RCA Corp. v. Applied Digital Data System, Inc.

OPINION STAPLETON, District Judge: In these consolidated actions, RCA charges Hazeltine, Lear Siegler and ADDS1 (collectively “HLA”) with willful and deliberate infringement of the Cole, et al U.S. Patent No. 3,345,458 (“the Cole patent”). There is no dispute that RCA owns the Cole patent, including the right to recover for past infringement, and there is no dispute as to subject matter jurisdiction, personal jurisdiction or venue. This opinion constitutes the Court’s findings of fact and conclusions of law following trial of the issues of validity, unenforceability and infringement. I. BACKGROUND FACTS. A. The Cole Patent Specifications. The Cole patent issued October 3, 1967, on an application filed October 16, 1963. Its specifications describe a system for decoding digital computer symbol codes representing a message and directly converting them into digital video control signals. These video signals are then used to turn an electron beam on and off to display the message as the beam scans a television raster scan pattern on the cathode ray tube (“CRT”) of a standard television set. In order to understand the Cole patent, it is necessary preliminarily to have an understanding of the formation of a picture on the CRT of a television set. A picture is formed by an electron beam which illuminates various points on the phosphor coating of the screen as it scans the area in which the image is to be displayed. Normally, the beam scans across one horizontal line at a time, starting at the top of the screen and moving sequentially down the screen to the bottom. This pattern of scan in which the beam proceeds across the entire width of the visible CRT screen before scanning a second horizontal line is referred to as a television raster scan pattern. By using a digital video control signal to appropriately control the points at which the [939] beam illuminates the screen during its scan, the beam can be used to form a recognizable message or image. Because of its speed, the beam’s movement is not detectable to the eye. The Cole specifications suggest that each character of the message can be represented by the dots in a rectangular dot matrix having fixed dimensions (e.g., 5 dots wide by 7 scan lines high). A character is displayed on the television screen within a character space which includes the dot matrix of the character and additional blank space to separate the characters on the screen (e.g., 8 dots wide by 8 scan lines high). Two such adjacent character spaces are shown in Figure 4 of the Cole patent reproduced below. As the beam tracing the television raster moves across the screen in a scan line, the computer codes for each of the characters to be written in a row across the screen are sequentially provided from a memory to a decoder or “character generator”. As shown in the figure below, a simplified portion of Cole patent Figure 1, timing and control circuitry produce count signals which represent the scan line of the raster and the dot positions along the scan line. The character-code information, the scan-line count signal, and the dot position count signal are applied to the character generator, labeled “Digital-To-Video Generator,” which converts these signals into a two-level, serial digital output. This output is applied to the television monitor circuitry as a video signal.2 One digital level of the sig [940] nal corresponds to a dot and turns on the electron beam to write a dot on the television screen. The other digital level corresponds to the absence of a dot, and leaves the electron beam turned off so that no dot is written. The dots thus produced as the electron beam moves along a scan line correspond to the dots in the appropriate horizontal slice of each of the characters to be displayed in the character row. Thus, in Figure 4 above, dots 103 through 105 (for the character “A”) and dots 106 through 109 (for the character “B”) will be illuminated sequentially as the electron beam moves along the top scan line. The timing and control circuitry also provide horizontal and vertical drive pulses to the monitor to synchronize the scanning motion of the beam with the video signal generated as described above. [941] After completing one scan line, the electron beam flies back to the starting side of the screen, and down one position, to start the next scan line. The sequential application of character codes, scan line count and dot position counts is then repeated, this time generating the video signals for the next dot slice of each of the characters in the row. After the appropriate number of scan lines (e.g., 8 in the 8X8 character space format) have been “written” onto the screen, a full row of characters is complete. The entire row has been written onto the screen, one scan line at a time, from top to bottom. In a like manner, the additional rows of characters making up the message are written on the screen. After the entire screen has been scanned — a process taking Vioth of a second — the procedure is repeated so as to “refresh” the screen and create a display which the human eye perceives as a persistent, non-flickering image. [942] The system described in the Cole specifications is said to provide an important advantage over most prior art devices because it can utilize a standard television receiver for the display without storing video signal output of the character generator. Most prior art systems write each character in its entirety on a CRT screen before going on to the next character to be displayed. The beam in these systems utilized what is referred to as a miniature (or “mini”) raster scan pattern; the beam scanned horizontally back and forth through only one character space at a time, writing the first slice of a character, then the second of that character, then the third, etc. to completion of the character, before going on to the next character space. This had two disadvantages. First, special circuits were needed to cause the electron beam to deflect in this pattern. Second, if the ultimate display in such systems was to be on a standard television receiver, a “scan conversion” step had to be added in which the message, written in character-by-character order, was stored in some sort of memory and then read out an entire scan line at a time so as to be compatible with a beam scanning in a TV raster scan pattern. The Cole system eliminated the need for special deflection circuitry and had no “intermediate storage” between the character generator and the CRT. Its character generator produced the video signal in a sequence and with timing compatible with a beam scanning in a TV raster scan pattern at the speed utilized by a standard television receiver. This “real time” or “on the fly” operation eliminated the need for an expensive intermediate memory for the video information. In addition, RCA urged before the Patent Office, and here urges, that the system described in the Cole specifications with its digital approach to character generation represented a substantial improvement over those prior art devices which employed analog techniques. Analog character generators had problems of accuracy and component tolerances which resulted from variations in temperature, component age, background noise and other impairments inherent in analog systems. The entirely digital Cole system avoided these problems. B. The Cole Patent Claims. Claims 1 through 3 of the Cole patent, the claims in suit, claim the concept for a system and not the particular implementation provided in the specifications as an example. In particular, they claim: 1. A display system for generating character patterns for display on a display device that exhibits a television raster scan-line pattern, each character pattern being displayed in one character space, means responsive to a certain character code for applying to a certain selected lead an output signal having a duration substantially equal to the scanning time in said scan-line direction through one character space, means for generating scan-line select counts in synchronism with the scan-lines of said raster, each scan-line count having a duration substantially equal to that of a raster scan-line, means for generating position counts which occur successively during a scan along a scan-line through a character space, and means for causing said output signal appearing on said selected lead, said scan-line counts and said position counts to supply to said display device a selected character pattern. 2. In a system for displaying a message comprising certain character patterns on a display device that exhibits a television raster scan-line pattern, wherein each different character pattern is manifested by a digitally coded data signal corresponding thereto, the improvement comprising generating means responsive to the data signal forming said message applied thereto for digitally generating a video signal for use in displaying said message on said display device, [943] and means for applying said data signals forming said message to said generating means. 3. The improvement defined in claim 2, wherein said generating means includes first means for producing as said video signal a signal which selectively has either a first level or a second level for the entire duration of each respective one of successive elemental time intervals all of which have the same predetermined duration, the duration of each television raster scan line being an integral multiple of said predetermined duration, and second means coupled to said first means for selecting which of said first and second levels, respectively, exists during each respective one of said successive elemental time intervals in accordance with the data signals forming said message. C. An Overview Of The Technology. In designing a system for receiving digital coded data and displaying it in decoded form on a CRT, there are certain parameters that must be considered. Among these are: 1. The type of scan pattern. As noted above, the two types of primary concern here include one in which the scan covers one character space at a time (a miniature raster scan pattern), and one in which each line of the scan covers a horizontal slice of each character in a row, as the beam scans across the entire width of a CRT screen (television raster scan pattern). 2. The type of CRT. The two principal types are the memory tube, which can hold a picture for minutes, and the non-memory type (including some with high-persistence phosphors), of which a TV tube is an example, that needs to be “refreshed” at a sufficient rate to make the picture appear continuous. 3. The type of character generator. The two broad categories are analog and digital, as related to the type of signal in and signal out, whether continuously variable (analog) or variable only in discrete increments (digital). 4. Storage. A storage or memory is required in a system employing a non-memory CRT because the video signal must be applied to the CRT a number of times a second. The memory may, however, be either one that stores the character code prior to decoding or one that stores the video bits produced by the translation process. When the former is used, the system is sometimes characterized as an “on-the-fly” system to indicate that the video bits are applied to the input of the CRT as each one is generated by the translator in contrast to a system that has storage of the video bits. Each type of raster scan pattern has its advantages and disadvantages. An advantage of the miniature raster scan over the TV raster scan is that the character code may be presented at a slower rate for the same number of characters per row and rows per screen. The two main advantages in using a TV scan are the cost saving in the display portion of the system (the CRT and deflection circuits), and the ability to superimpose characters on pictures or other video (such as maps, etc.). These advantages generally come into play, however, only if one can operate at speeds at least equal to “commercial” or “entertainment” TV rates. Then one can buy a mass produced off-the-shelf display system relatively inexpensively, or simply transmit messages to TV sets that are already being used for other purposes, and also mix character signals with other video signals operating at commercial TV rates. At least the first advantage is lost, however, if the cost of producing character video signals at commercial TV rates exceeds the cost saving in the display portion. As the beam of the CRT scans, the information controlling the on/off condition of the beam must be synchronized with the beam scan. This is true whether the beam follows a TV scan pattern or a miniature scan pattern. If, however, the TV scan [944] pattern is to operate at commercial TV rates, or faster, the requirement of synchronization means that the electronic components must work at rates which were prohibitive in the 1950’s and too expensive to be commercial throughout the early 1960’s. As earlier noted, the coded form of the input data must be translated to video data to control the on/off condition of the beam of the CRT as it sweeps in a mini raster or TV raster scan. The patent literature from the 1950’s discloses translators in the form of digital circuits, for translating from a 6-bit character code to a pulse train which will display a 5 X 7 character matrix on a CRT. These are shown in the Gordon3 and Jones4 patents. Among the analog translators known in the 1950’s was the monoscope character generator. The monoscope generally does not produce a train of 35 equal-length pulse positions, each of which can be on or off and, therefore, it is not generally associated with a character-matrix type of display. Rather, it receives a 6-bit character code and produces a pulse train in which the pulse lengths correspond to the precise width of the character to be displayed, at various locations along the character height. Physically, a monoscope is a small CRT, generally cylindrical in shape, which has a target imprinted or stencilled with characters, instead of a screen. The beam is deflected to a particular character on the target, in response to the 6-bit character code received, and it then scans the character. In response to this scan, the monoscope produces an output signal which is a pulse train having pulse lengths or durations corresponding to the time the beam is crossing the character. This pulse train closely replicates the character shape, and, consequently, produces much better quality characters than obtainable from a 5 X 7 character matrix. All of the character display CRT systems of interest here use signal storage, so that the CRT screen can be repeatedly “refreshed.” Thus, a full screen of character-representing signals (sometimes called a “page”) is stored and used repeatedly to “refresh” the screen. The storage is either of two types. The first type involves storing the 6-bit character codes that are received. The codes can then be read out from storage, and translated to the video signal, as the CRT beam is scanning. The second type of storage involves translating the 6-bit character codes, as they are received, to their corresponding video signals, and then storing the video bits. The video bits can then be read out from storage to the CRT, as the CRT beam is scanning. While both of these storage approaches involve the storage of codes, and both codes are generally binary in nature and represent characters, they can be conveniently referred to as “character code storage” and “video bit storage,” respectively. The disadvantage of video bit storage is that it takes more storage space, since it must store a screen or page of 35-bit character matrix codes — while the character code storage need only store a screen or page of 6-bit character codes. The advantage of video bit storage, however, is that the character codes are only translated once, and the video bits are then stored and read out concurrently with the beam scan— so that the character codes need not be translated concurrently with each scan of a scan line and, therefore, the translation need not keep pace with a high-speed beam scan. Further, video-bit storage is particularly suited to the variable placement of the characters on the screen. II. ANTICIPATION. A. The Dirks System. HLA assert initially that the Cole claims in suit read on a system reflected in Swiss, French and British patents (Nos. 341,665, 1,230,999 and 786,041 respectively) [945] issued to Dirks between 1948 and 1957, none of which were before the Patent Examiner during the prosecution of the Cole patent application. The Dirks system, HLA urge, is the Cole system implemented in 1940’s digital technology, and, since the Cole claims are drawn to cover all digital systems generically, as opposed to a new implementation, they are anticipated by these foreign Dirks’ patents. I agree. Before discussing the novelty of the claims in suit in light of the Dirks’ disclosure, however, a threshold issue must be resolved. RCA invokes that portion of 35 U.S.C. § 112 which permits claims in terms of “means and function” and which provides that they “shall be construed to cover the corresponding structure ... described in the specification and equivalents thereof.” RCA construes this provision to mean that Claims 1 through 3 of Cole should be read as being limited to the implementation disclosed in the specifications. Since that implementation and the implementation disclosed in Dirks are concededly distinct, RCA urges that Section 112 precludes any finding that Dirks anticipates Cole. The statutory language relied upon by RCA was added to the Patent Act by amendment in 1952. There is some debate as to the purpose of that amendment. See, e.g., 2 Chisum, Patents, § 8.04[2]. There is general agreement, however, that it was not intended to have the effect claimed by RCA. A contemporaneous commentary by the Assistant Commissioner, for example, points out as follows: The paragraph ends by stating that such a claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof. This relates primarily to the construction of such claims for the purpose of determining when the claim is infringed (note the use of the word “cover”), and would not appear to have much, if any applicability in determining the patentability of such claims over the prior art, that is, the Patent Office is not authorized to allow a claim which “reads on” the prior art. P.D. Federio, “Commentary on the New Patent Act,” 35 U.S.C.A. 25-26. This view has been endorsed by the courts. The lead ease is In re Lundberg, 44 CCPA 909, 244 F.2d 543 (1957); also in accord are Siegel v. Watson, Com’r, 267 F.2d 621 (D.C.Cir.1959), In re Edwards, 48 CCPA 808, 285 F.2d 811 (1961) and In re Henatsch, 49 CCPA 915, 298 F.2d 954 (1962). Thus, while the claims in suit must be read in the context of the specifications, they may not be read as limited to the implementation which they disclose. The Dirks patents disclose a completely digital, video display system using a TV raster scan pattern. It stores character codes on a rotating magnetic medium, and, on repeated rotations, the stored character codes are repeatedly translated. During the first revolution, the stored character codes are successively translated to form the video pulse train for the first slice of each character on a character row; then, on the second revolution, the second slice of each character, and so forth, until a character row is scanned. Other character rows can then follow. The Dirks system can best be understood by reference to Figure 2a of the British patent: [947] The character codes are stored on a moving magnetic medium 38, and sensed by a pick-up head 34, as the magnetic medium 38 rotates. Each code selects a different tube 36, which, in turn, selects a core matrix 50 to 59. The video signal from a selected matrix is supplied to an amplifier 72 and CRT 44, in synchronism with the scan of the beam of the CRT. The stepping switch 46, which can be an electronic switch, keeps track of the scan line that the CRT beam is traversing. The magnetic yokes 71° to 719 all move in unison, to scan the respective matrices 50 to 59, the successive dot positions of each matrix being scanned as the beam scans across a character space on the screen. By way of example, if a character code for a “0” is sensed on the moving magnetic medium 38, in Dirks Figure 2a, tube 36° will be actuated to select the core matrix 50 showing a “0” outline. As the yoke 71° then scans the matrix rows 50° to 509, in synchronism with the beam scan of positions 0 to 9 across a character space, with switch 46 in its left-most position, the top slice of the “0” will appear as bright dots at positions 3, 4, 5 and 6 across the character space. Then the matrices selected by the succeeding character codes on medium 38 will be scanned in turn and the top slices of each will be written as the beam proceeds across the CRT screen. As the magnetic medium 38 is then rotated a second time, switch 46 is moved one position to the right, and the CRT beam begins to scan a second scan line. During this second rotation, the video bits for the second slice of each character are produced in similar fashion, and supplied to the CRT 44, without any video bit storage, and this continues until a row of characters has been displayed. Claim 1 of the Dirks British patent refers to the generation of a video pulse train having successive pulse positions, each of these pulse positions corresponding to a dot position across the character space on the CRT screen. As indicated in this Dirks claim, the Dirks system includes means for generating pulses in any one of the pulse positions, to produce a pattern of pulses, different patterns being representative of different characters, together with a selecting means (tubes 36, and the associated input circuitry), controlled by the stored signal in the form of the character codes on medium 38, to determine in what positions the pulses are to be generated. While the implementation of the system is quite different in Dirks than in Cole, a comparison of the Cole claims with the Dirks’ disclosure reveals that the former describes the latter precisely. Such a comparison is made in Appendix A to this opinion. I will here discuss in detail only the three primary distinctions which I understand RCA to urge: (1) Dirks does not utilize a standard television display, (2) Dirks “machinery controls the CRT beam movement,” and (3) Dirks is not an “entirely digital” device. Each of the claims in suit calls for “a display system for generating character patterns for display on a display device that exhibits a television raster scan line pattern.” RCA finds in this language a limitation to a system utilizing a standard television display which is said to be unlike anything shown in Dirks. While it is true that Dirks does not teach the use of a standard television receiver, RCA’s argument is unsound because it misreads the Cole patent and ignores a portion of the Dirks disclosure. Each of the claims in suit refers, not to a standard television receiver or monitor, but to any “display device which exhibits a television raster scan line pattern.” This reference to a type of deflection pattern rather than to a standard television receiver was clearly deliberate. The specifications demonstrate that the drafter specified a “standard television tube” when that is what he meant and the file wrapper indicates that a system for displaying on a “television display” or “television monitor” was what he claimed when the application was originally filed. The original claims were rejected, however, and, when the amendments were drafted which added the claims in suit, the new draftsmen broadened the scope of the [948] invention claimed to include any device exhibiting a TV raster scan pattern. The effect of the decision to claim devices with a particular pattern was explained by RCA’s expert at trial. Professor Ward pointed out that, while the specifications stressed the advantage of being able to use the invention with a standard television receiver, the claims were not so limited. In the course of discussing the design parameters of a standard television set, he observed: Cole very carefully explains the operation of the system to the extent that anybody of ordinary competence can derive his own numbers of his own parameters. The system is not intended to be restricted to these numbers. It is not intended to be a system that will only work with the standard television raster scan ray, but is intended in fact to work with systems that have television raster scan like deflection patterns.5 The claims in suit thus cover any CRT which utilizes a TV raster scan line pattern and we must turn to the Dirks patents to determine whether the Dirks’ system discloses the use of a CRT having such a pattern. Dirks teaches the use of a CRT with a television raster scan line pattern. This is graphically illustrated in Figures 51b to 51d of the Swiss Dirks patent. (HLA Exh. 1). Those figures show, as the text of the other Dirks patents teach, the beam scan crossing the entire screen with the characters being built up slice-by-slice during the successive scans of the beam. While Figure 2b of the British patent, shown above, has only one line of numerals, the specifications expressly teach that the system may be employed to produce “a number of rows of characters instead of one row only, the ray after scanning the bottom line of characters in one row then being deflected still further downward to write the top line of the following row.” (PX 3, Tab. 34, p. 9, lines 39-44). Even if this suggestion had not been made by Dirks, however, it would nevertheless remain true that his CRT exhibited a television raster scan line pattern. In its briefing, RCA points out that “the Dirks machinery controls the CRT beam movement.” “In contrast,” it is said, “the Cole system produces a video output which is compatible with the independent timing of a standard television raster scan.” It may be that all that RCA is saying here is that Dirks does not use a standard television receiver. If so, its argument has already been answered. RCA seems to view this as a separate point, however, emphasizing that a crucial part in the Cole invention is “accepting the constraints of the independent timing” of a standard television receiver. As the above-quoted testimony of Professor Ward indicates, the Cole claims are not limited to the timing of a standard television receiver. Nor do they speak of the source of the timing of the CRT beam. All that is required is the generation of counts “in synchronism” with the travel of the beam. As previously indicated, the Dirks system is designed to do precisely that. Moreover, Dirks scans horizontally using a sawtooth waveform and, thus, just as Cole, discloses an autonomous or independent time base, notwithstanding his controlled stepping in a vertical direction.6 Finally, RCA finds, in Claim l’s requirement of a “means for generating position counts,” a limitation which excludes a system employing continuously moving yokes, said to be an “analog” or “continuously variable” component of the Dirks system. [949] Similarly, in connection with the reference in Claims 2 and 3 to a “means ... for digitally generating a video signal,” RCA argues that the Cole system is “all digital” while the Dirks system is not. I conclude, however, that RCA and its expert read the word “digital” more narrowly than did artisans of ordinary skill in the art in the early 60’s. The character generator of the Dirks system can produce only on/off signals, for each successive pulse position of the video pulse train. There is no value significance to any voltage derived from a Dirks core matrix until the voltage reaches a level to produce a pulse at the output of amplifier 72 in Figure 2a. Anything above that level produces an output pulse; anything below it does not. Amplifier 72 thus acts as a threshold device, producing “0” ’s for voltage input levels below the threshold, and “1” ’s for voltage input levels above its threshold. This is purely digital operation. In the terms of Professor Ward’s explanation of the concept of a digital device, the Dirks’ system abstracts from the voltage pulses produced by the Dirks matrices the binary levels of interest, and throws away any information in the form of any other variations in the pulse levels. All pulses have significance only in the respect that they are above or below the threshold, i.e. “1” or “0”. No analog information is needed, and none is retained. This on/off, discrete operation for each pulse position of the resulting pulse train is precisely the operation that occurs in the later magnetic core matrices, such as shown in the Gordon patent and Cole patent disclosure data sheet, and also in the block diagram shown in Figure 2 of the Cole patent. This is the concept which the contemporary art associated with the word “digital.” See, e.g., HLA 500-503; Larky Tr. 2673-74; HLA Ex. 514, p. 33.7 Thus, the artisan of ordinary skill in the art in the early 1960’s would characterize Dirks’ character generation as “digital,” even though there are some analog aspects to its components. As Professor Ward acknowledged, there are analog aspects to the components of all digital systems. While Professor Ward agreed that the Dirks’ system operates in the manner I have described, he maintained that it was not entirely digital because its timing is centered around a shaft that rotates in a continuous way for 360 degrees and because the positions of the coils and cores around the shaft are “continuously variable.” Professor Ward acknowledged, however, that the presence of a rotating element is not inconsistent with the concept of a digital system. The Amdahl system, for example, was described by Professor Ward as an all digital system, even though it employed a continuously rotating magnetic drum. Moreover, while it is true that the Dirks system could be designed with coils whose positions would be variable, the systems disclosed in the Dirks patents make no provision for varying their positions and, indeed, it is clear that they are intended to remain fixed in position. With regard to timing, since the Dirks drum is continuously rotating at a constant speed and since the cores of the matrices are fixed about the drum, there is no intentional variation in the timing of the pulse signals produced. Any such variation, if it occurs, will be unintended. The possibility that such unintended variation may occur does not change what is a digital system in concept to one that is analog. Indeed, the same possibility exists with the Cole system. RCA’s Mr. Wine agreed that variations in timing through the matrix of Cole’s character generator are inevitable. He also testified that this did not make Cole something other than a digital system. In addition to asserting that the claims in suit do not read on Dirks, RCA urges that Dirks is a “bizarre” piece of “farm machinery” which simply will not work. It relies upon case law indicating that an invention which is inoperable cannot anticipate. E.g., United States v. Adams, 383 U.S. 39, 50, 86 S.Ct. 708, 713, 15 L.Ed.2d 572 (1966). [950] It is true that the Dirks technology was passé even by the time of Cole and that it could not come close to the speed necessary to display a message upon a commercial television screen as the Cole system does. I am persuaded, however, that one of ordinary skill in the art in the 50’s could have implemented the Dirks design so as to display a message in a row or two of alpha numeric characters on a cathode ray tube. Professor Larky, who I believe is more familiar with the state of the art at that time than is Professor Ward, so testified. Moreover, I do not understand Professor Ward to have expressed an opinion to the contrary. He acknowledged that the Dirks system might achieve the speed necessary to display a row of characters and, while he believed the yokes would produce random pulses tending to obscure the desired pulse trains, he did not opine that this was a problem beyond the skill of mechanics from the Dirks’ era. The Cole invention was not a specific device; it was a system concept for direct production of digital video bits from coded data and was operable when appropriately implemented by those skilled in the art who would design the requisite digital networks and sub-assemblies, none of which was disclosed or claimed in the Cole patent. The Dirks patents disclosed the same concept of converting coded data directly into pulse trains representing the appropriate characters, timed and arranged by character slices for display in a television type raster scan pattern. Like Cole, the Dirks concept was operable when properly implemented by those skilled in the art. This is sufficient to qualify Dirks as prior art under Section 102, even though the Dirks’ implementation was crude in comparison to later embodiments. Zephyr American Corp. v. Bates, 128 F.2d 380, 385 (3d Cir.1942); Sutter Products Co. v. Pettibone Mulliken Corp., 428 F.2d 639, 646-47 (7th Cir.1970). B. Fenimore And Amdahl. HLA contend that the claims in suit are anticipated by the Fenimore8 and Amdahl9 patents as well as Dirks. I do not agree. The systems taught in the Fenimore and Amdahl patents used intermediate storage of the video bits from their character generators to perform the operation of converting from miniature raster scan to TV raster scan order. In these systems, the pattern for each character in a message was generated once, and only once, by the character generator. As each bit pattern was generated, one character at a time, it was written into intermediate storage to form a bit map of the entire message. After the bit map was formed, it was read out of intermediate storage, as an entirely independent operation, to generate the video signal for each refresh of the television screen. The character generator was not involved in this generation of video signals for each refresh cycle; it remained inactive until the message was changed. Claims 2 and 3 of Cole require that the “video signal generating” means be “responsive” to the “data signal” (i.e. character codes). I agree with Professor Ward that this requirement excludes intermediate video bit storage systems in which the video signal generating means is “responsive,” not to the character codes, but to an intermediate storage bit map. This conclusion is based in part on the fact that the specifications emphasize that the Cole invention operates “on the fly,” that is, that the video signal is generated directly by the character codes with no intermediate operation. III. OBVIOUSNESS. A. Level Of Skill In The Art. Many technically sophisticated engineering groups across the nation recognized the advent of the computer age and attempted to design display systems for computer generated messages in the 1950’s and early 1960’s. Among others, these included employees of the Hazeltine Technical Development Center, Raytheon, the Stanford Research Institute, Ford Aeronutronics, Stromberg-Carlson, Wang Laboratories, [951] ITT, CBS, Bell Telephone, and Hughes. Typically, these groups were composed of graduate engineers with substantial research and development experience in the display field. Those working at Hazeltine were recognized as leaders in this technology- B. The Relevant Prior Art. In support of their obviousness argument, HLA rely not only on Dirks but also upon Jones, et al.,10 Gordon, et al.11 and Evans, et al.12 RCA makes reference to an additional piece of prior art, the British Ellson13 patent, which it claims to be more pertinent than Jones or Gordon. Evans was cited by the Examiner during the prosecution of the Cole patent; Jones, Gordon and Ellson were not. While RCA did report to the Examiner that Ellson was cited during the prosecution of the British Cole application, it neither described that patent to him nor supplied him with a copy. Since I am unable to find it more likely than not that the Examiner considered Ellson, that patent must be disregarded in determining whether the Cole patent is entitled to the statutory presumption of validity.14 The Jones patent application was filed April 16, 1958, and the patent issued on June 6, 1961. It discloses a character generator for use in displaying alpha-numeric characters on cathode ray tubes. This character generator utilizes a matrix of diodes to translate the character codes and is acknowledged by both sides to employ “digital” character generation like that spoken of in Cole. Jones discusses the earlier use of monoscope character generators and indicates that a monoscope has certain disadvantages not found in its diode matrix, digital character generator. Jones’ character generator is designed for a display with a mini raster scan, rather than a TV raster scan, pattern. The Gordon patent issued on January 5, 1960, on an application filed August 13, 1953. It also mentions certain disadvantages of a monoscope character generator and suggests the use of a digital translator much like the magnetic core character generator of the patent disclosure data sheet prepared by the Cole inventors. Gordon refers to a video display system using a mini raster scan, rather than a TV raster scan, pattern but, like the Jones patent, focuses primarily on the process of character generation rather than an entire display system. Evans was filed on May 8, 1959, and issued on January 16, 1962. The Evans system includes a memory for storing digital character codes in the sequence in which the characters are to be displayed on the TV set and a monoscope character generator which translates the digital character codes to video signals to be supplied to a standard television receiver. Evans also includes digital timing circuits for producing row counts for each character row on the screen (one such count for each 27 scan line counts) and column counts. The Evans column counts can also be regarded as character counts. The Evans row and column counts reflect the position of the scanning beam of the TV set. They are used to access the memory so that the character codes are supplied from the memory to the monoscope translator and thence to the TV set, without intermediate storage, in synchronism with the scanning of the beam. Ellson discloses a printer which uses a digital character generator consisting of a matrix of magnetic cores. The printer operation of Ellson is such that the character [952] codes for a row of print are recirculated in a storage buffer and are repeatedly and sequentially applied to a decoder. The decoder enables the appropriate cores in the character generator by activating a character-shaped winding corresponding to the character code. A line counter input causes the appropriate dot slice to be read out of the character generator. The next character code is then decoded and the appropriate dot slice of that character is read out. In this manner character dot information for a slice through all characters in a line is generated. The dot slices for all the characters in a row of characters are collected and then sent to a group of styli for simultaneous printing, one stylus for each character. Each stylus moves in a miniature raster scan pattern. For present purposes, the most significant feature of Ellson is its digital character generator. C. Comparison Of Cole With The Prior Art And The Obviousness. It will be recalled that the Cole system stores the incoming character codes in a memory. They are thereafter read out to a digital character generator which, in response to the character code, the position count and the scan line count, forwards slices of each successive letter on a scan line, “on the fly,” to a CRT for display in a TV raster scan pattern. It thus reflects in Dr. Ward’s words, an “all digital approach to the translation of coded character information to television raster scan form without the use of intervening .. . storage” of video bits between the character generator and the CRT. Thus, Cole, like Evans, translates digitally coded information for display, without intermediate storage, in a TV raster scan pattern, but it accomplishes the translation with a digital character generator, rather than a monoscope. As earlier noted, Jones and Gordon teach digital character generators much like that in Figure 2 of Cole. They do so, however, in the context of mini raster scan systems which present character-at-a-time patterns and they do not confront the problem of converting the character shape information to full scan line sequence. The primary Section 103 issue15 presented by the parties is whether it would have been obvious to one of ordinary skill in the art in October of 196316 to combine Jones or Gordon with Evans in such a way as to produce the concept claimed in Cole. While the fact that the Examiner apparently did not consider a prior art patent disclosing a digital character generator deprives the Cole patent of its presumption of validity, I nevertheless conclude that Cole is not obvious based on Evans in light of Jones or Gordon. In support of their argument that the hypothetical artisan of ordinary skill would have combined these references, HLA stress that Jones and Gordon expressly suggest the use of a digital character generator as preferable to a monoscope and that Mr. Stocker, one of the inventors of the Cole patent, candidly acknowledged that magnetic core, digital character generators were sufficiently well known in 1963 to be a “designer’s choice” alternative to a mono-scope. HLA ignore the context in which the suggestions of Jones and Gordon and the acknowledgment of Mr. Stocker were made, however. When Gordon and Jones did the work reflected in the patents which bear their names, neither they nor their fellow artisans on this side of the Atlantic had been exposed to the concept of “on the fly” generation of video signals for display in a TV raster scan pattern. With the exception of Dirks and Amdahl, every prior art reference which teaches some form of digital character generation does so in the context of a mini raster scan system. Amdahl, while suggesting the use of a TV raster scan pattern, solved the problem of convert [953] ing from character-by-character order to scan line order by creating a representation of the message page in an intermediate storage which could be read out in scan line order. Given this state of the art at the time of Jones and Gordon, I think it highly unlikely that the artisan of ordinary skill in 1963 would have read Gordon’s or Jones’ comments about monoscopes as a suggestion that one could or should substitute a digital character generator for a monoscope in an environment involving a TV raster scan and no intermediate storage. Similarly, Mr. Stocker’s testimony was that digital character generation was well known in the 50’s in the context of mini raster scan systems.17 He said nothing in his deposition or at trial which would suggest that the use of a digital character generator was a matter of designer’s choice in any other context. When the Evans patent was issued in January of 1962, it thus taught something new to those working in the art, — on the fly generation of signals for display on a standard TV monitor. In this totally new environment, Evans taught the use of a mono-scope to generate the character shaping information. Nothing in Evans suggests, expressly or by implication, the possibility of using a digital character generator. While with the benefit of hindsight there appears to be no reason why the concepts of Gordon and Jones could not be used together with those of Evans, I find no suggestion in any of these references, or anywhere else in the prior art, that they could or should be so used. While HLA’s briefing suggest that even a technician would have perceived the feasibility of substituting a digital character generator for the monoscope in Evans’ system, the timing and circuitry of the Evans' system and the Gordon and Jones’ systems are sufficiently different that I do not believe this to have been the case. Indeed, there is no evidence from which one could conclude that an artisan of ordinary skill would have had such a realization when examining these patents. Significantly, while Dr. Larky gave detailed explanations about how one might rearrange Jones and about the creating of a series of digital timing circuits in the context of Jones, he was silent about how or whether one could or would combine Evans with Gordon or Jones or what that combination would be. Similarly, Evans gave no testimony as to how his system could or would be combined with a Jones or Gordon digital character generator. And while HLA counsel cross-examined Dr. Ward at length regarding the Cole patent and the prior art, they never brought up any combination of Gordon or Jones with Evans. Indeed, no HLA witness testified that it would have been obvious in 1963 to combine Evans with Gordon or Jones. While the prior art taught the use of digital character generators only with either a miniature raster scan or intermediate storage, I do not find, as RCA contends, that the art “taught away” from Cole by urging “isolation” of digital character generators from the timing of a television receiver. RCA suggests that there was a problem getting a digital character generator to function in synchrony with the autonomous time base of a standard TV receiver and that the art taught away from such an approach. I find no such teaching and no such problem evidenced by this record, however. While there was a substantial problem through the 50’s and early 60’s in achieving the speeds necessary for utilizing a standard TV monitor while keeping costs within feasible limits, I am not persuaded that the designing of a system which operated in synchrony with the timing of a TV monitor was any more difficult than designing a system utilizing any other kind of CRT functioning at the same speed. Nevertheless, in the absence of any suggestion in the prior art that the Evans system could or should be used with Jones or Gor [954] don character generator, I am unwilling to infer that such a combination was obvious in 1963. Such an inference would be particularly inappropriate when the record shows that those working in the art in the early 1960’s did not, in fact, find the Cole concept obvious from Evans, Gordon and Jones. I find this particularly significant in light of the fact that digital character generators had been a part of the common knowledge of the art for many years when Evans’ ideas became available. With the exception of a young man at Hazeltine named Wagner, none of the numerous experts working in the field came up with the Cole concept during the two years following the issuance, of Evans.18 There is no documentation suggesting that Evans himself thought of using a digital character generator in his system and I believe that he did not. Leland Nave, a co-patentee of the Fenimore patent, testified to the development of display systems at the Hazeltine Technical Development Center. His testimony and contemporaneous documents show that by the early 1960’s, he and his thirty colleagues were one of the most advanced and experienced display groups in the country. This group came up with three different systems in attempting to solve the problem of displaying computer generated messages. They failed, however, to see what Cole, et al. saw. While these Hazeltine experts were exposed to the Cole concept as a result of Wagner’s work, they believed that use of digital character generators for on the fly TV raster scan displays would require a character generator for every character position in a character line. (DX 323, pp. 1-6).19 These experts, accordingly, chose an intermediate storage television display approach as their solution to the problem of displaying full text messages. Despite the discovery which has been conducted in connection with this litigation, there is no documentary evidence of anyone in the field, other than Wagner, coming up with the Cole concept within two years after the issuance of Evans in January of 1962. If it had been obvious, one would certainly expect that there would have been more discerners of the “obvious” given the number of firms working in the computer display area.20 HLA attempt to explain away the failure of others to see the “obvious” by suggesting that digital character generation for on the fly TV display was too expensive to be commercially successful in the early 60's. While the cost of this approach was materially reduced with the advent of the integrated circuit and RCA cannot claim commercial success until a point in time after that advent, this does not explain the absence of any evidence of conceptions like that of Cole, et al. Engineers who conceive inventions in an art like this one write them down. They lay out the circuits and then investigate the cost and component speed requirements. This is particularly true where, as here, the conception involves, in Dr. Ward’s words, a “simple and elegant” solution to a problem on which the art is intently focused. Moreover, every witness who testified on the subject confirmed that the artisans of the early 60’s knew that the cost of components was coming down. No one was going to refrain from having an idea, or decide not to write it down, because of the cost or component speeds. We know, for example, that the Fenimore and Am955 dahl high-cost systems were conceived and written down in detail. Prices for Feni-more and ANG systems ranged from $60,-000 to $1.25 million. My conclusion regarding non-obviousness is further buttressed by the evidence concerning RCA’s licensing program. RCA’s major licensing discussions began with Ha-zeltine in June 1970. After many meetings and after a thorough review of prior art which included Evans and a number of digital character generator patents, Hazeltine agreed in November 1974 to pay RCA a royalty of $6.50 for each terminal using the invention of the Cole patent, effective as of January 1,1974. Hazeltine was well aware of the history of its own work on displays, its personnel were experienced in patent matters, and it signed the license only after careful consideration of the substantial royalties which would be owed under the license. Many others thereafter followed Hazel-tine’s lead. Prior to the end of 1980, more than 40 manufacturers of video display terminals had taken licenses under the Cole patent and had paid substantial royalties to RCA based upon licensed sales of more than one million terminals. As of November 1, 1981, the number of companies which had taken licenses had grown to 64 and RCA had received royalties of approximately $9.1 million based upon sales of almost 1.4 million terminals using the Cole invention. This industry acceptance of the Cole patent came despite the fact that, prior to the taking of their licenses, all of the licensees were aware of Evans and of the common use of digital magnetic core character generators throughout the 1950’s. Moreover, by 1974 the largest of these licensees were aware that the demand for display terminals utilizing the Cole concept would be such that very substantial amounts of royalties would be payable by them under the licenses they were being offered. I note in this connection that the five licensees with the highest volume paid approximately $5.6 million in royalties between January 1,1974 and November 1, 1981. This fact, together with the fact that the steady increase in demand was expected to continue throughout the life of the patent, indicates that there were those with a sufficient financial stake to litigate if the industry had believed that it was obvious in 1963 to substitute a digital character generator for the mono-scope in Evans.21 IV. ENFORCEABILITY HLA’s argument that the Cole patent is unenforceable rests primarily on the fact that RCA failed to call the Jones patent to the attention of the Examiner. This failure to focus attention on a patent involving a digital character generator is said to be a particularly egregious default because RCA, in 1967, successfully distinguished Evans before the Patent Office on the ground that Cole had a digital character generator which was superior to the monoscope used by Evans. While it is true that the Jones patent was known to RCA and undisclosed by it, I find no basis for concluding that there was deceptive intent or gross negligence on RCA’s part. George Seligsohn, the attorney who drafted the amended claims and “Remarks” in 1967, was unaware of Jones. While his work was reviewed by Edward Norton, now deceased, who had been cited to the Jones patent in 1963, there is no reason to believe that he focused his attention on it in 1963, much less that he remembered it in 1967 during the course of his review. The Jones patent was cited to Norton by number in a list of nine patents included in a report on an infringement study secured in connection with the marketing of RCA’s DIVCON system. That report never became a part of the RCA file on the Cole patent. [956] If Norton in fact read the Jones patent in 1968, there is reason to believe he may not have focused on its significance. As earlier noted, it does teach a digital character generator and suggests that it is superior to a monoscope. It does so, however, in the context of a mini raster system. While it is nevertheless pertinent when studied in connection with Evans, there is no reason to believe that Norton studied them together, or was even aware of Evans in August of 1963. I also note that RCA, prior to its attempts to distinguish Evans after the first office action, referred the Examiner to the British Ellson patent which discloses a digital character generator and is as pertinent as Jones. While it is not clear that either Seligsohn or Norton was knowledgeable about the Ellson patent, it was in the Cole file and if there had been some intentional scheme afoot to mislead the Examiner, one would certainly have expected them to have ascertained what had previously been referred to the Examiner. Having completed such a task, one would realize that pursuing an approach on the assumption that the Patent Examiner was unaware of digital character generators would be an uncertain course. I have considered HLA’s subsidiary charges of inequitable conduct on the part of RCA, but find no RCA activity which would justify a holding that the Cole patent, if valid, would be unenforceable. V. INFRINGEMENT. HLA urges that their terminals are distinguishable from the Cole invention on two grounds and that, accordingly, a finding of infringement is precluded. These distinctions are without substance, however, and the HLA terminals are the equivalents of the system claimed in the Cole patent. Thus, if the Cole patent were valid, HLA would be infringers. HLA first argue that the use of a read only memory (“ROM”) and shift register combination, rather than the illustrative “one example” of a digital-to-video generator shown in Cole Figure 2, requires a finding of non-infringement. The HLA ROM/shift register combination is the direct equivalent, however, of the digital-to-video generator shown in Cole. The HLA ROM/shift register receives the digital character code, scan line count and dot position count inputs, and generates the digital, two-level video signal output. So also does the Cole digital-to-video generator. Thus, consistent with the Cole patent description, Lear Siegler’s infringement expert witness, Douglas H. Thomson, characterized the ROM/shift register combination as a “char-aeter-to-video converter.” HLA argue that the character code, scan line count, dot position count and digital video output signals are not precisely simultaneous in a ROM/shift register digital-to-video generator. Nothing in the Cole patent disclosure or claims requires them to be, however. Moreover, contrary, to HLA’s suggestion, the shift register does not constitute intermediate storage and result in the “means for digitally generating” the video signal being anything other than “responsive to the data signal.” The shift register of the HLA terminals is functionally a part of the HLA character generator. In the HLA terminals, just as in the embodiment illustrated in the Cole patent, each character code in a row is repeatedly presented to the character generator, and the character generator generates the video signal directly in response to that character code in real time. The shift register is nothing more than a parallel to serial converter. It receives from the ROM, in parallel, a group of bits defining the dot pattern for a slice through a character and delivers them, in series, at its output terminal. The bits do not remain “stored” in the shift register. They cannot be repeatedly read out. Rather, the bits move through the shift register. No bit remains in the shift register for more than a microsecond and the bit slice is not delayed in the shift register for a character space. The combination of the ROM and shift register produces the video signal, in real time and in synchronization with the television raster scan. The one character [957] space “delay” referred to by HLA is the time it takes for the ROM/shift register combination to generate the video signal after, the character code is applied, due to propagation delays and the “pipelining” effect of the combination. Like the character generator of the Cole patent embodiment, HLA’s ROM/shift register combination is directly affected by and must operate in synchronization with the ■television raster scan. The ROM produces a slice of dots at the character rate and the shift register outputs those dots at the dot rate. Unlike intermediate storage systems, this process continues throughout the raster scan, and is repeated over and over again with each refresh cycle. HLA also contend that the signals applied to the shift register are a dot clock and a shift/load signal rather than dot position counts. But each of the HLA terminals has a dot counter. The dot counter cycles through a set of counts corresponding to the dot positions within a character space, generating the shift/load signal at an appropriate count. Following the shift/load signal, the dot clock signals shift the dot bits out of the shift register. HLA’s expert Thomson acknowledged that the sequence of each dot clock signal after the shift/load signal corresponds directly to a dot count of the dot counter. Thus, the dot clock and shift/load signal provide the dot position count information to the shift register.22 Significantly, neither Thomson, nor any of the other three HLA terminal engineers listed as witnesses, but not called by HLA, attempted to counter testimony of RCA’s Wine that the HLA terminals use the Cole invention and are described by Cole patent Claims 1, 2 and 3. VI. LACHES AND ESTOPPEL. The relevant questions raised by ADDS’ laches and estoppel defenses are: (1) Did RCA delay unreasonably in enforcing its rights against ADDS, (2) if so, was ADDS prejudiced by that delay, (3) did RCA mislead ADDS into believing that it would not enforce its rights, and (4) was ADDS thereby misled to its detriment? RCA advised ADDS of the existence of the Cole patent in late 1969 and licensing discussions ensued. At a July 1,1970 meeting, ADDS insisted it was not infringing and, in support of this position, agreed to supply RCA with a “detailed block diagram of the read only memory and the video signal generator, together with an explanation of the operation.” (PX 620). This information was not forthcoming, however, and RCA’s licensing personnel ultimately diverted their attention to bigger fish. Their efforts focused primarily on Hazel-tine, to whom much of the industry looked for leadership in licensing matters. After Hazeltine agreed to take a license, contact was reestablished between RCA and ADDS in October of 1974. At this point, ADDS’ patent counsel referred RCA to two ADDS’ patents which were represented as providing information about the structure and operation of the ADDS’ terminals. After RCA had reviewed the two ADDS’ patents and stated that it believed further discussions were advisable, ADDS’ counsel took the position that ADDS’ terminals did not necessarily operate in the manner described by the patents. After the renewal of contact, numerous meetings and correspondence followed. RCA continued to seek more information from ADDS about its terminals. While it [958] did not supply the requested information, ADDS continued to assert that its terminals did not infringe the Cole patent and, beginning in 1976, that the Cole patent was invalid. It took care, however, not to break off the discussions. In February of 1973, for example, ADDS’ counsel reported to its president as follows: RCA’s licensing representative, Peterson, seems to take the view that it is a foregone conclusion that ADDS will take a license. I did not encourage this view, but did little to discourage it, taking the posture that it was necessary for me to review the file in view of the long time that had elapsed since I heard from him. Licensing discussions continued until May of 1978, when RCA filed suit. By late 1969, RCA was on notice of the possibility that ADDS was infringing the Cole patent. By January 1, 1971, six months after the July 1, 1970 meeting, a reasonable patent owner would have taken steps to determine whether ADDS was infringing, either by pressing ADDS for the information it had requested or by securing that information in the marketplace. If RCA had made such efforts in late 1970 or early 1971, it would have been able to secure a manual which described the structure and operation of the ADDS terminals. In most Circuits, the passage of six years after a patent owner should know of infringing activity gives rise to a presumption that he has been guilty of unreasonable delay and that the alleged infringer has been prejudiced. The result in such jurisdictions is that the plaintiff assumes the burden of proving either an excuse for the delay or an absence of prejudice.23 The laches law applied by the Third Circuit in non-patent areas is consistent with this view24 and I predict that, when the question is squarely presented to that court, it will decide that this rule should be applied in patent cases as well. If the generally prevailing rules are applied to the facts of this case, the appropriate conclusions are that the six year statutory period commenced on January 1, 1971 and, since suit was not filed until May of 1978, that RCA has the burden of proof. See 4 Chisum, Patents, § 19.05. I agree with ADDS, however, that Minnesota Mining & Manufacturing Co. v. Berwick, Indus., Inc., 532 F.2d 330 (3d Cir.1976), in which the court stated that the burden of proof on laches issues rested with the alleged infringer, is factually indistinguishable from this case. Fortunately, however, the conclusions compelled by this record are the same regardless of where the burden of proof is placed. RCA has shown no excuse for the three year and ten month delay between January 1,1971 and October 29,1974 and, if ADDS had not led RCA to believe there was a realistic possibility of it taking a license during the period from October of 1974 until the eve of suit, I would hold that there had been unreasonable delay. I conclude, however, that RCA was reasonable in believing during this latter period that the matter be resolved without resort to litigation. Moreover, I am satisfied that there has been no prejudice to ADDS from the delay which occurred. ADDS believed from the outset that RCA would eventually conclude that ADDS was infringing and that RCA would sue if ADDS did not take a license. It knew RCA had too much invested in its licensing program to overlook any infringement by ADDS. Its strategy was to delay as long as possible the day of reckoning when ADDS would be forced to sign or be sued. Thus, ADDS was never misled by RCA into believing that RCA would not enforce its rights under the Cole patent. It was on notice, almost from the outset, that it was likely to be involved in litigation if it did not take a license and, indeed, knew that it could precipitate a suit at any time during the period from 1974 to 1978 simply by firmly refusing to take a license. [959] While ADDS expanded its terminal business substantially between 1969 and 1978, I think it clear that ADDS management, anticipating that ADDS would eventually take a license or be sued, conducted its business no differently during that period than it would have if a suit had actually been pending. Finally, I agree with RCA that the record does not support the proposition that significant documentary and testamentary evidence was lost during the delay to the prejudice of ADDS. Ample evidence, both testamentary and documentary, was available on every issue that it was possible to litigate. The only affirmative suggestion of possible prejudice in the record relates to the unavailability of Mr. Norton at trial due to his demise.25 Mr. Norton did not die, however, until August of 1980, fifteen months after suit was filed. ADDS had ample time to take his deposition prior to that date. Based on the foregoing facts, I conclude that, if the Cole patent were valid, RCA would not be foreclosed by laches or estop-pel from enforcing it against ADDS. VII. WILLFUL INFRINGEMENT. My finding of anticipation, of course, precludes any conclusion that HLA are guilty of willful infringement. Even if the Cole patent were valid, however, I would refuse to find that this is an exceptional case within the meaning of 35 U.S.C. § 285. While Hazeltine and Lear Siegler stopped paying royalties in 1981, I am not persuaded that they did so without a good faith belief in their ability to win the litigation that was sure to follow. VIII. CONCLUSION. In summary, I conclude that the Cole patent is anticipated by Dirks, that it was not obvious from Evans in light of Jones or Gordon, and that it is not unenforceable by virtue of RCA’s conduct before the Patent Office. I further find that, if the Cole patent were not invalid under Section 102, it would be infringed by the HLA terminals, though not under such circumstances as would make this an “exceptional case” within the meaning of 35 U.S.C. § 285. Finally, I conclude that neither laches nor estoppel preclude RCA from making the claims which it has here asserted against ADDS. APPENDIX A [961] *The Dirks system is also disclosed in Dirks French patent 1,230,999 and Dirks Swiss patent 341,665, which show the line-sequential build-up of the characters on the screen. SUPPLEMENTAL OPINION The parties have brought to my attention the fact that my original opinion does not decide Lear Siegler’s claim for a refund of the royalties which it paid after filing its declaratory judgment action challenging the validity of the Cole patent. I write to remedy this deficiency. In Lear v. Adkins, 395 U.S. 653, 89 S.Ct. 1902, 23 L.Ed.2d 610 (1969), the Supreme Court held that the contract principles underlying the doctrine of licensee es-toppel must give way to the conflicting federal interest in encouraging the early testing of arguably invalid patents. Relying on this federal interest, the court further declared that a licensee cannot be required to pay royalties during the time it is challenging the patent’s validity in court. The opinion in the Lear case does not reach the issue which is presented here: when a licensee elects to pay royalties while litigating the validity of the patent, may he, if successful, recover those royalties. I conclude that the answer to this question, in the absence of special circumstances should be “no”. When a licensee continues to pay royalties after filing a declaratory judgment action, it does so because it believes that course is in its best interests. As Professor McCarthy has pointed out, a licensee “hedges” its bet by continuing the payment of royalties and thereby continues to derive benefits from the license even while attacking the patent. McCarthy, “Unmuzzling” the Patent Licensee: Chaos in the Wake of Lear v. Adkins, 59 J.Pat.Off.Soc’y 475, 528-33 (1977). First, the license assures that the licensee will be able to continue its use of the patented invention during the litigation and, if it loses, thereafter. It has neither of these assurances if it chooses to cease paying and terminate the license. Moreover, continued payment assures the licensee that its use pendente lite and thereafter, if he is unsuccessful, will be at the license royalty rate, thereby providing insurance against the possibility of a higher, court determined “reasonable royalty” or a higher negotiated rate in a new license. Finally, continued payment provides insurance against the possibility of an award of attorneys’ fees or treble damages in the event the challenge of the patent is unfruitful. Id. Given the fact that the licensee reaps these benefits from the payment of royalties under the license while litigating, I believe equity is on the side of the patentee when recoupment is sought after a finding of patent invalidity. Moreover, I perceive no inconsistency between a result consistent with this equity and the policy considerations which underlie Lear.* Since I find no special circumstances favoring a recovery of royalties by Lear Siegler, judgment will be entered for RCA on this claim. Footnotes [1] Hazeltine refers to Hazeltine Corp., Lear Sie-gler to Lear Siegler, Inc., and ADDS to Applied Digital Data Systems, Inc. [2] If the binary codes used for identifying the various characters were supplied directly to the CRT, the pattern on the screen would not generally be recognizable. Thus, the 6-bit code 000111 might represent a “7”, but it would appear as three dark spots followed by three bright spots, or vice versa. Consequently, it is necessary to translate the 6-bit binary code into [940] a video signal which will present a normal-appearing character. For example, a number “1” might be translated as shown in HLA Exhibit 805, which is as follows: [3] U.S. Patent No. 2,920,312. [4] U.S. Patent No. 2,987,715. [5] The displays in the Amdahl and Fenimore patents have parameters differing from those of a commercial TV display and I understand RCA to acknowledge these as being displays which “exhibit a television raster scan line pattern.” [6] The use of both sawtooth and staircase (or stepping) waveforms for vertical and horizontal deflection in CRTs was common prior to 1960, as evidenced by the Brown and Gordon patents. PX 3, Tab. 31, p. 4, lines 24-29; PX 3, Tab. 8, col. 5, lines 39-59. It was a matter of designers’ choice. Thus, if Dirks is distinguishable on the basis of its stepping in a vertical direction, Cole was nevertheless obvious from the prior art and invalid under 35 U.S.C. § 103. [7] In particular, the art’s understanding did not confine the concept to synchronous, as contrasted with asynchronous, operation. HLA 410 at p. 163. [8] U.S. Patent No. 3,293,614. [9] U.S. Patent No. 3,241,120. [10] U.S. Patent No. 2,987,715. [11] U.S. Patent No. 2,920,312. [12] U.S. Patent No. 3,017,625. [13] U.K. Patent No. 905,951. [14] Accord: Park-Ohio Industries, Inc. v. Letica Corp., 617 F.2d 450, 453-54 (6th Cir.1980); Medical Laboratory Automation, Inc. v. Labcon, Inc., 670 F.2d 671, 673 (7th Cir.1981); Louis A. Grant, Inc. v. Keibler Industries, Inc., 377 F.Supp. 1069, 1081-82 (N.D.Ind.1973), aff’d, 191 USPQ 424, 541 F.2d 284 (7th Cir.1976); Ropat Corporation v. West Bend Company, 382 F.Supp. 1030, 1035 (N.D.Ill.1974) and Lundy Elec. & Sys., Inc. v. Optical Recognition Sys., Inc., 362 F.Supp. 130 (E.D.Va.1973), aff’d on other grounds, 493 F.2d 1222 (4th Cir.1974). [15] As explained in an earlier footnote, I conclude that Cole, if not anticipated by Dirks, is obvious from Dirks in light of Brown et al. and others. See p. 948, n. 6, supra. [16] While the Cole invention was conceived in early 1960, there is no evidence of a reduction to practice before the October 16, 1963 filing date. Cole’s invention date is therefore deemed to be this filing date. [17] Moreover, I am not persuaded that the 1963 artisan would take Jones’ and Gordon’s expressed preference at face value. The literature of the period, as a whole, suggests that there were some advantages to the monoscope and some advantages to the then state of the art digital character generators. [18] I agree with HLA that the Norden system was, in the vocabulary of the day, a “digital” system when it displayed straight line letters. I am not persuaded, however, that it tells us much about whether it was obvious to combine Jones or Gordon with Evans. Moreover, given the amount of activity in the area, if Cole had been obvious when Evans was published in January of 1962, one would expect people like those at Norden to have conceived of the combination before early 1964. [19] Similar systems with the same multiple character generator limitations were proposed by Raytheon and described in its Dutton patent and by Smith and Heggs. [20] Wagner’s conception in February of 1962 does not convince me of the obviousness of Cole. He appears to have had the same inventive insight as Cole et al. While he testified that he didn’t consider that conception to be exceptional, his contemporaneous conduct indicates otherwise. [21] RCA also tendered evidence that the Japanese patent office, with knowledge of Evans and of Gordon and Jones type character generators, found the Cole invention to be non-obvious. I have not relied upon that evidence in reaching my conclusion, however, because the record does not suggest that the level of skill in the relevant art in Japan in 1963 was the same or similar to that in the United States. [22] The duration of the scan line count signal in the HLA terminals is an integral multiple of the dot position count, as required by Cole Claim 3. At trial, Larky interpreted Claim 3’s requirement that “the duration of each television raster scan line [be] an integral multiple of [the dot period]” as referring to the entire horizontal raster scan cycle, including retrace. The Cole patent teaches that the integral multiple relationship between scan line and dot period need only extend across the displayable portion of the scan line. Cole specifically excepts the retrace time from this integral relationship requirement. However, nothing in Cole prevents the total scan line period from also being an integral multiple of the dot period. In fact, all of the HLA infringing terminals have this integral relationship, thereby infringing Claim 3 even under Larky’s interpretation. [23] See 4 Chisum, Patents, § 19.05. [24] See, e.g., Churma v. United States Steel Corporation, 514 F.2d 589 (3d Cir.1975); Gruca v. United States Steel Corporation, 495 F.2d 1252 (3d Cir.1974). [25] Those RCA employees identified by HLA in the pre-trial order who are now deceased had only administrative contact with the Cole application. [*] The policy considerations in this context are different from those involved when the issue is the recoupment of pre-challenge royalties. As Professor McCarthy points out, permitting the recovery of pre-challenged royalties would defeat the policy’s underlying Lear by encouraging licensees to postpone challenges until the end of the patent term. 59 J.Pat.Off.Soc’y. at 495. While permitting the recoupment of post-challenge royalties would not have the same detrimental effect, the principal that one who receives bargained for benefits should pay the consideration promised is sufficient to sustain a rule denying such recoupment in the absence of a conflict with the values affirmed in Lear.