Rixon Inc. v. Racal-Milgo, Inc.

551 F. Supp. 163 | D. Del. | 1982

551 F. Supp. 163 (1982)

RIXON INC., Plaintiff,
v.
RACAL-MILGO, INC., Defendant.

Civ. A. No. 78-18.

United States District Court, D. Delaware.

July 2, 1982.

Robert K. Payson, Peter M. Sieglaff, Potter, Anderson & Corroon, Wilmington, Del., David W. Plant, Laurence S. Rogers, Norman H. Beamer, Albert E. Fey, Robert C. Morgan, Kevin J. Arquit, Fish & Neave, New York City, for plaintiff.

James L. Holzman, Michael Hanrahan, Prickett, Jones, Elliott, Kristol & Schnee, Wilmington, Del., Harold L. Jackson, Stanley R. Jones, Albin H. Gess, Larry K. Roberts, Jackson, Jones & Price, Tustin, Cal., for defendant.

OPINION

STAPLETON, District Judge:

Rixon, Inc. here seeks a declaratory judgment that certain United States patents held by Racal-Milgo, Inc. ("Milgo") are invalid, unenforceable and uninfringed. Milgo has counterclaimed for relief against alleged infringements by Rixon, Inc. For present purposes our focus is on three patents — No. 3,524,023 ("Whang '023"), issued to Sang Whang; No. 3,643,023 ("Payne '023"), issued to Paul Payne and Robert Ragsdale; and No. 3,590,381 ("Ragsdale '381") issued to Robert Ragsdale. All involve electronic circuits used in devices known as "modems."

A modem, short for modulator/demodulator, is a device used to permit communication between computers at different sites, or between a computer and remote terminals, by converting binary digital computer data^[1] into analog signals which can pass over telephone lines. The modem circuits encode or translate a digital output by modulating the phase, frequency, or amplitude of a carrier pulse. Each digital "word", or sequence of digital bits, modulates the carrier in a different way, so that the demodulator at the receiving end can reconstruct the original word in accordance with a pre-selected convention.^[2]

The goal of modem designers in the mid-1960s was to increase the speed and reliability of data transmission. Computers process information much more rapidly than it was possible to transmit that information over telephone lines.^[3] One obstacle to higher speeds was the variation in delay and amplitude distortion on telephone lines. Not only did phone lines delay different frequencies more than others, and reduce the amplitude of some frequencies more than others, these differences also varied from one line, or combination of lines needed to complete a call, to the next. Some users, in search of high speed reliable data transmission, leased telephone lines and conditioned them so that their frequency response was more uniform and predictable. Another solution was the use of "variable equalizers" which could be adjusted to minimize distortion over any combination of lines. The leased line system was expensive and impractical for computer systems which did not communicate with any single remote location continuously. Variable equalizers were unreliable. So, engineers continued to search for something better.

In the burgeoning market for data communications equipment of the early 70s, Milgo^[4] enjoyed considerable success with a line of modems, including Model 4400/48. Milgo manufactured these modems under patents acquired by assignment from Milgo engineers, including the three involved in this case.^[5] Whang '023 claims an 8 phase modem using a "narrow" passband. The Payne '023 patent involves a detector circuit used to convert the transmitted signal pulses back into digital form. Ragsdale's patent covers an improved detector which is more resistant to noise interference because it embodies what is known as a phase lock loop.

In July 1971, Milgo filed suit against United Telecommunications, Inc. ("United"), and its subsidiary United Business Communications, Inc. ("UBC") in the United States District Court for the District of Kansas. Milgo alleged that UBC had infringed the Whang, Ragsdale and Payne patents by offering for sale the DS 4800 modem, manufactured by UBC's subsidiary, Rixon Electronics, Inc. Milgo did not join Rixon Electronics as a party; subsequently, UBC dissolved Rixon Electronics and transferred its assets to a new entity, Rixon, Inc., formed in a joint venture with Sangamo Electric Company ("Sangamo"). The Court held that Rixon Electronics was UBC's agent and alter ego. Milgo obtained a judgment in the Kansas Court against UBC;^[6] the Court awarded damages and an injunction against future infringement of the Milgo patents. The Tenth Circuit affirmed.^[7]

Rixon, Inc. commenced this action in January 1978. Milgo's position is that the Kansas judgment binds Rixon, Inc. by principles of res judicata and collateral estoppel and precludes it from pursuing its claims in this Court.

The United States District Court for the District of Massachusetts recently filed an Amended Opinion holding that Whang '023 patent invalid and awarding attorneys' fees against Milgo for its inequitable conduct with respect to that patent. Codex Corp. v. Milgo Electronic Corp., 534 F. Supp. 418 (D.Ma.1982). Relying upon principles of non-mutual collateral estoppel,^[8] Rixon has supplemented its complaint to allege that Milgo is precluded from asserting the validity or enforceability of Whang '023 here. A final judgment was entered in the Codex action on June 16, 1982. The parties to this action have not yet been heard on the issues raised by Rixon's supplemental complaint, however, and they are not ripe for resolution.

This Court severed for trial the issues relating to whether Rixon, Inc. is bound by the Kansas judgment and whether Milgo has engaged in inequitable conduct which renders one or more of the patents in suit unenforceable.

I. BACKGROUND.

In 1928, Harry Nyquist published his discovery that an information carrying pulse of duration T (T is also known as the period) required a bandwidth of 1/T Hz^[9] for accurate transmission.^[10] This relationship forms the basic constraint upon the speed of telephonic data transmission. Nyquist also described criteria for a filter which would enable the accurate transmission of data without interference between successive pulses.^[11] The ideal, or "brick wall", Nyquist filter passes only signals within a range of ± ½T Hz around the carrier frequency:^[12]

Because it is not possible to create such a brick wall filter, practical filter design requires the use of "raised cosine" filters:

The portions of the curve which extend beyond the limits of the brick wall are known as the "skirts". The "rolloff" of the filter is defined by the ratio of the width of the skirts to the width of the Nyquist ideal. Thus, a 50% rolloff filter is one in which the sum of the skirts is equal to one-half 1/T.

Another factor which governs the rate of data transmission, along with the range of frequency available, is the modulation scheme. An engineer can convey information by modulating a signal in three basic ways:

One can design a system to encode data using any modulation scheme.

Transmission speeds depend both upon how many impulses can be transmitted over a given bandwidth and how many "bits" of data are communicated in each impulse. The number of available "codes" depends upon the ability of detection circuitry to distinguish one modulation pattern from another in the presence of random "noise" and phase jitter which can be reduced, but not entirely eliminated, from the transmission line.^[13]

As noted, a third constraint on transmission was the ability of the demodulator portion of the system to translate the transmitted analog signals back into a sequence of 1s and 0s. Accurate recovery of data required the model designer to "read" the signal pulse at its most stable point. In some systems, this most reliable point was at the center of the period. By transmitting a constant pulse at a rate equal to 1/T, an engineer could maintain a constant sampling rate, timed to coincide with the center of the interval. To do this, however, the demodulator had to recover an accurate "clock", or timing signal, from the transmitted data to maintain synchrony.

Once the receiver detected the phase of the incoming signal, the task of translation remained. One method, that was employed in the Payne '023 patent, involved the comparison of the phase of the incoming signal at time t1 with the last signal received at time t0. In this scheme, the detector obtains the transmitted data by calculating the difference, or change in phase between the t1 and t0 signals. This is known as differential phase detection. "Coherent" phase detection, used in the Ragsdale '381, "phase locks" the carrier wave to a higher frequency wave which yields a more stable and therefore reliable point of comparison.

Milgo successfully marketed 8 phase modems which operated at 2400 and 4800 bps and labeled those modems with the Whang, Payne and Ragsdale patents. Rixon, Inc. argues that each of those patents are invalid and unenforceable. According to Rixon, Whang '023 is invalid because it is an old combination of elements already established in the prior art, and it is unenforceable because Whang and Milgo concocted, and successfully urged upon the Kansas Court, a theory of validity they knew to be untrue. Payne '023 is invalid, in Rixon's view, because it is obvious from the prior art; it is unenforceable because Milgo took a position in the Kansas litigation with respect to the scope of the Payne patent which was logically irreconcilable with one it was taking simultaneously in a foreign patent proceeding. Rixon attacks Ragsdale '381 as invalid because it is anticipated in the prior art, or obvious from it, and unenforceable because Milgo suppressed evidence that it had made a commercial use or sale of the Ragsdale invention more than one year prior to filing the patent application.

II. THE EFFECT OF THE KANSAS JUDGMENT.

Rixon was not a party to the Kansas action, but "[b]y the usual formulation of the rule, the effects of res judicata also extend to those in privity with the parties to an action." Note, Developments in the Law — Res Judicata, 65 Harv.L.Rev. 818, 855 (1952). See Regal Knitwear Co. v. NLRB, 324 U.S. 9, 14, 65 S. Ct. 478, 481, 89 L. Ed. 661 (1945); Golden State Bottling Co. v. NLRB, 414 U.S. 168, 179, 94 S. Ct. 414, 422, 38 L. Ed. 2d 388 (1973); Montana v. United States, 440 U.S. 147, 153, 99 S. Ct. 970, 973, 59 L. Ed. 2d 210 (1979); Restatement of Judgments § 83 (1942). Unfortunately, "privity" itself is an ambiguous and protean term; binding a non-party by finding it in privity with a party to a prior action amounts to little more than proof of assertion. As Judge Goodrich wrote, concurring in Bruszewski v. United States, 181 F.2d 419, 423 (3d Cir.), cert. denied, 340 U.S. 865, 71 S. Ct. 87, 95 L. Ed. 632 (1950):

Privity states no reason for including or excluding one from the estoppel of a judgment. It is merely a word used to say that the relationship between one who is a party on the record and another is close enough to include that other within the res judicata.

See also Nash County Bd. of Ed. v. Biltmore Co., 640 F.2d 484, 494 (4th Cir.1981); Developments in the Law—Res Judicata, supra, 65 Harv.L.Rev. at 856; Vulcan, Inc. v. Fordees Corp., 658 F.2d 1106 (6th Cir. 1981); Aerojet General Corp. v. Askew, 511 F.2d 710 (5th Cir.1975), cert. denied, 423 U.S. 908, 96 S. Ct. 210, 46 L. Ed. 2d 137 (1976). What we must proceed to determine then is whether the relationship between UBC and Rixon, Inc. is sufficiently close that the preclusive effects of the Kansas judgment against the former should apply to the latter as well.

To bind Rixon, Inc. (hereinafter sometimes "Rixon III") to the Kansas judgment, Milgo needs a chain with at least two links. No one contends that Rixon III was in privity with UBC. Rather, Milgo relies on the "successor in interest" doctrine to link Rixon III to Rixon II. As for Rixon II, Milgo premises its res judicata theory on two alternative arguments. Milgo contends that Rixon II is the successor in interest to UBC's modem marketing activities, and is therefore bound by the judgment in Kansas. Milgo also reaches the same result by characterizing Rixon II as UBC's "alter ego".

A. Rixon's Corporate History.

At the time that Milgo first began to manufacture and sell modems embodying the systems claimed in the Whang, Payne and Ragsdale patents, Rixon Electronics, Inc. ("Rixon I") was an independent corporation organized under the laws of Maryland, which sold modems and other electronic equipment to commercial purchasers. Rixon I sold "Sebit" amplitude modulated modems, and had begun research and development work on a phase modulated model, the PM-48.

As part of a far-reaching plan to compete with AT & T by marketing its own electronic devices for interconnection with the telephone network, United Utilities agreed to purchase Rixon I in October 1968. On July 3, 1969, Rixon I transferred its manufacturing assets to a wholly-owned subsidiary of United, New Rixelco, Inc., later renamed Rixon Electronics, Inc. ("Rixon II"). United envisioned Rixon as a manufacturing supplier for a marketing entity, tentatively named United Data Products, which would distribute a complete line of business communications products, purchased from United subsidiaries such as Rixon and from non-affiliated companies.

Although retaining Rixon's independent management, from the beginning United laid down policy guidelines for Rixon II management. Darwin Deaver instructed Maurice Horrell, the President of Rixon, to inform him of any decisions within certain enumerated categories before implementing them. Deaver also told the Rixon management that the Rixon Board of Directors would resolve any policy conflicts between Deaver and management, unless the issue was one involving basic United policy.

In January 1970, United formed a new subsidiary, United Business Communications, Inc. UBC's Robert Liepold assumed responsibility for coordinating Rixon II's operations with those of the parent, and UBC's marketing department head, William Lombardi, began to play an increasing role in the supervision of Rixon's marketing. In preparation for the planned transfer of Rixon's marketing organization to UBC, Lombardi assumed control over Rixon marketing efforts in April 1970. United transferred its Rixon stock to UBC on June 24, 1970. The formal transfer of Rixon's marketing operations to UBC followed on July 1, 1970.

As Rixon's primary commercial customer, UBC communicated customer specifications to Rixon II. Rixon and UBC also coordinated research and development activities through technical policy committees which included representatives of both firms. The original members of the UBC board of directors — Deaver, Liepold, and James Gleason — served on the Rixon II board as well. The transfer of July 1, 1970 occasioned no change in Rixon II's managerial personnel, however.

Following UBC's acquisition of Rixon II's stock, Rixon II and UBC were run as separate corporate entities. No officer of UBC served as an officer of Rixon II and no director of UBC, other than Rixon's Horrell, ever served as an officer of Rixon II. Each had its own employees whom it hired and fired and compensated from a separate payroll. Each maintained its own books and bank accounts, and prepared its own budget. The two corporations occupied different premises and when one performed services for the other compensation was paid. UBC did not provide financing for Rixon. Sales between the two were at negotiated transfer prices designed to yield to Rixon II a 20% pre-tax return on investment.^[14] Responsibility for the day-to-day management of Rixon's affairs rested with its officers, not those of UBC.

United ultimately became dissatisfied with Rixon's performance and approved Darwin Deaver's proposal that it sell the company. It reached this decision in August of 1971, about a month after Milgo filed suit in Kansas. Believing that Rixon would be a more attractive purchase if it had its own marketing capability, UBC restored Rixon's marketing staff and function. Thereafter, Rixon continued these marketing operations, filling its orders in part by repurchasing modems which it had previously sold to UBC and which UBC held in its inventory when the marketing business was retransferred. Rixon operated for a time as "United Business Communications — Rixon Division" and fulfilled outstanding UBC sales contracts.

Sangamo Electric Company agreed to enter a joint venture under the name Rixon, Inc. on September 29, 1972. United contributed the assets of Rixon II, and received a 40% ownership interest; Sangamo contributed its own modem manufacturing operation, for a 60% share. United also agreed to indemnify the joint venture in the event that it could not market Rixon's DS 4800 modem, or a substitute, in the event that Milgo obtained an injunction in the pending Kansas case. The joint venture, Rixon III, continued the business operations of Rixon II without interruption.

B. Rixon II And Rixon III.

Rixon III succeeded to Rixon II's entire modem business: marketing, designing, and manufacturing. It did so after the commencement of the Milgo patent litigation and with actual knowledge of the pendency of that case. It is a well settled principle of law that a party, after the commencement of an action, may not circumvent, defeat or evade the judgment by transferring its interest to a non-party. Restatement (2d) of Judgments § 44. If that were not so, a party could always avoid the consequences of a judicial defeat by transferring its interests to someone not bound by the earlier judgment, putting an aggrieved party to the expense of pursuing its rights as the interests to which they attach pass from hand to hand. Under this well settled doctrine, Rixon III is bound by the Kansas judgment to the extent that its transferor, Rixon II, is so bound. Brunswick Corp. v. Chrysler Corp., 408 F.2d 335 (7th Cir.1969).

C. Rixon II And UBC.

1. The alter ego theory.

Corporations are presumptively distinct juridical entities. Indeed, the primary function of the corporate form is to define a legal "person" with rights and interests which are separate from those of its owners. It follows that the measure of control an owner must exert over the operations of a subsidiary to make it his or her alter ego must be substantially greater than that which inherently results from stock ownership. If this were not so, any corporate subsidiary would become, by definition, the alter ego of its parent.

Control over the composition of the board of directors gives the parent the authority to make broad policy decisions, to approve major planned expenditures, and to consult with management. UBC exercised each of these perogatives. UBC chose Rixon's board of directors and placed some of its own directors on that board. But these meetings were not ritualistic, "rubber stamp" endorsements of UBC sponsored proposals. Through Liepold, UBC insisted on being informed of decisions likely to have a major effect on Rixon's profitability before implementation, but there is nothing in the record from which I could infer that this amounted to anything more than a process of "advice and consent." Neither Liepold's conduct, nor the Deaver letter which outlined the relationship between Rixon and United before the creation of UBC, indicates that UBC managed Rixon's affairs.

I do not agree with Milgo that the consultations between Rixon and UBC through the medium of the technical policy committees involved UBC in the day-to-day engineering and manufacturing operations at Rixon. UBC delivered information about customer design preferences to Rixon, as it did to other, unaffiliated suppliers, but did not deliver instructions to Rixon engineers. Nor am I persuaded that UBC's designation of Rixon II as "UBC — Rixon Division" after the transfer of the marketing operation back to Rixon amounts to an admission that the two should not be treated as separate corporate entities. Such a designation was utilized during a transition period in order to avoid customer confusion. This is, of course, some evidence tending to show that those involved regarded UBC and Rixon II as a single enterprise. When considered in light of its purpose and the other evidence regarding the manner in which Rixon II was run, however, it does not establish an "alter ego" relationship.

The evidence does support Milgo's view that Rixon's affairs were conducted with an eye to their potential impact on United's interests as a whole and that the terms of transactions between Rixon II and UBC were not always the same as those which would have been produced by bargaining between two independent firms. I do not view this as controlling, however, in a case where the parent and subsidiary have entirely separate operations and the parent has refrained from exercising control over the subsidiary's day-to-day management. Japan Petroleum Co. (Nigeria, Ltd.) v. Ashland Oil Co., 456 F. Supp. 831 (D.Del.1978).

2. Successor in interest.

As noted above, the successor in interest doctrine is well established. A judgment against a transferor of property which affects its rights in that property affects the transferee in the same manner, so long as the transfer takes place after the commencement of the action and the transferee has notice that the action is pending. Restatement (2d) Judgments § 44. Rixon III disputes neither the validity of this proposition nor its general application to patent infringement litigation. It points out, however, that UBC did nothing other than market the accused modem during the time the Kansas litigation was in progress and argues that the transfer of its marketing activities to Rixon II on January 1, 1972 should not carry with it any obligations arising from that litigation. The suggested distinction between a marketer and a manufacturer or a manufacturer-seller is without substance, however. UBC was in the business of selling modems. Milgo asserted in the Kansas Court that these sales infringed the Whang '023 patent. On January 1, 1972, UBC effectively transferred its entire modem marketing business to Rixon II. Rixon II accepted that business with knowledge of the pending action and continued to operate it. Under these circumstances Rixon II is bound by the ultimate determination of the Kansas Court with respect to all claims pending at the time of the transfer.

I agree with Rixon III, however, that it cannot be held bound, as a successor in interest, with respect to any claim which had not been asserted in the Kansas action when the transfer of the marketing business occurred. Since Milgo did not amend its Kansas complaint to assert claims based on the Payne or Ragsdale patents until August 11, 1972. Rixon II, and accordingly Rixon III, stands in UBC's shoes for res judicata and collateral estoppel purposes only with respect to the Kansas adjudication of Milgo's Whang '023 claims.

III. UNENFORCEABILITY.

One accused of patent infringement may defend by challenging the enforceability of the patent. "Unclean hands" in the procurement of the patent from the Patent and Trademark Office or in prior enforcement litigation, for example, may render the patent unenforceable. See, e.g., Precision Instrument Mfg. v. Automotive Main. Mach. Co., 324 U.S. 806, 65 S. Ct. 993, 89 L. Ed. 1381 (1945); Keystone Driller v. General Excavator, 290 U.S. 240, 54 S. Ct. 146, 78 L. Ed. 293 (1933); 4 Chisum, Patents, § 19.03. To establish an unenforceability defense, however, the allegedly infringing party must prove, by clear and convincing evidence, that the patent holder has wrongfully obtained or misused its patent.

A. Whang '023.

The story of Whang '023 is the centerpiece of the Milgo modem litigation. In Kansas, Milgo touted Whang '023 as a breakthrough on a par with the invention of the transistor, and persuaded the Kansas Court that the Whang invention was responsible for the commercial success of the Milgo modem line during the 1970s. In persuading the Kansas Court that Whang had invented something patentable rather than arrived at an obvious improvement over the prior art, Milgo relied on representations of Whang and other witnesses that the Whang '023 patent taught the use of "sharp skirted" filters. As explained hereafter, Whang, Milgo and its Kansas trial attorney knew at that time that the Whang '023 did not teach "sharp skirted" filters and that the Milgo modems touted as having revolutionized the industry did not have sharp skirts. In its successive effort to sell this "sharp skirt" theory, Milgo deliberately concealed documents and information from its Kansas opponents and the Kansas Court which were needed to fairly assess the patent against the statutory criteria of anticipation and obviousness.

1. The development of the Whang modem.

Whang developed the modem embodying the claims of the '023 patent in response to a "challenge" to Milgo made by Western Union. After rejecting a modem originally designed for aerospace communications, Western Union told Milgo it would consider a modem that could transmit 2400 bps in a useable bandwidth of 1000 Hz over Western Union's BEX lines which were comparable to the commercial telephone network. Whang was a filter designer by trade and training, with no previous experience engineering data modems. There is much in the record which suggests that inexperience was an advantage, not a handicap.

Whang knew that the usual approach to increasing the speed of telephonic data transmission was to use delay and amplitude compensating networks to equalize the phase and amplitude characteristics across the entire transmitted frequency spectrum. Modem designers preferred to use a wide frequency spectrum, because the wider the bandwidth, the more stable and free from error the modem tended to be. Modem designers also regarded two level or four level systems as preferable to those with more levels which could transmit more bits per baud, but required greater accuracy in the detection circuitry.

Whang's conceptual starting point was the 1000 Hz limitation on bandwidth. Rather than trying to figure out how best to improve the quality of the line so that he could develop a wide band system, Whang experimented and was able to determine that all telephone lines had much the same characteristics within a range centered around 1700 Hz or so. His experimental results were consistent with the profile of telephone line characteristics in the Alexander, Gryb and Nast article published in the Bell System Technical Journal in 1960. Because within this range of about 1000 Hz the characteristics of lines on the switched telephone network did not vary significantly, Whang concluded that he could do away with the expense and inconvenience of variable equalizers. In that sense, the Western Union specification proved advantageous.

But the 1000 Hz limitations posed troublesome problems. Whang was not trained in information theory, and was apparently not familiar with Nyquist's paper on the minimum bandwidth needed to transmit information at a rate 1/T. His own experimental work, however, confirmed the Nyquist principle. Whang realized that — with 1000 Hz to work in — he could transmit information at a theoretical maximum of 1000 baud per second. If each baud contained a single bit, as with a two phase system, or even two bits, with four phase modulation, the modem would fall short of Western Union's demands. This impelled Whang to try an 8 phase system. 8 phases, or 3 bits per baud, gave Whang a theoretical capacity of 3000 bps.

Patient experimentation and testing showed Whang that he could transmit 8 phase shifts with sufficient accuracy, although his error rate was higher than usually experienced with two or four phase systems. Whang also determined that he could transmit enough energy in the signal "envelope" to maintain an accurate clock at the receiving end. With an accurate clock, Whang found that he could reliably sample shifts in phase at the center of each pulse interval, and reduce the likelihood of an error.

The system Whang devised met the Western Union criteria. It could transmit 2400 bps over a nominal bandwidth of 800 Hz, and do so with sufficient reliability to be commercially practicable. Whang achieved this result by combining and altering techniques to fit externally imposed constraints — the paradigmatic role of engineering. Whang's system combined (1) 8 differential phase modulation and hence 3 bit bauds; (2) center sampling to determine phase shifts at the point of greatest phase integrity; and (3) a reliable clock derived from the signal envelope. The careful balance among the elements of the system produced a practical and commercially successful modem. That does not, however, mean that it was a patentable invention.

Each of the elements listed above, along with the additional observation that the minimum bandwidth required for transmission at a rate 1/T was T Hz, was known in the prior art. Whang skillfully combined these elements to produce a superior practical system. But the record shows that the subsequent confirmation of a monopoly in Kansas rested on misinformation provided to the Court by Milgo.

2. The patent office prosecution.

In his application and the patent which eventually issued in his name, Whang explained:

[T]he basic philosophy of the prior art and current development is in upgrading each telephone line individually through automatic or manually operated equalization devices each time data is to be transmitted. The basic philosophy of the present invention lies in the discovery that if the signals are appropriately band limited, as described hereinafter, all the lines appear the same within that bandwidth. Most telephone lines do not have the desired characteristic, but since they are approximately the same within a selected narrow bandwidth as hereinafter described, a fixed equalizer will make them suitable and essentially the same for the purpose of transmitting digital data at a high bit rate, and almost all lines will behave as a linear phase bandpass filter with a known response irrespective of variations between lines. In one respect, the invention may be though of as adapting the signal format to accommodate an unequalized line, rather than custom adapting a line to the signal, by narrowing the signal spectrum appropriately to provide that within the bandwidth all lines appear similar.^[15]

As promised, Whang proceeded to describe the way to achieve the "requisite narrow bandpass characteristic." "It is preferred," he wrote, "that filters 2 and 8 both be used to form a composite network having a 1/T bandpass under 1000 Hz." Later on in the application, Whang explains that 1/T Hz is the minimum bandwidth required to transmit the information contained in the original signal. He illustrates this with three figures. Figure 3 is the pulse shape transmitted through a filter of width 1/T Hz. This pulse preserves all of the phase information of the original signal. Figure 4 illustrates the consequences of a band which is too narrow, while Figure 5 depicts an overly wide composite filter. Whang summarized:

If the line characteristic is good within the bandwidth of 1/T Hz and poor outside this range, it is better to pass only 1/T Hz bandwidth and get an output shape like Fig. 3 than to pass a wider bandwidth and lose critical features of the response.

In Whang's specification 1/T Hz was equal to 800 Hz, but his patent taught that the 1/T band could extend up to 1000 Hz, assuming that the good part of the telephone line was that wide. The 800 Hz bandwidth allowed him to meet Western Union's specifications without testing the tolerance of his system at the edges of the high quality portion of the line.

Whang's "discovery" was the usefulness of adapting the modem to the telephone line rather than vice versa. He implemented this discovery with a confirmation of the Nyquist theory of the "necessary and sufficient" bandwidth to preserve phase separation. The Whang invention, as set forth in the application and the final patent, did not teach anything about the rolloff of the composite filter, by rolloff or any other name. If one gives "passband" its usual and accepted meaning, that is, the width of the Nyquist 1/T Hz band, Whang's patent is entirely self-consistent and unambiguous.

Thus, in Claim 1 of the patent, Whang claims, "filter means connected in the signal transmission link and characterized as passing signals in the frequency range defined as 500 Hz on either side of a center frequency, fo, selected from 1600 Hz through 1800 Hz, the filter means characterized as having a passband width of 1/T Hz and having a center frequency of fo." The 500 Hz from either 1600 or 1800 Hz defines the range of frequencies within the reliable part of the telephone line. This limitation applies not only to the 2400 bps embodiment, but to the entire Whang invention. It goes on to specify a "passband" of 1/T Hz, where T is the modulation period. In short, Whang applies Nyquist principles to the knowledge of the phone lines acquired from Alexander, Gryb and Nast.

The patent examiner originally rejected Whang's application as unpatentable over a combination of patents relating to the Bell 201B modem. The 201B could transmit 2000 bps over switched telephone lines, and used four phase modulation with center sampling and differential phase detection at the receiver. Whang had access to the Bell System Technical Reference describing the 201B while he was doing his modem design. The examiner concluded that an article by Irland, "A High Speed Data Signalling System," taught the use of a fixed equalizer over a filtered bandwidth of 800 Hz centered at 1600 Hz. He also noted that a patent to Baker describes a system for phase modulating a 1750 Hz carrier and recovering the signal at the receiver. (U.S. Pat. No. 3,128,343).

Whang's attorney submitted an amendment in which he attempted to distinguish the Baker and Irland references from the subject of the invention. The amendment distinguished Irland by stating that the 800 Hz limitation of the band taught in the Irland article "is dictated by his low data rate of 1600 bps which is halved by group[ing] bits in pairs." But this was nothing more than a restatement of the Nyquist principle which governs data transmission modems, including Whang's. Whang's 800 Hz bandwidth was dictated by his desire to achieve 2400 bps transmission within less than 1000 Hz and his decision to implement an 8 phase modulation scheme. Additionally, the amendment notes, "The phase equalizer discussed in Ireland [sic] is solely to compensate for the delay introduced by his filter and does not concern itself with also compensating an `average' amplitude and delay amount to make all phone lines look alike."

Rixon challenges the propriety of these comments attempting to distinguish the prior art and one can certainly argue about the substance of the distinctions. But some room must be left for advocacy. The patent examiner had the Irland reference and technical expertise to evaluate it. Unlike a case in which the patentee fails to disclose a known and pertinent reference, or presents experimental results in a misleading way, a case in which the patentee disputes the examiner's reading of the prior art does not prevent the examiner from applying his or her expertise to the patent. To the contrary, it calls forth precisely the kind of analysis and comparison of prior art and patent application which the law requires.

Whang's attorney also distinguished Whang's modem from the Baker system, as elaborated in a separate patent to Logan (U.S. Pat. No. 3,020,475), which indicates that the four phase modem with a Nyquist passband of 1000 Hz the modem uses a bandwidth of 2000 Hz. He went on to explain:

Baker uses convention[al] bandwidths in which his modulation period must contain the phase information for the entire duration of the modulation period. In decoding, Baker integrates the signal over the entire interval. See column 4, lines 23 through 33. In applicant's invention the narrow band limiting affords phase integrity only for the middle portion of the modulation period. Thus the two approaches are entirely different.

This language strongly suggests that Whang's use of center sampling, as opposed to full interval decoding, was a novel advance over the art described in the Baker patent. In his Kansas testimony (Tr. 955) Whang stated that he had possession of the Bell System Technical Reference when he was designing his modem. The Technical Reference disclosed an aspect of the Western Electric modem which Baker did not: it used center sampling.

Whang and his attorney had a personal interview with the examiner. In this interview "an understanding was reached that applicant's modem concept was patentable over the Baker, Logan or Ireland [sic] references alone or in combination. In order more adequately to distinguish applicant's novel modem structure over the prior art, it was agreed to point out in the claim language the various problems overcome in a modem system embodying applicant's narrow band-limiting feature." The main problem resulting from the use of a narrowband, as Whang indicated in the original patent application, was that the signal exhibited phase integrity only over the middle part of the band. Consequently, center sampling was necessary, and one needed a reliable clock to maintain the sampling rate. The amendment which followed this understanding added center sampling to the patent claims. It is fair to conclude that Whang and his attorney persuaded the examiner that it was the use of center sampling, in conjunction with the narrow passband, which distinguished the Milgo modem from the prior art.

If Rixon's burden of proof were a preponderance of the evidence, I would be willing to conclude that Whang and Milgo committed fraud on the Patent Office in connection with their description of the prior art. I do not, however, find clear and convincing evidence of scienter. Whang admitted that he had the Bell Technical Reference, but his attitude towards reference materials appears to have been somewhat casual. His emphasis appears to have been on results in the laboratory rather than current developments in the art. The record as a whole does not rule out the possibility he did not carefully consult the Technical Reference, and that he was not aware that it disclosed center sampling.

3. The Kansas litigation.

While I conclude that the evidence of fraud on the Patent Office is not clear and convincing, there can be no doubt that the evidence of fraud on the Kansas Court is accurately so categorized. Whang testified in Kansas that his patent taught the use of filters having 50% rolloff or less. The Kansas Court accepted this statement, and relied upon it as a non-obvious advance over the prior art. (Finding of Fact ("F.F.") 61, 76). This testimony was false, and Whang and Milgo's trial counsel knew it was false at the time. Whang also testified that the Milgo 24LSI and 20LSI modems, although using more advanced circuitry than the original 4400 modem series, incorporated his sharp rolloff invention. The commercial success of these modems was one of the secondary considerations contributing to the Kansas Court's finding of non-obviousness. (F.F. 39). This testimony likewise was deliberately false.

The text of the patent itself is irreconcilable with Milgo's "sharp skirt" theory. Whang, as noted before, was concerned to accomplish one thing: to accommodate his data transmission to the good part of a telephone line. To do this, in turn, he found he needed to sample received signals at the center of the modulation period, to use 8 phases, to recover timing from the carrier envelope, and to filter the signal so that signals passing through the bad part of the line would not interfere with the transmission of the message. He believed that the good part of the line was the 1000 Hz range between 1200 and 2200 Hz. So, for example, Whang included in Claim 39 the requirement that the composite filter "pass signals only between ± 500 Hz on either side of the center frequency." One could interpret this Claim as instructing a modem engineer to use a "brick wall" filter 1000 Hz wide, but Whang knew this to be impossible. Alternatively, one could perhaps understand Whang to be claiming a filter with 25% rolloff, attenuating the signal to a virtual zero level at the plus or minus 500 Hz points. In the context of the patent as a whole, however, this interpretation is untenable. Whang needed to reduce the amplitude of the signals outside the 1000 Hz portion of the telephone line sufficiently so as to prevent interference with data transmission, but he did not disclose the degree of filtering required to accomplish that purpose.^[16] Whang was not describing a composite filter system; he was telling modem engineers that they could count on only 1000 Hz of reliable bandwidth by using his fixed equalizer.

Furthermore, Whang wrote, "By band limiting the signal and rejecting all such unnecessary components and passing only the 800 Hz bandwidth from 1300 Hz to 2100 Hz the signal shape shown in Fig. 3 will be produced." Figure 3 depicts a received signal which has passed through a 100% rolloff filter. Milgo belittles this evidence as a "sketch," but Whang plainly took care to draw Figure 3, and to distinguish it from Figures 4-7. While Whang was not necessarily endorsing the use of a 100% rolloff filter, it seems evident that he, like Bennett and Davey, recognized that any filter which obeyed Nyquist's criteria for avoiding inter-symbol interference would work in his modem.

Since the Kansas trial, Milgo has had two separate opportunities to provide support for its "narrow skirts" reading of the Whang patent. It has failed on both occasions to articulate a coherent story. In the Codex case in Boston, Whang again testified that his patent taught "severe band limiting" and a composite filter with a rolloff less than 50%, although he was unable to point to any language in the patent teaching 50% or less rolloff. The Massachusetts District Court rejected Whang's testimony, citing the absence in the patent of any reference, direct or indirect, to a 50% rolloff, the 100% rolloff of the pulse in Figure 3 of the patent, and the filters actually used in the original Whang modems.

Whang did not testify at trial in this Court. Milgo's technical expert, Dr. Lytle, had not reviewed Whang's previous testimony and was therefore unable to corroborate or be cross-examined on his interpretation of the patent. I cannot accept Dr. Lytle's attempt to identify narrow rolloff teachings in Whang `023. He testified that "passband" means different things in different parts of the patent, that Figure 3 is not — as it purports to be — a representation of the pulse after passing through the composite filter, and was unable to identify a single element in the patent which unambiguously points toward narrow rolloff. In sum, I cannot accept the proposition that Whang '023 communicates narrow rolloff to persons skilled in the relevant engineering art.

The evidence of Whang's own research and development also strongly supports the conclusion that narrow rolloff was not an element of Whang '023. In Kansas, UBC's expert, Dr. Walter Beam, noted that he had not had an opportunity to determine the bandwidth of the modem Milgo built for Western Union which purported to incorporate the Whang invention. If he had done so, UBC would have learned that the original Whang modem, the one built specifically to meet Western Union's "challenge," did not incorporate a composite filter having less than 50% rolloff. This information would also have been disclosed if Milgo had responded in good faith to UBC's discovery. Milgo did not disclose Whang's filter files to UBC during the Kansas litigation, even though those documents fell within categories of documents which UBC sought by discovery.

Whang originally demonstrated his modem to Western Union on June 27, 1966. He held a second demonstration on July 13, the day before he filed his patent application. William Waggener testified at the Delaware trial that the filters which Whang identified as those used in the June and July tests had rolloffs of over more than 70%. Milgo has contested the accuracy of Waggener's method of estimating rolloff, but it has not put forward any alternative estimate of the rolloff of these filters. I find the evidence clear and convincing that the June and July demonstration modems used filters which did not have less than 50% rolloff.

The evidence also shows that Whang subsequently designed narrower filters in an attempt to produce a modem capable of simultaneously transmitting voice and data.^[17] He employed these in the production model 2247 modem delivered to Western Union in March 1967, but this modem did not work successfully and Milgo returned to wider filters for the 175 modems of the 2247 model it sold to Western Union.

I do not accept Milgo's explanation that Whang simply "forgot" what filters were used in the Western Union modem. The design of those filters was contemporaneous with the patent application; Whang thought the patent covered the Western Union modems. A filter designer, by training, Whang was intimately involved in the design and testing of the filters for the Western Union modems. Whang himself said in Kansas that he remembered the filters he designed because they were like his own children. (Tr. 297-98). He could not have "forgotten."

Milgo had ample means available to it to correct the record, or to refresh Whang's ostensibly failing memory. It had both Whang's filter files and internal documents which reflected the decision to widen the Western Union filters. (PX 185). Whang misrepresented the facts about the Western Union modems, and he did so with Milgo's blessing.

Finally, if more were needed, Milgo's corporate conduct shows that narrow rolloff filters were no part of the Whang patent. Milgo marked its 24 and 20 LSI ("large scale integrated [circuit]") modems with the Whang patent designation. But these modems, as Milgo was forced to admit in a letter to the Tenth Circuit Court of Appeals, used filters with wide rolloff.

Despite this corporate recognition of the scope of the Whang '023, when the time subsequently arrived for litigating the validity of the patent in Kansas, Whang specifically testified that these modems used "fifty percent rolloff or less." (Tr. 2326). He did so in response to a specific question posed by Milgo's counsel. Milgo let the answer stand because it helped to establish a "breakthrough" in modem design built on narrow rolloff filters. And Milgo did more than simply let Whang stand uncorrected. Its Proposed Finding of Fact 37, which the Kansas Court adopted verbatim, stated:

37. Subsequent introduction of additional modem series by Milgo (Models 4600, 3300, 2200, 24/LSI and 20/LSI) as well as the 4400 series embody the claimed invention of the patents in suit. (citations to record omitted). The patented Milgo modems are today sold on a worldwide basis. In less than a decade from its entry into the modem market, Milgo/ICC has emerged as one of the acknowledged leaders of the independent modem manufacturers. The growth of Milgo is attributable to the patented modems....

This finding of commercial success is of the sort endorsed in Graham v. John Deere, Co., 383 U.S. 1, 86 S. Ct. 684, 15 L. Ed. 2d 545 (1966) as secondary evidence of a non-obvious invention. As such, it was material to the Kansas Court's application of the statutory criteria, and Milgo's misrepresentations made it impossible for the Kansas Court to apply those criteria correctly to the issue before it.

B. Payne '023.

Whang's original modem design included means for directing shifts in phase from one modulation period to the next. This prototype detector used a circuit known as an analog delay line to store the phase of a sample at time 1 with the next phase sample taken at time 2. Basically, a delay line will — as its name implies — delay a transmitted signal by a certain amount of time. The distance the signal must travel over the line determines the length of the delay. The two outputs from the delay line permit comparison of the phases of successive modulation intervals. The detector then translates the phase shift into one of the 8 three bit words associated with it.

The Payne '023 detector improved on the original Whang modem by substituting digital circuitry for the analog delay line. Although on the subatomic level all electrical phenomena are analog, digital circuits rely upon communications conventions to produce — in many uses — more reliable results. Electrical impulses are not instantaneous, thus any change in voltage will pass through some "grey zone" before reaching the desired level. Digital systems ignore the electrical values in the grey zones, and react only to impulses which the system, by convention, deems "on" or "off," 1 or 0.

The system in the Payne '023 patent uses a digital storage register to hold the phase reading for time. It samples the phase at time 2, and then compares the two phases in a device known as a parallel adder. From this comparison the detector derives the phase shift, which it then converts into the three bit digital word, as in the Whang design. Robert Ragsdale, a co-patentee on the Payne '023 and a Milgo technical expert at the trial in this Court, agreed that it was the digital storage register and parallel adder circuitry which distinguished Payne '023 from the prior art.

In Kansas, Milgo charged that the Rixon DS-2400 modem infringed the Payne '023 patent. Unlike the continuously running counter in the Payne '023, the DS-2400 used a counter which reset once each modulation period. The result was that the DS-2400 determined the phase shift directly from the output of the counter, rather than by means of storage registers and a parallel adder. The Kansas Court found that the DS-2400 and Payne '023 counters were equivalents under the patent law, and held UBC liable for infringement of the Payne '023. The Court noted, "Plaintiff does not contest the defendant's position that the DS-2400 does not include a storage register and comparing means which is physically separated from the counting means, but rather relies on the doctrine of equivalents...." (F.F. 89). "Ragsdale showed that the counter of the DS-2400 when stopped acted as a register and that the reset and timing operation for the counter of the DS-2400 performed substantially the same function, in substantially the same manner, to obtain the same end results as the claims of Payne '023." (F.F. 92). Last, the Court stated that "[t]he defendants failed to prove anything in the file wrapper of the Payne '023 patent prevented the plaintiff from reading the claims in issue on the DS-2400 data set." (F.F. 94).

While the Kansas defendants may have failed to make a case for file wrapper estoppel based on representations made in the United States Patent Office, there was considerable ammunition for an estoppel argument based upon the file history of the German equivalent patent. Siemons cited a United States patent to Secretan, No. 3,020,485, as prior art against the Payne device. The Secretan patent covered a detector in a much slower modem which computed the signal phase by integrating over an entire modulation interval, rather than by sampling at the center of the baud. Despite these differences, however, it used digital circuits to perform much the same function as the Payne '023.

Milgo first learned about Secretan in 1973. Siemons cited Secretan in the German Patent Office in February 1975, six months before the Kansas liability trial. In Kansas, Ragsdale told the Court that the Payne '023 "was the first original digital detector." (Tr. 1085). Ragsdale had seen Secretan and discussed it with Milgo's counsel before the trial, and must therefore have known that Secretan's digital system predated Payne's. At trial in this Court, Ragsdale explained that he meant to say only that Payne was the first digital detector used in the Milgo 4400 modems. But in its proposed Finding of Fact No. 34, Milgo told the Kansas Court that Payne was the first detector using digital circuitry.

Shortly after trial, and before Milgo submitted its post trial proposed Findings of Fact, Milgo distinguished the Payne detector from Secretan in the German patent prosecution by pointing out that Secretan used a reset counter, rather than the register and adder circuits of Payne. This admission that the Payne detector did not read on "reset" devices such as Secretan, if it had been disclosed to the Kansas Court, might well have convinced that Court that the DS-2400 reset counter was not within the range of equivalents to the Payne '023 system.

Milgo argues that it did disclose Secretan to counsel for UBC. The argument is disingenuous at best. The Secretan patent was among those cited on a list of 116 patents which Milgo's counsel sent to UBC's attorney before trial. In response to this submission, UBC asked Milgo to specify which of the 116 patents its trial lawyers had reviewed. Milgo responded with a list of ten patents, which did not include Secretan. As a result, UBC's trial counsel never focused on the Secretan patent during the Kansas litigation.

Moreover, Milgo's disclosure argument misses the point. The problem here is far more serious than a failure to provide opposing counsel with a pertinent prior art reference. If only that were involved, Rixon's unenforceability argument would clearly be insufficient. When Milgo was seeking to obtain monopoly rights, it suited its interests to define its invention as excluding DS-2400 type devices, and it did so. Almost simultaneously, however, in attempting to prove infringement in a court, Milgo adopted exactly the opposite position. Courts have repeatedly held in a file wrapper estoppel context, as well as in non-patent contexts, that one may not be permitted to play "fast and loose" with the courts in this manner. Scarano v. Central R. Co. New Jersey, 203 F.2d 510, 513 (3d Cir.1953). Having taken the position it did in the German patent office proceeding, Milgo, at least in the absence of disclosure of that fact, was precluded from urging upon the Kansas Court an inconsistent infringement theory.

Milgo thus comes to this Court with "unclean hands." It does not follow, however, that the Payne '023 should be held to be unenforceable. Unclean hands is an equitable doctrine and the remedy for inequitable conduct should be tailored to the offense. Traditionally, in analogous file wrapper estoppel cases, the remedy is not a forfeiture of the patent holder's monopoly. Under the approach taken in that line of cases, Milgo should simply be foreclosed from asserting infringement on a theory which reads the Payne '023 to include DS-2400 type devices.^[18]

C. Ragsdale '381.

Even after the development of the Whang modem, and the improvement of its detection circuits, Milgo engineers continued to experiment. Whang met with a group of Milgo designers, including Robert Ragsdale, and reported to them on an article by the Russian Telecommunications Administration which explained how to phase lock a stable high frequency oscillator to an incoming carrier signal to improve the quality of differential signal detection. Following the discussion with Whang, Ragsdale set out to develop a "coherent" detector for the Milgo modem.

The starting point for Ragsdale's work was the digital Payne detector. Payne converted the incoming pulses to a "squared and shifted" intermediate frequency ("IF") square wave, and then used a counter to determine how long it took to reach a zero crossing of the IF signal. The output of this counter fed into a storage register, which held the count for comparison with the count of the following pulse in a parallel adder. This digital process yielded a seven digital bit sequence which encoded the phase change between the first sample and the second. Errors in calculating the phase shift could result in this system if random noise shifted one of the samples and not the other, or shifted both to a different degree. In that case, the phase shift read out from the parallel added would include the difference attributable to line interference as well as the intended shift in phase.

"Coherent" detection solved this problem by standardizing each IF signal as it came into the detector. The Russian CCIT article applied coherent techniques to a quadrature amplitude modulated system, distinguishable in its overall operation from the Milgo modems. It did, however, explain how to phase lock an incoming IF signal to an "N" (where N = the number of possible levels of phase or amplitude) times the IF pulse emitted by a stable oscillator. Ragsdale combined this teaching with the use of digital circuits and center sampling, previously found in the Whang and Payne patents, to create the system claimed in Ragsdale '381.

Along the way, Ragsdale also found a way to maintain a stable clock for center sampling, even in the presence of repeated phase shifts. Bennett & Davey, a standard reference work published in 1965, expressed the view that it was "impossible" to maintain stable timing in a phase modulated system in the presence of repeated phase shifts. Repeated phase shifts alter the apparent frequency of the incoming signal. Consequently, there is no received energy at the carrier frequency, and no way to maintain the clock. Ragsdale found that by adding a small amount of amplitude modulation to the carrier, he could retain a stable clock even if the data pattern was one with an endless series of identical phase shifts. The Ragsdale patent does not, however, disclose the use of amplitude modulation to retain the clock.

Milgo did not reveal the Russian article, and its influence on Ragsdale's creative process, in Kansas, although it had acknowledged the article in its own contribution to the CCIT in April 1968. Instead, Ragsdale claimed for his own the improved signal to noise performance of coherent detection. In its proposed Findings of Fact in Kansas, Milgo touted:

47. The Payne '023 patent led Ragsdale to the development of the Ragsdale '381 digital coherent detection patent. (R. 1109-1117). The digital detector of the Ragsdale '381 patent accomplished phase locking of two signals of different frequencies in a restricted bandwidth modem in spite of the presence of repeated phase shifts as acknowledged by the defendant's expert Dr. Beam. (R. 2086). Originally, Dr. Beam asserted that the Kawai patent (DX Z) and the Lender patent (DX A-1) showed the method of deriving a reference carrier for coherent detection. However, Dr. Beam finally admitted that phase locking two such sinusoidal signals of different frequencies was a difficult task, (R. 2066), and further admitted that he knew of no prior art that did what Ragsdale did in the '381 patent and that included all the technical things that Dr. Beam could think of. Defendant's own technical expert admitted that the Ragsdale '381 patent distinguishes over the prior art he was aware of, and the claimed features of the Ragsdale '381 patent represents significant and nonobvious improvements over all prior art known by Defendant's technical expert.

The Kansas Court adopted this Finding verbatim. (F.F. 63).

Milgo also persuaded the Kansas Court that the Ragsdale '381 patent enabled Milgo to do what Bennett and Davey had proclaimed impossible. (F.F. 64). This claim to have done the impossible appears to have influenced the Kansas Court's finding of patentability.

But neither Dr. Beam, UBC's expert, nor the Kansas Court knew about the Russian article. Unlike the Puente patent, on which the Kansas defendants relied, which disclosed phase locking two signals of the same frequency to one another, the Russian article, in contrast, illustrated phase locking signals in precisely the frequency relationship necessary to give Ragsdale's coherent phase detection.

Ragsdale admitted in this Court that he did not originate the use of amplitude modulation to retain the carrier signal in the presence of repeated phase shifts. Since neither phase locking coherent detection, nor the use of amplitude modulation to preserve the carrier were Ragsdale's invention, neither afforded a basis for patentability.

Although the evidence shows that persons holding responsible positions at Milgo, including Whang and Bleckner, Milgo's president, were aware of the Milgo CCIT contribution, which in turn cited the Russian article, Milgo did not disclose this article to the Kansas Court, or to the Patent Office. Nor did Milgo present Ragsdale '381 as a patentable combination of old elements. Rather, Milgo relied upon a feature known in the prior art, and which Milgo knew to be old, to persuade the Kansas Court of patentability. Whether Ragsdale's combination was or was not patentable, it was not what Milgo held it out to be. Once again the problem is not limited to Milgo's having withheld a possibly pertinent prior art reference. It deliberately undertook to convince the Kansas Court of something it knew was untrue. Milgo was guilty, not merely of withholding valuable information, but of fabricating a theory of patentability. The result was that the Kansas Court was unable to make a fair judgment on patentability.

As if what has gone before were not enough, there are additional, and extremely serious, claims of inequitable conduct which extend from the initial Ragsdale application to the conduct of discovery in this Court. Rixon accuses Milgo of attempting to conceal a commercial use or sale of the Ragsdale detector more than a year before the application date of March 17, 1969. Although Rixon cites other commercial dealings, the crucial factual question for me to decide is when Milgo agreed to incorporate Ragsdale detectors in the modems it had contracted to sell to Western Union, and, if it made such an agreement prior to the critical date, whether the agreement was for experimental purposes.

Rixon, as with all fraud claims, bears the burden of clear and convincing proof, although once Rixon has established a use or sale before the critical date, Milgo has the burden of showing that it was for experimental, not commercial reasons. Paeco, Inc. v. Applied Moldings, Inc., 562 F.2d 870, 872 (3d Cir.1977); Carborundum Co. v. Combustion Engineering, Inc., 505 F. Supp. 1011, 1026 (D.Del.1981). My review of the evidence convinces me beyond any shadow of a doubt that such a sale took place, that it was a commercial sale, and that Milgo deliberately sought to conceal the truth from this Court by withholding documents in discovery.

The story begins with Whang's conversation with Ragsdale about coherent detection. We have no precise date for that event, but Milgo's engineering records show that Ragsdale had constructed a coherent detector by late January, 1968. Western Union had already agreed to purchase over $600,000 worth of Milgo series 4400 modems, using the Payne digital detector. The circuitry for the Payne detectors was located on two circuit cards in the Western Union modems, 107 and 108. By February 29, 1968, Milgo had modified 107 and 108 circuit cards to incorporate the Ragsdale phase lock loop and shipped one of each of the modified cards to its customers. (PX 254). An internal Western Union memorandum dated March 1, 1968 reflects the understanding that:

Milgo has implemented a new engineering change in their 4800 baud modem using phase lock loop techniques. A four dB improvement in signal to white-noise performance on the average line has been achieved compared with the models now under acceptance test. We have confirmed this improvement using replacement cards loaned to us by Milgo. This corresponds to a maximum theoretical improvement in error rate of about 100 to one. Milgo will furnish us three sets of conversion cards at no charge. One set will be used in the modem now being evaluated and two sets for the modems to be used in the Autodin tests. Negotiations should now be initiated with Milgo to retrofit the 15-4800 baud sets now being delivered. Similar consideration should be given to retrofitting the 2400 baud units currently in stock. This improvement will automatically be included in the balance of the present order for 4800 baud units and in all future orders. (Emphasis supplied).

(PX 341). Milgo shipped two additional sets of B cards on March 5. (PX 256). A telefax from William McCormick of Western Union to E.L. Claire of Milgo, dated March 5, 1968, confirmed McCormick's understanding "that the 4800 baud sets will have improved detection system (phase lock detector) as discussed and demonstrated to our J.E. Boughtwood Feb. 27." (PX 255). A Milgo party list for the 4402 modem was changed on March 11, 1968 to reflect the substitution of B cards for the A detector. (PX 319). Western Union and Milgo formally amended the contract to acknowledge the substitution of a phase lock detector on March 21, 1968. (PX 264).

Milgo shipped the first 20 units incorporating the Ragsdale detector to Western Union on March 22. (PX 266). On April 22, 1968, Milgo advised Western Union that it would cost $2700 to update the first 15 data sets shipped without Ragsdale detectors before March 14. "This modification," wrote Milgo's marketing chief, "would bring these modems completely up-to-date and would include incorporation of the phase lock detector as well as a number of smaller changes." (PX 305).

Although the formal contract amendment occurred after the critical date, as did the first shipment of completed modems with Ragsdale detectors, Milgo's documents unequivocally establish that Milgo offered the new detector for commercial sale no later than March 1, 1968, a full two weeks before the critical date. Doubtless Milgo was eager to find out how the phase lock detector would perform in service, but there is nothing in the documents which suggests that Western Union purchased the new detector as an "experimental device." Nor was this a case of an agreement to sell a device not yet reduced to practice. Milgo had already developed the new detector, had demonstrated it to Western Union's Boughtwood, and had shipped three sets of cards to Western Union long before the critical date.

"When an invention is offered for sale more than a year before the patent application is filed, it is `on sale' within the meaning of § 102(b), even if an actual sale has not occurred. This is true, even when (a) but one offer has been made to but one customer; (b) the prices are only estimated rather than established; (c) no commercial production runs have ever been made; and (d) the alleged invention is never sold." Chromalloy American Corp. v. Alloy Surfaces Co., 339 F. Supp. 859, 865 (D.Del.1972). The negotiations between Western Union and Milgo went far beyond this minimum prior to the crucial date.

"The experimental use doctrine `allows an inventor a reasonable period wherein he may perfect his ideas, provided the inventor truly has utilized the public use or sale to that laudatory end, not as a competitive tool to exploit his invention and gain an advantage over others.'" Carborundum Co. v. Combustion Engineering, Inc., supra, 505 F.Supp. at 1028, quoting Paeco, Inc. v. Applied Moldings, Inc., supra, 562 F.2d at 874. The evidence shows that Milgo's objective was commercial, not experimental.

This is only the beginning, however. Rixon cannot prove fraud merely by demonstrating a prior use or sale. True, Ragsdale signed an oath certifying that the invention disclosed in his application had not been placed on sale more than a year earlier. But "the mere misstatement of a material fact on the patent application is [not], by itself, sufficient to support a finding of fraud. Fraud requires something more. There must be an affirmative intent to deceive or at least a gross negligence in misrepresenting the truth." DeLong Corp. v. Raymond Int'l, Inc., 622 F.2d 1135, 1146 (3d Cir.1980). See also In re Coordinated Pretrial Proceedings in Antibiotic Antitrust Actions, 676 F.2d 51 (3d Cir.1982). Ragsdale's factually incorrect oath does not prove fraud. Regrettably, in this case there is something more, however.

Milgo failed to produce in discovery all of the documents on which this Court has relied in the foregoing discussion of the Western Union sale. These documents came to light only because Rixon obtained many of them from other sources and surprised Milgo's president, Mr. Bleckner, with questions based on documents which Milgo thought it had kept hidden. While it appears that Milgo's counsel showed the March 5 telefax confirming the agreement to sell B detectors to Milgo's president before his deposition, Milgo produced the documents to Rixon only after I granted Rixon's motion to compel following the Bleckner deposition.

I simply cannot accept the explanation of Milgo's associate general counsel that the documents were lost in the shuffle of papers from Milgo's warehouses to its adversaries in a number of pending civil actions. The omissions were too convenient for Milgo. Moreover, there are inconsistencies in this explanation. The files containing the documents just "appeared" in his office one day. Although he believed the documents had already been turned over in discovery, he turned over the file to Milgo's trial counsel. I am not sure who is personally at fault, but I am persuaded clearly and convincingly that Milgo deliberately concealed documents which prove that the Ragsdale '381 patent is invalid under 35 U.S.C. § 102(b). This is the kind of conduct which corrupts the truth finding ability of a court and which cannot be countenanced. Independent of any misconduct with respect to the patent in Kansas, therefore, I hold Ragsdale '381 unenforceable on the ground of fraud.

IV. CLAIM AND ISSUE PRECLUSION AND THE WHANG '023.

I have earlier concluded that Rixon III is free to assert its claims that the Payne and Ragsdale patents are unenforceable and have, accordingly, made findings of fact with respect to those claims. I have also made findings of fact with respect to the Whang '023 unenforceability claim, despite my conclusion that Rixon III stands in UBC's shoes with respect to the Kansas judgment on Milgo's '023 infringement claim, for a number of reasons. First, this issue has been fully tried before me and, even if I were to conclude that Rixon III is precluded from litigating its unenforceability claim with respect to the Whang '023, it would nevertheless be appropriate for me to record my conclusion in light of the possibility of an appeal.

In addition, Milgo stands before me seeking affirmative relief by way of an accounting and an injunction on the basis of the Whang '023 patent. Whatever may be the res judicata and collateral estoppel effects of the Kansas proceedings in the context of this case, those effects do not require this Court to provide aid to one which, by manufacturing favorable evidence and suppressing unfavorable evidence, has sought to reap the benefits of a monopoly on "narrow skirt" modems to which it knew it was not entitled. The courts, mindful of the special character of a patent monopoly and of the public interest in restricting such monopolies to their intended role, have applied traditional claim and issue preclusion doctrines in a flexible manner in similar contexts. E.g., Mercoid Corp. v. Mid-Continent Invest. Co., 320 U.S. 661, 64 S. Ct. 268, 88 L. Ed. 376 (1944); Bros. Incorporation v. W.E. Grace Manufacturing Company, 320 F.2d 594 (5th Cir.1963) and 351 F.2d 208 (5th Cir.1965). In particular, the Supreme Court has held in circumstances very similar to those now before this Court that res judicata should not be applied to bar litigation of whether judicial relief should be denied to one which has allegedly misused its patent. Mercoid Corp. v. Mid-Continent Invest. Co., supra.

In the Mercoid case, Mid-Continent sued Mercoid for contributory infringement of its patent. In defense, Mercoid asserted that Mid-Continent had attempted by tying a patented and an unpatented product to expand its monopoly beyond the scope of the patent and, accordingly, that it had unclean hands. Mid-Continent pleaded res judicata relying on a judgment sustaining the validity of the patent in an action in which Mercoid had provided the defense. It pointed out that Mercoid knew of the alleged tying agreement at the time of the earlier suit and could have asserted it by way of defense. While the opinion contains references to the anti-trust aspects of Mid-Continent's conduct, a theory Rixon does not here press, the rationale of the Court's holding is equally applicable here:

Respondents ask the equity court for an injunction against infringement by petitioner of the patent in question and for an accounting. Should such a decree be entered, the Court would be placing its imprimatur on a scheme which involves a misuse of the patent privilege and a violation of the anti-trust laws. It would aid in the consummation of a conspiracy to expand a patent beyond its legitimate scope. But patentees and licensees cannot secure aid from the court to bring such an event to pass, "unless it is in accordance with policy to grant that help." Beasley v. Texas & Pacific Ry. Co., 191 U.S. 492, 497 [24 S. Ct. 164, 166, 48 L. Ed. 274]. And the determination of that policy is not "at the mercy" of the parties (id., p. 498 [24 S.Ct. at p. 166]) nor dependent on the usual rules governing the settlement of private litigation. "Courts of equity may, and frequently do, go much farther both to give and withhold hold relief in furtherance of the public interest than they are accustomed to go when only private interests are involved." Virginian Ry. Co. v. System Federation, 300 U.S. 515, 552 [57 S. Ct. 592, 601, 81 L. Ed. 789]. "Where an important public interest would be prejudiced," the reasons for denying injunctive relief "may be compelling." Harrisonville v. Dickey Clay Co., 289 U.S. 334, 338 [53 S. Ct. 602, 603, 77 L. Ed. 1208]. And see United States v. Morgan, 307 U.S. 183, 194 [59 S. Ct. 795, 801, 83 L. Ed. 1211]. That is the principle which has led this Court in the past to withhold aid from a patentee in suits for either direct or indirect infringement where the patent was being misused. Morton Salt Co. v. G.S. Suppiger Co., supra, [314 U.S. 488 at] p. 492 [62 S. Ct. 402 at p. 405, 86 L. Ed. 363]. That principle is controlling here. ...

Under this authority, I conclude that the facts found above require that relief be denied to Milgo on any Whang '023 based claim. As in Hazel-Atlas Glass Co. v. Hartford-Empire Co., 322 U.S. 238, 64 S. Ct. 997, 88 L. Ed. 1250 (1944), "this is not simply a case of a judgment obtained with the aid of a witness who, on the basis of after-discovered evidence, is believed possibly to have been guilty of perjury. Here, ... we find a deliberately planned and carefully executed scheme to defraud ... [a] Court ... through the presentation of a theory of patentability manufactured by a patent holder out of whole cloth." 322 U.S. at 245, 64 S.Ct. at 1001.^[19]

Finally, the foregoing factual findings with respect to the Whang `023 are appropriate because they may have significance in this case in contexts other than Milgo's counterclaim prayers for affirmative relief. I conclude that that possibility should be explored further by counsel and the Court along with the possibly related issue of what effect, if any, the finding and conclusions of the Codex court have on the issues in this case. I will confer with counsel about these matters promptly.

V. CONCLUSION.

The relationship between Rixon II and UBC was not such that Rixon II was bound by the Kansas judgment as the "alter ego" of UBC. With respect to the claims in the Kansas action at the time of the transfer of UBC's modem marketing business to Rixon II on January 1, 1972, however, Rixon II was the successor in interest to UBC and was bound by the Kansas judgment to the same extent as UBC. Since Rixon III ultimately became the successor in interest to Rixon II, it likewise stands in the shoes of UBC with respect to all claims asserted in Kansas before January 1, 1972. The Kansas judgment has no effect on Rixon III, however, with respect to claims asserted by Milgo after January 1, 1972.

Finally, based on Milgo's conduct before the Kansas Court and before this Court, the Ragsdale '381 patent is wholly unenforceable and the Payne '023 is unenforceable in part. Also, based upon that conduct, Milgo is entitled to no affirmative relief from this Court with respect to the Whang '023 patent.

NOTES

[1] A binary digital signal is one which has only two possible electrical states, e.g. "on" or "off", which are usually interpreted as ls and 0s. Computers store and compute with sequences of ls and 0s which may represent numbers or letters. To a computer, therefore, information takes the form of a "square wave" such as:

[2] For instance, in the Whang 8-phase modulated system each word contains 3 bits (each 1 or 0 is a bit). There are 8 possible 3 bit words: 000, 001, 010, 011, 100, 101, 110, 111. Whang represents each of these by the 8 possible 45° phase shifts of a given carrier pulse.

[3] In a seminal article in the field, Alexander, Gryb and Nast explained that transmission of a digital pulse at high speeds would require an infinite frequency bandwidth. Their survey went on to demonstrate that the usable portion of the commercial telephone network had a bandwidth of only about 3000 Hz. Today commercially available modems operate at speeds of up to 9600 bits per second ("bps"), four times the capacity of the Bell 201 modem in existence in 1966.

[4] In 1978 Milgo Electronics merged into Racal-Milgo, Inc., the defendant in this action.

[5] There are five patents involved in this action altogether, but the res judicata issues implicate only three described above.

[6] The Kansas Court rejected Milgo's claim that United was liable as the alter ego of Rixon II, and found that United had not itself committed any infringing activities. Milgo Electronics Corp. v. United Telecommunications, Inc., 189 U.S.P.Q. 160, 190 (D.Kan.1976), aff'd, 623 F.2d 645 (10th Cir.1978) (per curiam).

[7] Milgo Electronics Corp. v. United Telecommunications, Inc., supra. After this affirmance, Milgo filed a motion that Rixon be found in contempt of the Kansas injunction. Prior to the trial in this case, proceedings on that motion were stayed by the Kansas Court.

[8] See Blonder-Tongue Laboratories v. University of Illinois Foundation, 402 U.S. 313, 91 S. Ct. 1434, 28 L. Ed. 2d 788 (1971).

[9] One Hertz is equal to one cycle per second.

[10] Nyquist's work dealt with the transmission of telegraph signals, but his mathematical analysis was known to apply to the transmission of digital data by analog pulses as well. W. Bennett and J. Davey, Data Transmission (1965) 61-66.

[11] Just as plucking the string of a musical instrument excites vibrations at a central frequency and its harmonic frequencies, so too does creating an electrical impulse propagate pulses of lower amplitude at the harmonic frequencies.

[12] A baseband system uses only the pulses on one side of the carrier. Phase modulation requires using both side bands.

[13] For example, Whang expressed some misgivings in his patent about the feasibility of a 16 phase system which require the detector to be accurate within 11° of phase.

[14] The transfer prices were fixed with the recognition that Rixon II would incur no marketing costs.

[15] PX 2, Application dated July 14, 1966, at p. 3.

[16] The width of the passband and the width of the skirts depend upon where you measure them. The width of the Nyquist brick wall filter is the same at all amplitude levels, but such filters cannot be realized in practice. Whang feared that signals passing outside the equalized portion of the telephone line might interfere with data transmission, but this depended upon how strong those signals were. Whang's patent does not disclose how weak those frequencies outside the good portion of the line must become before they no longer post a danger of interference. Thus, when the patent speaks of a ± 500 Hz bandwidth measured from a center frequency, it does not say whether this bandwidth is defined at 100% power and amplitude (the "top") or at close to 0% (the "bottom").

Rixon's experts testified that engineering convention describes the passband of a filter at the halfpower, or -3dB points, unless otherwise specified. See Tr. A194-196 (Kretzmer); Tr. G20-22 (Forney). Whang, in the Codex trial, agreed that one measured the "nominal bandwidth" of a filter at the -3dB attenuation level. PX 396, Tr. 6, 79-81, 84. I conclude that, in the absence of specification, Whang must have intended his references to passband and bandwidth to be measured at -3dB, and that, in any event, he did not — as Milgo suggests — define bandwidth to be measured at the zero amplitude and power points.

[17] The new filters narrowed the data bandwidth to clear a channel for voice signals at the lower end of the telephone frequency spectrum.

[18] Milgo's conduct with respect to Secretan may be a basis for obtaining some relief from the Kansas judgment but that issue is not before me.

[19] The Hazel-Atlas case is helpful on another point as well. It is true that if UBC's Kansas expert had tested the allegedly "breakthrough" Milgo modems, he would have discovered that they did not have narrow skirts and Milgo's suppression of the Whang filter files would have been rendered immaterial. This fact is no aid to Milgo, however. In Hazel-Atlas, Justice Black made the following observations about a similar situation:

[E]ven if Hazel did not exercise the highest degree of diligence, Hartford's fraud cannot be condoned for that reason alone. This matter does not concern only private parties. There are issues of great moment to the public in a patent suit. Mercoid Corporation v. Mid-Continent Investment Co., 320 U.S. 661 [64 S. Ct. 268, 88 L. Ed. 376]; Morton Salt Co. v. G.S. Suppiger Co., 314 U.S. 488 [62 S. Ct. 402, 86 L. Ed. 363]. Furthermore, tampering with the administration of justice in the manner indisputably shown here involves far more than an injury to a single litigant. It is a wrong against the institutions set up to protect and safeguard the public, institutions in which fraud cannot complacently be tolerated consistently with the good order of society. Surely, it cannot be that preservation of the integrity of the judicial process must always wait upon the diligence of litigants. The public welfare demands that the agencies of public justice be not so impotent that they must always be mute and helpless victims of deception and fraud.