EUGENE A. WRIGHT, Circuit Judge.
Beckman Instruments appeals a decision of the district court which held that its analog computer check circuit infringed claims 7, 8, and 13 of U. S. Patent No. 2,967,997. We affirm.
I. THE PATENT IN SUIT.
U. S. Patent No. 2,967,997 (hereinafter the "McCoy patent") was issued to Rawley D. McCoy on January 10, 1961 as the result of an application filed April 20, 1955. All right, title and interest in the McCoy patent has been assigned to appellee Dynamics Corporation of America, of which appellee Reeves Instrument is a division.
The McCoy patent is entitled "Method and Apparatus for Checking Electronic Analog Computers" and deals generally with the problem of checking the operation of the numerous elements of an analog computer prior to utilization of the computer to solve complex industrial problems of a generally mathematical nature.
Electronic analog computers were initially developed following World War II and played an important role in the design of complex aircraft, weapons systems and space vehicles. Such computers are designed to solve a variety of problems by performing the mathematical operations required to analyze complex physical relationships.
Since a general purpose analog computer is designed to solve a variety of problems, it consists of a number of building blocks which can be interconnected in accordance with the mathematical definition of the problem to be solved. These building blocks include multipliers, summers, function generators and integrators. Perhaps the single most prevalent element in such computers is the integrator which can also be connected to perform other desired mathematical functions.
From this brief review, it is apparent that the interconnection of the elements as well as the potentiometer settings are vitally important to proper operation. Because such computers are designed to solve complex problems, it usually is not possible to determine whether the computer is properly programmed by examining the answers obtained. Rather, it is necessary to make some sort of initial verification of the programming before one may have confidence in the correctness of the solution. Certain visual checks of interconnection, potentiometer settings, etc., can be used to perform limited checking of the programming but in the final analysis it is necessary to perform some kind of operational check since connections that appear sound can be faulty, potentiometer settings may not reflect the actual value of the resistance, etc.
There were numerous attempts prior to McCoy to derive a satisfactory method of checking the computer prior to attempted solution of a problem. During trial, defendant Beckman introduced evidence of some 15 different prior art approaches to computer checking of which three are relied upon on appeal and discussed in detail in Section II, infra. Suffice it to say that the problem of checking analog computers was born with the development of such computers and has since occupied much time and effort in the computer industry.
The McCoy patent discloses and claims two different types of checks for analog computers: a static check and a dynamic check. The claims involved in this appeal deal only with the static check portion of the McCoy patent. Figure 1 hereof is a reproduction of Fig. 2 of the McCoy patent and illustrates an exemplary embodiment, including the static check portion of the McCoy disclosure.
In actual operation, the signal to be integrated by amplifier 39 and condenser 40 is generally the sum of other signals as determined by the program. These other signals must first be summed prior to integration and this is done by the summing resistors 25, 26, 27, 28. The summed signal passes through the contacts 31, 33 of the "Y" relay and forms the input to the electronic integrator. The integrated output signal, in turn, passes through contacts 46, 47 of the "P.C." relay to output terminal 48 from where it is relayed to other sections of the computer in accordance with the problem program.
During operation of the computer for problem solving, the X, Y, and P.C. relays are de-energized and the circuit is operated as shown. To perform the static check, the McCoy patent utilizes the X and P.C. relays to modify operation and provide for checking of the program prior to actual problem solving. The Y relay is used only for the so-called "dynamic check" which is not involved in the claims whose validity and infringement are now in question.
To carry out the desired static check, the X and P.C. relays are energized at the appropriate time by a timer 63. With the X and P.C. relays energized, the amplifier 39 no longer acts as an integrator since a shunt resistor 41 is placed in parallel with the integrating condenser 40 via contacts 55, 57 of the X relay and contacts 43, 45 of the P.C. relay. When the shunt resistor 41 is placed in parallel with the integrating condenser 40, amplifier 39 acts essentially as a simple electronic multiplier.
In addition, energization of the P.C. relay disconnects the output of amplifier 39 from output terminal 48 and substitutes a fixed voltage from voltage source 53 in its place. This is accomplished by connecting movable contact 47 to contact 50 and disconnecting movable contact 47 from contact 46. Since the X and P.C. relays are simultaneously energized for all integrators in the computer, the outputs of all of the integrators (as they appear on output terminal 48) become fixed voltages established by the pre-determined settings of the potentiometers 53 in each integrator.
In this way, the interconnections between the integrating amplifiers are broken to "open the loop." Fixed voltages are substituted for the outputs of the integrating amplifiers and these fixed voltages circulate through the interconnections to form the inputs to other integrators in the circuit. Since these voltages are both fixed and known, it is possible to predict what the input to each integrator must be if the problem is properly programmed and the potentiometer values properly established.
In the static check, the input to each integrator is checked. In the exemplary embodiment illustrated in the McCoy patent, the input to each integrator is checked by a digital converter and recorder 64 which is connected to the output of amplifier 39. Amplifier 39 no longer acts as an integrator because of the shunt resistor 41, but rather as a simple multiplier circuit. Its output voltage is thus representative of the input signal to the integrator. This voltage may be digitized and either manually or automatically checked with previously computed values to ascertain whether computer interconnections, potentiometer settings, etc., are proper.
The claims in issue on appeal deal with the static check portion of the McCoy patent. Claim 7 is the broadest of the three and provides:
The essence of McCoy's contribution is thus defined in claim 7 as including switching means to open the loop and render the integrator ineffective, followed by circulation of a fixed voltage through the open loop system and measurement of the voltage which appears at the input to the integrator. Claims 8 and 13 are similar but more detailed in their recitation of the elements of the claimed combination.
II. THE PRIOR ART.
a. The Initial Condition (I.C.) Static Check.
The first item of prior art relied upon by Beckman is the so-called Initial Condition (I.C.) static check. This method of checking analog computers originated admittedly more than one year prior to the filing date of the McCoy patent, to constitute a valid reference under 35 U.S.C. §§ 102, 103.
The initial condition circuitry of an analog computer is used primarily for the purpose of establishing certain pre-determined initial conditions in conformity with the problem to be solved by the computer. That is, certain variables in the problem will have non-zero initial conditions so that the voltage appearing on some (but usually not all) integrators in the system must begin at some value other than zero. An exemplary circuit for establishing the required initial conditions is illustrated in Figure 2, a reproduction of Figure 7.2(a) of Korn & Korn, Electronic Analog Computers (1952).
Under ordinary circumstances, the input resistors are connected directly to the input of the D-C amplifier because the relay is not energized. In order to establish the desired initial condition, the relay is energized. This places the leakage resistor r
The initial condition voltage circuit is also present in the disclosure of the McCoy patent and is illustrated by the Initial Condition Voltage source 59, resistor 41, input resistor 42, and relays X and Y which jointly accomplish the function of the relay of Figure 2.
It was established at trial that certain aspects of an analog computer could be checked by applying the initial condition voltage as set forth above and then making certain measurements within the loop (which is opened at the input to each integrator) prior to de-energizing the relay and proceeding with the desired computation. It was, however, equally established that this approach to checking had two substantial disadvantages.
The first disadvantage to this approach stems from the fact that not all integrators in a given problem will require an initial condition voltage so that, in order to check the circuitry associated with these integrators, it would be necessary to establish some initial condition voltage for checking purposes. This voltage would, however, have to be removed prior to running the problem. The chance that the computer operator might forget to remove these checking voltages adds an extra source of possible human error into the system.
The second disadvantage to this approach stems from the fact that it can check only limited portions of the loop. Particularly troublesome is the fact that it was not possible to check the crucial connections on the patchboard which connect the input resistors to the integrators nor was it possible to check the integrator summing junctions since the loop is opened at the input to each integrator.
b. The Martin Check.
The second item of prior art relied upon by Beckman is the so-called "Martin check" developed by Betty Lou Wilder of the Martin Company in 1953 and which constitutes an item of prior art in existence more than one year prior to the filing date of the McCoy patent.
Without detailed illustration, it suffices to say that an integrator circuit in an analog computer ordinarily serves two separate functions. It first sums the signals applied to its input resistors and then integrates that sum. The Martin technique separated these two functions by providing an additional amplifier which acted as a summer. The output of this amplifier was the sum of the signals applied to the input resistors and was fed through a single input resistor to a second amplifier which integrated that sum.
The Martin check then followed the basic approach of the I.C. check but had the additional capability of allowing a check of the input resistors since it was possible to check the output of the summing amplifier when the initial condition voltage was applied to the preceding integrator in the circuit.
Thus, while the Martin check did somewhat expand the capabilities of the I.C. check it retained all of the first disadvantage (disturbance of the problem setup) referred to above in that it was still necessary to apply some sort of initial condition voltage to those integrators where the problem called for none. Further, the Martin check was still incomplete in that it failed to check the crucial interconnection between the summing amplifier and the integrator and its input resistor.
The district court found that the Martin check was "basically the I.C. check and has the same shortcomings." We agree completely with that analysis of this item of prior art and conclude that the differences between it and the McCoy patent are essentially the same as those differences discussed above with respect to the prior art I.C. check.
c. The M.I.T. Check and the Jatras Thesis.
In 1945, Massachusetts Institute of Technology began construction of a large analog computer facility, ultimately completed in 1948. That facility functioned essentially as do more contemporary versions of the analog computer but utilized electromechanical integrators rather than the electrical integrator circuits now in almost universal use. This difference does not, however, mean that the efforts and/or results at M.I.T. are not pertinent as to the patentability of McCoy's invention since the McCoy patent is not limited to electrical integrators and the problems which require checking are essentially the same whether the integrators are electrical or electromechanical.
Figure 3 hereof illustrates one of the integrators of the M.I.T. facility with its associated circuitry.
The electromechanical integrator of the M.I.T. facility includes the amplifier A, the tach, and the servomotor clutch and pot. The input signals to be integrated are first fed to a summing amplifier AR whose output is connected to a first switch (shown enclosed in the box with the letter H) which operates to open the loop. The output of this switch is fed to a second switch SW which is, in turn, connected to a first attenuator ATT-2 and a second attenuator ATT. The output of this second attenuator ATT forms the input to the electromechanical integrator.
The static check used at M.I.T. was essentially the same as that used in the I.C. and Martin checks referred to above. A fixed voltage was applied from the block labelled "I.C. Sum" to the input of the amplifier A which resulted in a rotation of the servomotor to establish some initial condition voltage. This voltage then circulated through other elements of the system to the input sum amplifier AR of the next integrator circuit where it was measured, among other places, at point W of switch H.
There was some controversy in the testimony on whether the checking approach used at M.I.T. required disturbance of the computer in order to supply check voltages for those integrators whose initial condition did not require such voltage. We need not, however, attempt to resolve this controversy since it is admitted by both parties that a Master's degree thesis written by Stephen Jatras, an M.I.T. graduate student, proposed a system where checking could be accomplished automatically without disturbing the problem set-up. This thesis was published in 1952 and is thus a part of the prior art as regards the McCoy patent.
As noted above, the M.I.T. system, even as automated by the Jatras thesis, checked only as far as the output of the input sum amplifier AR. This meant that, for each of the many integrators involved in a particular problem set-up, the switch SW, the attenuators ATT-2 and ATT, and their interconnection as well as the connections to the electromechanical integrator were not checked. The result is that the M.I.T. and Jatras checks failed to check the voltage appearing at the input to the integrating circuit, a specific requirement of each of the claims in
The district court found that "[i]n principle and in result the MIT procedures were the same as the I.C. check." We agree.
III. THE QUESTION OF OBVIOUSNESS.
On appeal, Beckman's primary contention is that the McCoy patent claims involved are invalid for obviousness as defined in 35 U.S.C. § 103. Other defenses, including late claiming, were urged before the district court but are not relied upon in this appeal and thus will not be discussed.
a. The Applicable Standard.
To be patentable, an alleged invention must be shown to be new and useful per 35 U.S.C. § 101 and nonobvious under 35 U.S.C. § 103. The requirement of novelty is more specifically defined in 35 U.S.C. § 102 and Beckman suggests in passing that we might find the McCoy patent invalid as fully anticipated by one or more of the prior art check systems discussed in full in Section II, supra. We think it clear that the McCoy patent is not anticipated by the prior art and pass on to the more substantial question of obviousness.
The statutory requirement of nonobviousness is found in 35 U.S.C. § 103 which prohibits the patenting of an invention, even though not identically shown in the prior art, if "the differences between the subject matter sought to be patented and the prior art are such that the subject matter as a whole would have been obvious at the time the invention was made to a person having ordinary skill in the art to which said subject matter pertains."
Beckman's initial contention here is that the McCoy patent is somehow invalid on its face, without regard to the prior art, in view of Great Atlantic & Pacific Tea Co. v. Supermarket Equip. Corp., 340 U.S. 147, 71 S.Ct. 127, 95 L. Ed. 162 (1950). Specifically, we are warned to scrutinize carefully those patents which include combinations of elements which are individually old in light of the alleged improbability of finding invention in such a combination. The argument is that the McCoy patent fails to satisfy the constitutional standard of patentability enunciated in A & P because the claimed invention consists of old elements (amplifiers, potentiometers, switches, etc.) which operate within the claimed combination in the same fashion as they have always operated in the prior art.
This argument fails on two grounds. First, it misconstrues the import of the A & P decision. Second, it suggests an analytical approach to patentability which is directly contrary to the statutory language of 35 U.S.C. § 103 which provides that the inquiry into patentability must be drawn toward the "subject matter as a whole" and not to the elements of a claimed combination and their individual novelty.
The claims in issue here are, in statutory language, "machine" claims.
The Court's decision in A & P recognizes that non-obviousness is less likely to occur in combination patents, and therefore applies a somewhat more stringent rule of non-obviousness to such patents. It is not inconsistent with, but merely in application of, the general rule stated in § 103. The Court's more recent decision in Anderson's-Black Rock, Inc. v. Pavement Salvage Co., 396 U.S. 57, 90 S.Ct. 305, 24 L.Ed.2d 258 (1969) supports this interpretation of the A & P decision.
Black Rock involved a similarly simple combination of old elements. The Court found that there was no invention when viewed by the "obvious-nonobvious standard." Id. at 63, 90 S.Ct. 305. Moreover, the Court in Black Rock reaffirmed the position it had previously taken in Graham v. John Deere Co., 383 U.S. 1, 86 S.Ct. 684, 15 L. Ed.2d 545 (1966) to make clear that the crucial question in determining obviousness is whether the differences between the subject matter sought to be patented, as a whole, and the prior art are obvious to a man of ordinary skill in the art.
From an analytical view, the most definitive statement of the requirement of nonobviousness and the approach to be taken by federal courts in determining this question is found in Graham v. John Deere Co., 383 U.S. 1, 86 S.Ct. 684 (1966).
This court has consistently followed the analysis as thus prescribed in Graham, noting that "the entire tenor of the Graham decision is that there should be strict observance of all three explicit conditions precedent to the issuance of a patent, namely novelty, utility and non-obviousness, with special emphasis being placed on the latter." Jeddeloh Brothers Sweed Mills, Inc. v. Coe Manufacturing Co., 375 F.2d 85, 87 (9th Cir. 1967). See also Ashcroft v. Paper Mate Mfg. Co., 434 F.2d 910 (9th Cir. 1970); Schwinn Bicycle Co. v. Goodyear Tire & Rubber Co., 444 F.2d 295 (9th Cir. 1970).
The claims involved in this case appear in Section I, supra, and the prior art relied upon by Beckman is set forth in Section II. As for the differences, we have already noted that the prior art methods of accomplishing the claimed static check failed to check at the input to the integrator. The claims in issue here all require checking at the integrator input.
Whether this difference rises to the level of patentability depends upon the level of ordinary skill in the pertinent art. It is difficult to set forth any meaningful quantitative evaluation of the level of skill in a given art. Rather, such determination can be made only by an analysis of the problem allegedly solved by the invention and the efforts of others to arrive at a satisfactory solution. In this respect, the Supreme Court has noted that "[s]uch secondary considerations as commercial success, long felt but unsolved needs, failure of others, etc., might be utilized to give light to the circumstances surrounding the origin of the
The magnitude of the checking problem in general purpose analog computers is well demonstrated by the number of prior art attempts presented to the district court. Our review of the record reveals some fifteen approaches to the problem solved by the McCoy patent. When the evidence shows that several others in the art have attempted to solve the same problem and have not arrived at the solution claimed by the patent in suit, the statutory presumption of validity
Evidence of the magnitude of this problem and the efforts directed at solving it is particularly strong as regards the research conducted at one of the nation's leading technical institutions, Massachusetts Institute of Technology. As noted above, M.I.T. developed and constructed one of the first general purpose analog computers.
There was extensive testimony during the trial in this case of the existence of this problem and the efforts of personnel at M.I.T. to derive a satisfactory method of checking its computer which did not disturb the problem set-up and constituted a thorough check of interconnections, potentiometer setting, etc.
We have discussed the work of an M.I. T. graduate student, Stephen Jatras.
His efforts, along with those of Tremblath, Mathews and Seifert, clearly establish the existence of a problem which had plagued the computer industry since the development of complex analog computers. Substantial efforts by others in the art which fail to accomplish the result achieved by the patented invention are persuasive indications of nonobviousness.
The "circumstances surrounding the origin of the subject matter" of the McCoy patent are perhaps best demonstrated by the sequence of events surrounding the construction of a large analog computer facility at Wright Air Development Center (WADC). In 1954, WADC was contemplating the construction of an analog computer facility of a size and complexity far in excess of anything previously built.
L. M. Warshawsky was the chief of the computer operation at WADC and a man with considerable experience and expertise in the field of analog computers. In July, 1954, Warshawsky wrote to Reeves, Beckman, and other large analog computer manufacturers and invited their proposals for specifications for the new WADC machine. The request called for a better approach to computer checking to minimize human and machine errors. The impetus to produce something new for the project was substantial since the opportunity to construct the world's largest and most advanced analog computer was at stake.
In response to the request, Reeves, Beckman, and three others proffered their ideas. While there were other matters involved in the proposal, the portions dealing with computer checking as submitted by Beckman and the other companies were essentially limited to prior art approaches. The Reeves proposal, on the other hand, included the problem
Warshawsky testified at the trial to his reaction to the problem check proposal of McCoy. In addition to recognition of the potential and simplicity of the McCoy approach, Warshawsky said that he and his colleagues had "wondered why didn't we think of this ourselves."
In short, we believe that the extensive efforts at M.I.T., the additional efforts in response to the WADC request and the acceptance of McCoy's proposal demonstrate that the admittedly simple approach to computer checking involved in this appeal was such that the differences between that approach and the prior art were not obvious to one of ordinary skill in the art, and did produce the unusual or surprising result referred to in A & P, supra.
Beckman's final contention is that the check circuitry of the Beckman 2100 and 2200 computers does not infringe claims 7, 8 and 13 of the McCoy patent. Figure 4 hereof is a simplified schematic diagram of the accused Beckman circuit. For convenience, the elements of the Beckman circuit have been labeled with the same numbers used for corresponding elements in the McCoy patent.
As illustrated in Figure 4, the Beckman check circuit opens the loop at the input to the integrating amplifier 39 by opening the contact labeled . The integrating amplifier 39 is disabled and converted to a simple multiplier by closing the contact labeled which places the shunt resistor 41 in parallel with the integrating condensor 40.
A fixed voltage is introduced into the opened loop by closing contacts 47,50 which connects a fixed voltage source, represented by potentiometer 53a, to the input of the multiplier circuit consisting of input resistor 53b, amplifier 39 and shunt resistor 41. This fixed voltage then circulates through the "Other Computing Elements" of Figure 4 and appears
From this brief explanation, it can be seen that the Beckman circuit varies from the exemplary circuit of the McCoy patent in two ways. First, the loop is broken at the input to the integrating amplifier 39 by contact as opposed to opening the loop at the output of the integrating amplifier as is done by contacts 46,47 of the exemplary circuit of the McCoy patent shown in Figure 1 hereof. Second, the McCoy patent uses the converted integrating amplifier 39 as the measuring amplifier whereas the Beckman circuit provides an additional measuring amplifier 39a.
These differences do not, however, support Beckman's claim of non-infringement. It is, of course, well settled that the specification and drawings of a patent are exemplary only and the true measure of the invention is defined by the claims.
With regard to the fact that the Beckman circuit opens the loop at the input to each integrator rather than at the output, it suffices to note that only claims 7 and 8 specifically require opening the loop and neither is limited to opening the loop at the output of the integrating amplifier. Rather, the claim language speaks only of "opening the loop."
As regards the use of a separate measuring amplifier in the Beckman circuit, the three claims in issue do not require that the disabled integrating amplifier be used as a part of the measuring system.
In sum, it is apparent that the accused Beckman circuit literally infringes claims 7, 8 and 13 of the McCoy patent. The minor differences between the Beckman circuit and the exemplary embodiment of the McCoy patent do not avoid literal infringement of the claims. "[T]o permit imitation of a patented invention which does not copy every literal detail would be to convert the protection of the patent grant into a hollow and useless thing." Graver Tank & Mfg. Co. v. Linde Air Products Co., 339 U.S. 605, 607, 70 S.Ct. 854, 856, 94 L.Ed. 1097 (1950).
8. In an analog computer including an integrating circuit having an amplifier with input and output terminals and a common terminal, said integrating circuit including condenser means coupled between the input and output terminals of said amplifier and further including an input resistor coupled in series with the input of said amplifier, said integrating circuit being coupled to the remaining elements of said analog computer for forming a closed-loop feed-back system, the combination for testing the analog computer system comprising shunt resistor means, switching means coupled to said integrating circuit and said shunt resistor means, said switching means opening the loop of the feedback system and coupling said shunt resistor means across said condenser means, means supplying a fixed voltage to the opened-loop system, said fixed voltage being translated through said opened-loop system to the input of said integrating circuit, and means coupled to the opened-loop system for measuring the voltage appearing between the input terminal of said input resistor and said common terminal.
13. Apparatus for testing an analog computer including at least one integrating circuit, said integrating circuit having an input coupled to first terminal means of the remaining elements of the computer and an output coupled to second terminal means of the remaining elements of the computer to form a closed-loop system, the input of said integrating circuit including a series input resistor, comprising in combination, means including switching means coupled to said integrating circuit, a source of known voltage coupled to said switching means, said switching means rendering said integrating circuit ineffective and supplying a known voltage to said second terminal means of the computer said known voltage being translated through the remaining elements of the computer to the series input resistor of said integrating circuit, and measuring means including amplifier means coupled to said series input resistor for measuring the response of the remaining elements of the computer to said known voltage.