria, VA 22314
`7 703.412.6383
`8 tbaker@oblon.com
`9 Representing - Dr. Michael Farmwald and RPX
`10 Corporation
`11
`12
`13 LAW OFFICE OF JAMES T. BAILEY
`14 BY: JAMES T. BAILEY, ESQ.
`15 504 West 136th Street, Suite 1B
`16 New York, New York 10031
`17 917.626.1356
`18 jtb@jtbaileylaw.com
`19 Representing - Dr. Michael Farmwald and RPX
`20 Corporation
`21 Veritext Legal Solutions
` Mid-Atlantic Region
` 1250 Eye Street NW - Suite 1201
`22 Washington, D.C. 20005
`23
`24
`25
`
`Page 80
`
`Page 82
`
`1 Whereupon --
`2 BRUCE FETTE, Ph.D.,
`3 called to testify, having been first duly sworn
`4 or affirmed, was examined and testified as
`5 follows:
`6 EXAMINATION (Cont'd)
`7 BY MR. BAILEY:
`8 Q. Good morning again, Dr. Fette.
`9 A. Good morning, sir.
`10 Q. A week ago when we were talking,
`11 you mentioned several times the Friis Equation.
`12 When you say Friis Equation, what are you
`13 referring to?
`14 A. In the publications that he made back
`15 late '40s and early '50s, an equation came to be
`16 known as the Friis Equation, by which it was the
`17 standard methodology for calculating noise
`18 figure. It was subsequently adopted by the
`19 IEEE as the standard methodology and equation
`20 for performing calculation of noise figure of
`21 the front end.
`22 Q. So at least, from your opinion,
`23 the Friis Equation is the definition of noise
`24 figure, as opposed to an equation dealing with
`25 how you calculate cascaded noise figure over
`
`2 (Pages 79 - 82)
`
`1 A P P E A R A N C E S (Cont'd):
`
`2 3
`
` STERNE, KESSLER, GOLDSTEIN & FOX P.L.L.C
`4 BY: MICHAEL Q. LEE, ESQ.
`5 JOHN HARRIS CURRY, ESQ.
`6 CHRISTOPHER B. TOKARCZYK, ESQ.
`7 1100 New York Avenue, NW
`8 Washington, D.C. 20005
`9 202.772.8674
`10 mlee@skgf.com
`11 jcurry@skgf.com
`12 ctokarczyk@skgf.com
`13 Representing - ParkerVision, Inc.
`14
`15
`16 ALSO PRESENT:
`17 Thomas F. Presson, Esq. - ParkerVision
`18 Gregory L. Pollaro
`19
`20
`21
`22
`23
`24
`25
`
`Veritext Legal Solutions
`215-241-1000 ~ 610-434-8588 ~ 302-571-0510 ~ 202-803-8830
`
`Farmwald and RPX Exhibit 1066, pg. 2
`Farmwald and RPX v. ParkerVision
`IPR2014-00947
`
`

`

`Page 83
`
`Page 85
`
`1 multiple circuit units?
`2 A. Well, the Friis Equation explicitly
`3 explains how you calculate cascaded noise figure
`4 of multiple blocks in a circuit.
`5 Q. And at least as far as the analysis
`6 that you did and the circuits that are relevant
`7 here, we don't have to cascade it. We're just
`8 looking at the one circuit, right?
`9 A. I would say that, when dealing with
`10 sample and holds, you're not just dealing with
`11 a single circuit. And the reason is that the
`12 output impedance of a sample and hold is quite
`13 high and requires some form of buffering in
`14 order to make that signal available for further
`15 use. So, in general, there is always a second
`16 block in a sample-and-hold circuit, and,
`17 consequently, the Friis Equation will require
`18 to deal with that second element.
`19 Q. None of these patents claim any other
`20 circuit blocks that are cascaded along, do they?
`21 A. The ParkerVision patents are
`22 capable of driving a load impedance, subsequent
`23 functionality, if you wish to call it that,
`24 without buffering because of their lower
`25 impedance; however, subsequent circuitry could
`
`1 of that?
`2 A. Certainly, algorithmic errors can
`3 arise in a variety of ways, including poor
`4 implementations of the simulator, which, in
`5 this case of Cadence Spectre, it would not be
`6 expected, but also in the form of implementing
`7 incorrectly the function intended when
`8 describing it to the simulator. Whereas,
`9 numerical errors arise because of other
`10 problems, such as the inability of the circuit
`11 simulator to solve the matrix equations, which
`12 are typically sparse matrix equations, the
`13 inability of the circuit simulator to adjust the
`14 parameters to a small enough time steps to get
`15 numerical stability in the numbers, and, most
`16 commonly, the greatest problem with such
`17 simulators is extremely High Q resonant circuits
`18 where it's not able to accurately model Qs up in
`19 the thousands.
`20 Q. Okay. Now, I think I heard you say,
`21 SpectreRF, we assume that, in solving equations,
`22 since it's kind of an industry standard system,
`23 that it doesn't have any algorithmic errors,
`24 right?
`25 A. I would not expect it to have
`
`Page 84
`
`Page 86
`
`1 be applied.
`2 Q. So the answer to my question is, no,
`3 the claims are not addressed in the subsequent
`4 circuitry, correct?
`5 A. The claims are not addressed in
`6 subsequent circuitry.
`7 Q. Okay. Now, in the simulations that we
`8 talked about last week, you had added to Exhibit
`9 2026 the RLC circuit on the control circuit
`10 board.
`11 Do you recall discussing that?
`12 A. Yes, sir.
`13 Q. Okay. Now, do you think that the
`14 simulation would fail without that RLC?
`15 A. No, I do not. The RLC was simply to
`16 soften the edges of the control circuit.
`17 Q. Right. Because the simulation already
`18 has to deal with the sharp transition, because
`19 it's in there before you get to the RLC, right?
`20 A. The digital drive, yes.
`21 Q. When we're talking about simulations,
`22 do you understand the distinction between
`23 numerical errors and algorithmic errors?
`24 A. I do.
`25 Q. All right. What's your understanding
`
`1 algorithmic errors.
`2 Q. You described last week that you
`3 worked with Mike Rawlins of ParkerVision on the
`4 simulations, correct?
`5 A. Yes, sir.
`6 Q. And Mike Rawlins is a ParkerVision
`7 employee, correct?
`8 A. Yes.
`9 Q. What's his experience with the
`10 SpectreRF simulator?
`11 A. I did not delve into his experience
`12 level with the tool. I would assume that his
`13 experience level is quite high, since he has
`14 been designing integrated circuits using the
`15 tool and analyzing their performance.
`16 Q. Well, how do you know that?
`17 A. Because he's built a number of
`18 integrated circuits using that tool.
`19 Q. How do you know that?
`20 A. Pictures on the walls.
`21 Q. Actual circuits?
`22 A. Yes.
`23 Q. Does ParkerVision have any products?
`24 A. Oh, yes.
`25 Q. Have they sold any?
`
`Veritext Legal Solutions
`215-241-1000 ~ 610-434-8588 ~ 302-571-0510 ~ 202-803-8830
`
`3 (Pages 83 - 86)
`
`Farmwald and RPX Exhibit 1066, pg. 3
`Farmwald and RPX v. ParkerVision
`IPR2014-00947
`
`

`

`Page 87
`
`Page 89
`
`1 A. I don't know what the -- their sheet
`2 is on that.
`3 Q. But based on the pictures on the wall,
`4 you believe Mike Rawlins is the one who designed
`5 them using Spectre?
`6 A. Yes. I also saw components in the lab
`7 being measured.
`8 Q. Okay. So why do you think Mike
`9 Rawlins designed them using Spectre?
`10 A. He seemed to be extremely familiar
`11 with all of the circuits on the walls and what
`12 they did and how they worked.
`13 Q. Does Mike Rawlins have stock options
`14 in ParkerVision?
`15 A. I have no way of knowing, sir.
`16 Q. Do you know if Mike Rawlins has a
`17 bonus contingent on the outcome of these
`18 proceedings?
`19 A. I have no way of knowing, sir.
`20 Q. So whether Mike Rawlins has a dog in
`21 the fight, you just don't know?
`22 A. That's correct.
`23 Q. But you do know he's employed by
`24 ParkerVision and has been for some time?
`25 A. Yes.
`
`Page 88
`
`1 Q. Do you know how long?
`2 A. I do not know.
`3 Q. Do you know what a noise figure meter
`4 is?
`5 A. We used to have those such things 40
`6 years ago at Texas Instruments, but I haven't
`7 seen one lately.
`8 Q. It's a standard tool, though, right?
`9 A. Yes.
`10 Q. It measures the noise figure?
`11 A. Yep.
`12 Q. Still widely used, right?
`13 A. I think so.
`14 Q. Do you know what the input, typically,
`15 is into a noise figure meter?
`16 A. Typically, it's capable of
`17 synthesizing both a noise component and a clean
`18 signal component, such that the -- at the output
`19 of the circuit, the energy level of the clean
`20 signal component and the noise figure components
`21 can be separately measured.
`22 Q. Do you know whether a noise figure
`23 meter can make a measurement of noise figure
`24 with only a noise input?
`25 A. I would expect that that would be the
`
`1 case.
`2 Q. Now, there was some discussion last
`3 week about available powering. And I believe
`4 you said in order to calculate noise figure you
`5 have to take into account available power; is
`6 that correct?
`7 A. Yes, sir.
`8 Q. Okay. So available power is the
`9 maximum amount of power that can be delivered
`10 from some circuit, correct?
`11 A. During the proper operation of that
`12 circuit. So let me exemplify you: In a sample
`13 and hold where the hold functionality only works
`14 properly when there's a high -- a very high load
`15 impedance, such that the out-proposed does not
`16 decay, then you cannot apply a lower impedance
`17 to that circuit to measure its available output
`18 power.
`19 So, for example, sample and hold, if,
`20 as we discussed last time, the load impedance
`21 should be, let's say, 550 K or 700 K. Then, if
`22 you apply a lower load impedance in order to
`23 measure the available output power, you are
`24 asking the sample-and-hold circuit to perform
`25 incorrectly as relates to its functionality.
`
`Page 90
`1 It then ceases to be operating properly as a
`2 sample and hold.
`3 Abidi, himself, mentions the fact that
`4 he doesn't want voltage on a sample and hold to
`5 decay during that period of time that it's in
`6 hold mode.
`7 So I believe that answers your
`8 question.
`9 Q. So how did you take into account
`10 available power in your simulations?
`11 A. When measuring sample and holds,
`12 we measure that with the output impedance
`13 associated with the appropriate sample and
`14 hold load impedance. As we mentioned,
`15 Weisskopf and Avitabile both discuss very high
`16 load impedances. And, as we also mentioned,
`17 they didn't specify in either case what exact
`18 load impedances they expected to use, and we
`19 assumed load impedances that would deliver one
`20 percent accuracies by not drooping the energy
`21 more than a half of a percent. So we measured
`22 the available power with the available practical
`23 output impedance associated with sample and
`24 hold.
`25 Q. So you're measuring the actual power?
`
`4 (Pages 87 - 90)
`
`Veritext Legal Solutions
`215-241-1000 ~ 610-434-8588 ~ 302-571-0510 ~ 202-803-8830
`
`Farmwald and RPX Exhibit 1066, pg. 4
`Farmwald and RPX v. ParkerVision
`IPR2014-00947
`
`

`

`Page 91
`
`1 A. Excuse me?
`2 Q. You're measuring the actual power?
`3 A. Yes, sir.
`4 Q. All right.
`5 A. As appropriate for the circuit to
`6 function properly.
`7 However, when measuring the energy
`8 transfer circuits, we measure them with the
`9 power delivered to the load impedances as
`10 described, which, of course, is a lower load
`11 impedance, and, therefore, it delivers energy
`12 rather than simply voltage.
`13 Q. But under the true definition of
`14 available power, the available power definition
`15 doesn't really apply to discharging a capacitor,
`16 because, in theory, I could discharge a
`17 capacitor in an infinitely short amount of time,
`18 thereby delivering infinite power, regardless of
`19 the amount of energy stored on that capacitor,
`20 correct.
`21 A. Sir, it sounds like a hypothetical
`22 question. While you can short-circuit a
`23 capacitor and make sparks, the appropriate
`24 objective in analyzing receiver performance as
`25 regards noise figure is to use the energy as I
`
`Page 92
`
`1 have described it.
`2 Q. Okay. My question is, using the
`3 standard definition of available power, the
`4 maximum amount of power that can be delivered,
`5 that is not a well-defined quantity when you're
`6 talking about discharging a capacitor because it
`7 could theoretically be infinite, correct?
`8 A. As I stated before, the notion of
`9 discharging that capacitor, in any fashion other
`10 than as the circuit requires, is not addressing
`11 the Friis Equation properly with regards to
`12 available power.
`13 Q. Now, in your simulations, what you
`14 actually reported in your declaration, were two
`15 results for noise figure, correct?
`16 A. Yes, sir.
`17 Q. You did a lot of other simulations,
`18 but you chose not to record the results,
`19 correct?
`20 A. Yes, sir.
`21 Q. And sitting here today, you don't
`22 recall those results. Is that correct, too?
`23 A. That is correct, sir.
`24 Q. Okay. Now, you mentioned Dr. Abidi
`25 said he would like -- he thought -- the common
`
`Page 93
`1 wisdom was to try to hold the value on the
`2 capacitor constant or relatively constant,
`3 correct?
`4 A. During the period of time when the
`5 switch is open.
`6 Q. Right. Okay. So that notion, trying
`7 to maintain a relatively constant voltage during
`8 the hold period, that's old and well-known,
`9 right?
`10 A. That is traditional for a
`11 sample-and-hold circuit.
`12 Q. Okay.
`13 A. And that is why it's called sample and
`14 hold.
`15 Q. Okay.
`16 A. And the hold of the hold word implies
`17 holding the voltage -- voltage throughout the
`18 period of time when the switch is open.
`19 As I mentioned in my experience basis,
`20 when I was a co-op student at Texas Instruments
`21 in 1965 through '68, one of my first projects
`22 was to build sample and holds to measure the
`23 rise time of transistors on the transistor
`24 production line, which we called the central
`25 automated tester. And it required us to have
`
`Page 94
`1 two sampling circuits that would measure the
`2 voltage at the output of the collector of a
`3 transistor to measure the rise time and turn
`4 around and, on a separate device, two
`5 sample-and-hold circuits that would measure the
`6 fall time in the transistor where switching time
`7 was an important feature parameter.
`8 Q. So if I understand your report, your
`9 testimony so far, you think that the inventions
`10 in the patents we're dealing with are different
`11 from sample and hold because they want to
`12 discharge, and you think that gives a better
`13 result, right?
`14 A. Sir, it's more than just a better
`15 result. Sample-and-hold circuits have a place
`16 in circuits where you're making instrumentation,
`17 such as oscilloscopes and volt meters and so
`18 forth and so on, where you're going to make a
`19 measurement with an A to D converter of a
`20 voltage.
`21 The energy transfer circuits, on
`22 the other hand, are a optimized -- a means of
`23 optimizing designs for receiver circuits, where
`24 the intent is to measure energy per bit in a
`25 receiver, and energy per bit is the standardized
`
`5 (Pages 91 - 94)
`
`Veritext Legal Solutions
`215-241-1000 ~ 610-434-8588 ~ 302-571-0510 ~ 202-803-8830
`
`Farmwald and RPX Exhibit 1066, pg. 5
`Farmwald and RPX v. ParkerVision
`IPR2014-00947
`
`

`

`Page 95
`1 methodology of understanding the performance of
`2 the receiver, and it's measured in energy not in
`3 voltage.
`4 So the ParkerVision energy transfer
`5 patents all address themselves to doing as good
`6 a job as possible with energy transfer from the
`7 carrier to the circuits that then convert that
`8 into information bits.
`9 Q. Okay. Just -- there's the '551
`10 patent, Exhibit 1001.
`11 If I look at Figure 82-B, that's what
`12 I've been told is an energy transfer sampler by
`13 you and by others from ParkerVision. Where does
`14 it measure energy per bit?
`15 A. As with any patent, there are many
`16 pieces to the story. The part you're seeing in
`17 82-B is showing the sampling function. The
`18 sampling function, 82.06, is controlled by the
`19 energy transfer signal, 82.10, --
`20 Q All right. Could you please --
`21 A. -- which is described as a pulse of a
`22 non-negligible aperture.
`23 Q. Sir, I don't want to interrupt, but I
`24 want you to answer my question. I've had to
`25 come back here a second week. Answer my
`
`Page 97
`1 A. Actually, sir, I would actually go to
`2 his computer, and we would sit together at the
`3 computer and see the simulations, the conditions
`4 of the simulations, perhaps do additional
`5 analysis that would show performance versus
`6 various parametric choices, to see how the
`7 performance of the system met all the
`8 requirements of the program. The requirements
`9 in the program can be many, and so there's often
`10 a notion of finding a way to fit inside
`11 multidimensional polygon of requirements.
`12 Q. So a short answer would be, yes, you'd
`13 go do more work?
`14 A. I'd go sit with him, and we'd work
`15 through it, yes, sir.
`16 Q. Okay. All right. We tried this last
`17 week. Let's see if we can do it.
`18 MR. BAILEY: Can I get a sticker?
`19 THE REPORTER: Sure.
`20 (Deposition Exhibit No. 1054, a graph
`21 by Mr. Bailey, was marked.)
`22 BY MR. BAILEY:
`23 Q. I want to mark as 1054 -- I've plotted
`24 on this graph the two points that you have in
`25 your report for simulating using a one-puff
`
`Page 96
`
`Page 98
`
`1 question.
`2 Does this figure show where you
`3 measure energy per bit, that you discussed
`4 about?
`5 A. This figure, not in its own, shows
`6 that. And it's necessary to understand it in
`7 the context of figures such as 83-A and B, which
`8 are showing how the circuit delivers energy,
`9 which can then be interpreted by a modem into
`10 bits, and subsequent figures that walk through
`11 that part of the basis.
`12 Q. All right. Now, you discussed last
`13 week about, when you were at Motorola, you had
`14 design reviews, correct? That was one of the
`15 things you used to go to?
`16 A. Yes, sir.
`17 Q. If a young engineer came into a design
`18 review and said, I want to convince you to
`19 change a design on a product, we should do
`20 something different, and I have two simulations.
`21 I did some others, but I didn't record them, and
`22 I don't remember what the result was, but I do
`23 have two points. Let's change the product.
`24 Wouldn't you send him back to the drawing board
`25 to do some more work?
`
`1 capacitor. So the lower point is the energy
`2 transfer system that you said, which is
`3 identical to Weisskopf in all respects except
`4 for the load.
`5 So my question is, based on your
`6 experience and the simulations you've done, can
`7 you tell me what the noise figure graph would
`8 look like between those two points?
`9 A. I believe you would find -- how -- I'm
`10 not quite sure how to go about this in the legal
`11 sense, but you would find that there's an
`12 optimal region in and around the region from,
`13 oh, say, 900 ohms to 2000 ohms, and you would
`14 see the noise figure performance rise
`15 dramatically above a hundred thousand ohms, up
`16 to the megaohm and gigaohm-type numbers that you
`17 can easily imagine --
`18 Q. You can draw it on there, if you want.
`19 A. That's okay. I prefer not to.
`20 Q. Why?
`21 A. Because without doing accurate
`22 analysis, I'm just reluctant to put real numbers
`23 on here.
`24 Q. Okay. But all nature has trends,
`25 right?
`
`6 (Pages 95 - 98)
`
`Veritext Legal Solutions
`215-241-1000 ~ 610-434-8588 ~ 302-571-0510 ~ 202-803-8830
`
`Farmwald and RPX Exhibit 1066, pg. 6
`Farmwald and RPX v. ParkerVision
`IPR2014-00947
`
`

`

`Page 99
`
`Page 101
`
`1 A. Yes.
`2 Q. Okay. So there's going to be some
`3 trend, but beyond what you've just said to me,
`4 you think it will be -- optimal is somewhere
`5 around 900 to 2000 ohms, and it will get worse
`6 as you go up. That's as much as you're
`7 comfortable with based on the analysis that
`8 you've done?
`9 A. And let me remind you that the notion
`10 is that, when you have high impedance, you're
`11 delivering very small energy to subsequent
`12 circuits, and when you have low impedance,
`13 you're delivering substantially greater energy
`14 to subsequent circuits, which, generally, are
`15 demodulators.
`16 In a receiver design, the important
`17 measurement is energy per bit delivered compared
`18 with the noise delivered during that same bit
`19 time. And so the significance of improved
`20 energy per bit versus noise per benefit is a
`21 very significant non-negligible factor.
`22 (Deposition Exhibit No. 1055, a graph
`23 by Mr. Bailey, was marked.)
`24 BY MR. BAILEY:
`25 Q. Let me give you Exhibit 1055. I'm not
`
`1 ParkerVision patents, I kept my energies
`2 focused on those questions.
`3 (Deposition Exhibit No. 1056, a graph
`4 by Mr. Bailey, was marked.)
`5 BY MR. BAILEY:
`6 Q. All right. Let me give you Exhibit
`7 1056. I drew another curve on there.
`8 So if you swept across load and you
`9 got a smooth curve with a bunch of spikes on it,
`10 would you rely on the peaks of those spikes?
`11 A. No, sir, I would not.
`12 Q. Why not?
`13 A. It would not be apparent to me what
`14 nonlinear phenomenon was causing that as regards
`15 the changes of impedance, and so I would not
`16 rely on that.
`17 Q. Okay.
`18 A. I'm, by the way, a strong believer in
`19 nonlinear phenomena and understanding nonlinear
`20 phenomena.
`21 Q. So if you have a nonlinear phenomena,
`22 you should have a reason to explain it, correct?
`23 A. Yes.
`24 Q. Now, going back in your stack to
`25 Dr. Abidi's report, Exhibit 1004, --
`
`Page 100
`
`Page 102
`
`1 shy. I drew a line on there.
`2 Is that -- I'm not trying to lock you
`3 down to numerical precision, but is that the
`4 type of shape you would expect?
`5 A. I really have no way of understanding
`6 the shape of the curve in the region below 900
`7 ohms, having done no experience [sic] with that
`8 whatsoever.
`9 The shape of the curve going upwards
`10 is trending the right direction. That's about
`11 as much as I can say.
`12 Q. But going back to your simulations in
`13 1026, the only thing that varied between what
`14 you were calling the energy transfer sampler and
`15 what you're calling Weisskopf is the impedance
`16 of the load, correct?
`17 A. In those simulations, right.
`18 Q. Wouldn't it make sense to do
`19 simulations sweeping across a variety of loads?
`20 A. Sir, it would be scientifically
`21 educational for purposes, particularly, of
`22 a scientific paper. But as I was asked to
`23 opine on the difference between sample-and-hold
`24 circuits, such as Weisskopf and Avitabile,
`25 and energy transfer circuits, such as the
`
`1 A. So is this 1004?
`2 Q. It is.
`3 Could you give me 1004, just to make
`4 life easier?
`5 A. Just like to be sure we doing
`6 everything according to Hoyle.
`7 Q. I'm not even sure who Hoyle is.
`8 A. Just some guy who wrote a book about
`9 cards.
`10 Q. Oh, the guy that makes the poker
`11 cards?
`12 A. Yeah.
`13 Q. All right. So we talked a little bit
`14 about Dr. Abidi's work last week, and you had
`15 one issue where you said he didn't take into
`16 account temperature. I said, doesn't
`17 temperature cancel out. Do you have an opinion,
`18 like, if we start on Page 11, you know, of
`19 Dr. Abidi's report, he has equation 5.4. That's
`20 a standard equation for noise factor, right?
`21 A. Yes, sir.
`22 Q. And then he moves through and he adds
`23 in 5.5. That's a standard equation for noise,
`24 right?
`25 A. Noise voltage.
`
`Veritext Legal Solutions
`215-241-1000 ~ 610-434-8588 ~ 302-571-0510 ~ 202-803-8830
`
`7 (Pages 99 - 102)
`
`Farmwald and RPX Exhibit 1066, pg. 7
`Farmwald and RPX v. ParkerVision
`IPR2014-00947
`
`

`

`Page 103
`1 Q. And then he moves through and he has
`2 Equation 5.6 and ultimately he comes to 5.7 for
`3 noise factor. So do you have an opinion that he
`4 did this wrong, or just no opinion?
`5 A. I found it impossible to follow the
`6 derivation between the early steps, 5.5, and the
`7 applicable circuitry of Weisskopf and Avitabile.
`8 The steps in between are incomplete, and so I'm
`9 incapable of validating derivation of those
`10 remaining equations.
`11 Q. Now, simulations are often used as a
`12 tool to validate analysis, correct?
`13 A. Yes, sir.
`14 Q. All right. So that's not the way you
`15 used simulations in your declaration, correct?
`16 A. That's correct. I used simulations to
`17 calculate the actual numbers that the best tools
`18 in the industry can provide.
`19 Q. So you did not do an analysis prior to
`20 the simulation. You're relying exclusively on
`21 the simulation, correct?
`22 A. Yes, sir. Which, by the way, allows
`23 us to address any nonlinear artifacts.
`24 Q. And Dr. Abidi actually swept his
`25 simulation over multiple parameters, showing
`
`Page 105
`1 impedances, we expect the noise figure to be
`2 higher, as shown in the analysis, and yet it's
`3 not showing that way here. So we don't quite
`4 understand how this figure relates to the proper
`5 implementation of noise figure calculations.
`6 Q. Well, didn't everybody in the world
`7 before ParkerVision think the higher the load
`8 resistance, the better your signal would be?
`9 A. In the days before the ParkerVision
`10 patents, everybody believed in sample-and-hold
`11 circuits, and that was commonly and routinely
`12 practiced throughout the semiconductor industry
`13 for receivers. And in sample-and-hold circuits,
`14 everybody believed that it was important to
`15 maintain voltage for the A to D converters
`16 that were subsequent to the sampling process.
`17 Maintaining voltage is important to an A to D
`18 converter because you don't want the voltage to
`19 change while the A to D converter is performing
`20 its A to D conversion calculations. So, yes,
`21 everybody wanted high impedance in their
`22 sample-and-hold circuits in the days prior to,
`23 approximately, 2001. That was common in the
`24 industry.
`25 The ParkerVision patents, where
`
`Page 104
`1 that it matched with his analysis, correct?
`2 A. It's difficult for me to tell you that
`3 swept data of his Figure 5.2 and/or his Figure
`4 5.4 maps to his Equations 5.6 and 5.7. Again, I
`5 cannot affirm that.
`6 Q. It's what he shows on the graphs,
`7 though.
`8 A. He does show graphs.
`9 Q. Okay. If you would go to your --
`10 A. I thought it was somewhat interesting
`11 that he did not graph noise figure as a function
`12 of load impedance, which, of course, is the
`13 subject of the substantial difference between
`14 sample and hold and energy transfer design.
`15 Q. Do you want to turn to Page 39?
`16 A. Okay.
`17 Q. Isn't that exactly what he did, graph
`18 noise figures swept across load resistance?
`19 A. But then I can't tell the exact
`20 circuit that he's used. Sorry.
`21 Q. So you -- you -- let me understand.
`22 You don't know what he's using here?
`23 A. That's correct. Because, for example,
`24 at high load impedances, which all he's listed
`25 here is leakage resistance, at high load
`
`Page 106
`1 energy transfer explained how to deliver energy,
`2 rather than voltage directly to the demodulator.
`3 It changed that process in the industry.
`4 Q. But what Dr. Abidi's analysis here
`5 in Figure 10.1 is showing is that he thinks
`6 everybody was always right?
`7 A. Well, I think, but I can't tell for
`8 sure, because he's not showing the -- the
`9 circuitry in the way he's performing this
`10 analysis. But I believe what he's showing here
`11 is droop voltage versus load resistance, because
`12 he discusses it in that fashion.
`13 Again, I can't prove it, because I
`14 can't follow the arithmetic and how he relates
`15 it to the circuits.
`16 Q. Okay. Let's go back to your
`17 declaration. And if you turn to page -- it's
`18 Exhibit 2024. Turn to Page 118, and you have a
`19 Figure 30 there?
`20 A. Yes, sir.
`21 Q. How is that generated?
`22 A. That was generated using the partial
`23 differential equations of circuit reflecting the
`24 sample and hold functionality, and a circuit
`25 representing an energy transfer functionality,
`
`Veritext Legal Solutions
`215-241-1000 ~ 610-434-8588 ~ 302-571-0510 ~ 202-803-8830
`
`8 (Pages 103 - 106)
`
`Farmwald and RPX Exhibit 1066, pg. 8
`Farmwald and RPX v. ParkerVision
`IPR2014-00947
`
`

`

`Page 107
`1 and a control signal. As you can see here, the
`2 control signal is shown in purple, with a minus
`3 two volt offset so you can see it without it
`4 interrupting the desired signal voltage, shown
`5 in blue, and the output of a sample and hold
`6 functionality, shown in red, and the output
`7 showing the energy transfer integration, shown
`8 in green.
`9 Q. Okay. So let me just make sure I got
`10 that all right. So in Figure 30, purple control
`11 signal, that's what turns the switch on and off,
`12 right?
`13 A. Yes, it does.
`14 Q. Blue is the input signal, correct?
`15 A. Yes, sir.
`16 Q. Red is the down-converted signal
`17 output by sample and hold?
`18 A. Yes, sir.
`19 Q. And green is the down-converted
`20 signal output by what you're calling an energy
`21 transfer signal?
`22 A. Yes, sir.
`23 Q. Okay. And if I look on the next page,
`24 Footnote 16, you said you did the analysis for
`25 Figure 30 using a mathematical model of a simple
`
`Page 109
`1 circuits are the same, other than changing some
`2 parameters, correct?
`3 A. Sir, there are some differences in
`4 parameters, but there's also major difference
`5 in functionality, as well. So let's discuss the
`6 important differences between these circuits.
`7 An energy transfer sampler is
`8 intended to integrate energy over multiple
`9 apertures, it is intended to use a substantially
`10 larger non-negligible aperture timing, and is
`11 intended to provide greater e