dissenting.
With all respect to my colleagues on this panel, they have erred in law and in procedure, in their reversal of the findings and rulings of the Commission on the issues presented in this appeal. The Commission found infringement by some of the Ground Fault Circuit Interrupters of some respon*1314dents, In re Certain Ground Fault Circuit Interrupters and Prods. Containing Same, Inv. No. 337-TA-615, 2009 WL 962585 (Int’l Trade Comm’n Mar. 9, 2009), affirming with modifications the findings and rulings of the ALJ in In re Certain Ground Fault Circuit Interrupters and Products Containing Same, Inv. No. 337-TA-615 (Int’l Trade Comm’n Sept. 24, 2008) (“Initial Decision” or “I.D.”). The Initial Decision of 170 pages and the Commission’s supplemental rulings of 32 pages present a full understanding of the technology, with rigorous discussion of the evidence and extensive analysis, findings, and conclusions. This court now finds its own facts, applies theories that were not raised by any party, uses incorrect standards of review, and creates its own electrical technology contrary to the uniform and unchallenged expert testimony.
In the companion cross-appeal decided today by separate opinion, Appeal Nos. 2009-1338, -1369, taken by the patentee Pass & Seymour from the Commission’s rulings adverse to it, the court deferentially affirms the Commission’s findings of non-infringement, while in this appeal my colleagues act de novo to reverse the Commission’s findings of infringement. In reversing the Commission, the court does not discuss the substantial evidence supporting the Commission’s findings. In contrast to the companion appeal, the court bypasses the standards of APA review, and makes new finding and rulings on new theories, some of which were not presented and not argued by any party, and which are conspicuously incorrect. Further, despite its changes in claim construction, its new findings, and its de novo applications of law to fact, the court provides no opportunity for the patentee or the Commission to respond to the court’s new arguments and adjudications. I must, respectfully, dissent.
Discussion
The three appellants are Shanghai ELE Manufacturing Corp. (“ELE”), General Protecht Group (“GPG”), and Wenzhou Trimone Science and Technology Electric Co. (“Trimone”). The claim elements upon which my colleagues reverse the Commission are the “detection circuit” and “load terminals” of the '340 patent, and the “latching means” of the '398 patent.
A
The '340 patent: “detection circuit ... to generate a predetermined signal”
The Commission found infringement of the '340 patent by the 2006 model of the Ground Fault Circuit Interrupter (“Interrupter” or “GFCI”) of ELE, the 2003 and 2006 models of GPG, and the 2006 model of Trimone. Claim 14 is representative; the clause at issue is:
at least one detection circuit including a circuit segment coupled between the line terminals and configured to generate a predetermined signal in response to detecting a proper wiring condition....
'340 patent col.10 11.9-12. The ALJ heard testimony of expert witnesses for all parties, and found that “the concept of ‘detection’ is well understood in the context of circuits. In that regard, a circuit reacts in a particular way to a particular stimulus.” I.D. at 85 (citing testimony of Pass & Seymour’s expert Dr. Harman,1 Hr’g Tr. 1036:20-1040:13, J.A. 40444-45). Extensive testimony concerned the detection circuitry as described in the '340 patent and *1315as embodied in each of the accused circuit interrupters. The ALJ found infringement as to this element, upon the same findings and arguments now adopted by my colleagues to contrary effect. However, sufficient evidence to reverse these findings and rulings has not been shown.
The '340 patent describes miswire protection through circuitry that repeatedly trips the Interrupter until a predetermined signal is provided that indicates proper wiring. The predetermined signal is an electrical signal that cuts off the circuit path that was causing the repeated tripping of the device. In patent Figure 4, shown below, the detection circuit consists of switch SI, fuse FI, and resistors RIO and R13:
[[Image here]]
'310 Patent, Figure k
When current flows through the detection circuit to the gate of rectifier 404, the device will trip by energizing solenoid 406, which opens interrupting contacts 408. '340 patent col.7 11.11-16. When power is correctly wired to the line terminals, current passes through the detection circuit regardless of whether the interrupting contacts 408 are open or closed. With correct wiring, current will continue to flow “until such time as fuse FI clears, after which it is possible to accomplish a resetting of the interrupting contacts 408.” Id. col. 7 11.24-26. The '340 specification explains that the detection circuit uses power from the line terminal to create a response, and this response is present only when the device is properly wired. When properly wired, the device is allowed to be reset. The embodiment shown in Figure 1 of the '340 patent illustrates the same principle, where the circuit trips in response to a simulated fault that is present until proper wiring has cleared the resistive elements, thereby permitting resetting. See id. col. 4 11.9-67.
The experts agreed on the technology as set forth in the patent and as embodied in the devices that the Commission found to be infringing. In ELE’s device the reset button is blocked until, when power is correctly wired to the line terminals, current passes through a circuit segment that signals a solenoid that unblocks the reset button. See J.A. 56301-09. The signal is not produced if power is incorrectly wired. Dr. Harman explained that the portion of *1316the ELE circuitry configured to provide the signal when the device is properly wired is the detection circuit of the '340 patent, and generates the signal that allows the device to be reset. See Hr’g Tr. 844:13-850:6, J.A. 40396-97.
The Commission found that the ELE device indeed has a detection circuit that generates a signal. See I.D. at 91. “This circuit provides a predetermined signal upon proper wiring, and this signal does not simulate a fault condition.” Id. The presence of that signal allows ELE’s device to be reset, as described and claimed in the '340 patent. Reversing the Commission, my colleagues find that the ELE device does not detect whether the wiring is proper and generate a predetermined signal in response; my colleagues find that ELE simply uses the available line voltage to allow the device to be reset when properly wired. However, as the Commission found, in ELE’s device the reset state is prevented until power from a properly wired line terminal is used to energize a solenoid that moves the block, requiring detection of proper wiring. Substantial evidence supports the Commission’s finding that the ELE device meets the “detection circuit” and “predetermined signal” limitations of the claims. Indeed no other finding is plausible; that is the purpose of the device. My colleagues do not explain their rejection of the Commission’s findings, and indeed, I can discern no basis for such rejection.
The Commission also sustained the ALJ’s findings with respect to the GPG devices. As GPG explained, if power is properly connected to the line terminals, then when the reset button is pushed a “reset solenoid” is energized that resets the device. See GPG Br. 15-16. GPG does not dispute that its Interrupters generate a predetermined signal. GPG states in its brief that “the predetermined signal is generated when GPG’s products are mi-swired,” GPG Br. 36 (citing J.A. 40969-70, testimony of GPG expert Dr. Roberge). GPG also states that “GPG does not take issue with the words used in the ID’s construction of ‘detection circuit.’ ” GPG Br. 27. Nonetheless, my colleagues negate the Commission’s construction. This is inappropriate, for the Commission’s construction is not disputed, as the experts agreed.
GPG did argue non-infringement on the Commission’s construction, contending that its GFCIs do not infringe because one could theoretically avoid miswire protection by the following expedient: buy a GPG GFCI (sold in the tripped state), properly wire it, reset it, and then uninstall and miswire it. GPG argued that the claim clause “configured to generate a predetermined signal in response to detecting a proper wiring condition,” which occurs when the line terminals are connected to a source of AC power, is not infringed if the signal can be sent when the AC power is connected to the load terminals. See GPG Br. 35-36. The Commission found that this theory did not avoid infringement. I.D. at 99 & n. 23 (citing z4. Techs., Inc. v. Microsoft Corp., 507 F.3d 1340, 1350 (Fed. Cir.2007) (“[Ijnfringement is not avoided merely because a non-infringing mode of operation is possible.”); Golden Blount, Inc. v. Robert H. Peterson Co., 438 F.3d 1354, 1363 (Fed.Cir.2006) (“[I]t matters not that the assembled device can be manipulated into a non-infringing configuration.”)).
The witnesses explained that in GPG’s devices the reset state is prevented until power from a properly wired line terminal is used to energize a solenoid that resets the device. The Commission found that “it is not possible to close the contacts to reset the GFCI without the predetermined signal.” I.D. at 97. While GPG describes the operation of its device as “inherent” miswire protection, the Commission deter*1317mined that this is not different from the claimed detection, for the circuit generates a signal only when power is properly connected to the line terminals. I.D. at 97-98.
My colleagues do not discuss the support for the Commission’s findings that the ELE and GPG Interrupters have a detection circuit that generates a predetermined signal in response to proper wiring. These systems require detection of proper wiring and generation of a predetermined signal in response. Instead, the court creates its own, flawed, definition of “generate,” and finds non-infringement by ELE and GPG based on this definition. However, the definition of “generate” was not disputed by any party before the Commission; thus the record is sparse and argument is nil.2 The expert witnesses all understood the term in the same way, and directed them testimony to their opinions as to whether the term was infringed. Thus the ELE expert was of the opinion that the ELE device does not generate a “predetermined signal” because the ELE device does not sense a current threshold or time threshold. However, ELE’s expert admitted that the ELE circuit creates (i.e., generates) a current. J.A. 40818 (“Shanghai ELE ... simply uses the available line voltage to create a current....”). The panel majority does not discuss this evidence.
Instead, the court creates a theory not proposed by any party, and rules that a signal that originates from the line terminal is not generated by the detection circuit and thus is not “generated.” See Maj. Op. at 1308 (“The identified circuit does not generate this current; it is the current that comes from the AC power connection.”). There is no support for this theory. The only signal generated in response to proper wiring as described in the '340 patent is “generated” in the same way as in the accused devices, that is, the detection circuit uses power from the properly wired line terminals to supply a signal (claim 14) or response (claim 30) that allows the device to be reset. That is how the Commission, and the parties, interpreted the term. The court’s new interpretation of “generate” was not debated and is not briefed.
On the record and proceedings in the Commission, substantial evidence supports the Commission’s findings and rulings as to this claim clause. The court’s reversal of the Commission is unwarranted. At a minimum, this new (incorrect) definition requires remand for the views of the parties and the Commission.
B
The '340 patent: “interrupting contacts ... load terminals”
Another finding on which the court improperly reverses the Commission relates to the claim clause:
*1318interrupting contact assembly including four sets of interrupting contacts that are configured to provide electrical continuity between the line terminals and the load terminals in a reset state and configured to interrupt the electrical continuity in tripped state ....
'340 patent col. 10 11.17-22. The Commission construed this claim element as “four pairs of electrical contacts that can separate from each other to interrupt the flow of electricity,” I.D. at 89. ELE’s expert, Dr. Engel, stated that ELE’s Interrupters “absolutely” have “four sets of interrupting contacts under the plain meaning of that phrase.” J.A. 40840. The Commission’s findings are well supported by substantial evidence.
Neither ELE nor Trimone disputes that its devices have four sets of interrupting contacts between the line and load. However, the court holds that two of these contacts are not connected to “load terminals” as required by the claim. The '340 specification explains that the purpose of the Ground Fault Circuit Interrupter is to remove power from the load terminals, by interrupting the circuit from the power source. '340 patent col.2 11.8-10 (“Circuit interruption is typically effected by opening a set of contacts disposed between the source of power and the load.”). A person of ordinary skill in this field would understand that the “interrupting contact assembly” of the clause is configured to interrupt the continuity between the power source (the line terminals) and load terminals such as receptacle outlets.
The '340 patent states:
The electric circuits may typically include one or more receptacle outlets and may further transmit AC power to one or more electrically powered devices, commonly referred to in the art as load circuits. The receptacle outlets provide power to user-accessible loads that include a power cord and plug, the plug being insertable into the receptacle outlet.
Col.l 11.27-33. No one disputes that receptacle outlets are referred to in the electrical art as “user load terminals” or “user-accessible load terminals.” Multiple witnesses testified that “load terminals” include user-accessible load terminals such as receptacle outlets, explaining that receptacle outlets are sometimes more specifically referred to as “user load terminals” or “user-accessible load terminals.” Dr. Horenstein, an expert testifying on behalf of respondent Meihao, stated that “receptacles are also referred to as user accessible load terminals,” J.A. 40938. Dr. Harman explained that “the load terminals would be the terminals for any other devices or outlets or appliances that would be connected to the device,” J.A. 40408; “The loads are on the right side labeled load and outlet receptacle,” J.A. 40401. ELE’s expert witness, Dr. Engel, likewise testified that in ELE’s design “there, are two load terminals, so this, Your Honor, shows a wire going to a receptacle or a load terminal available on the face of the device, and this shows a wire going to a load terminal available on the side of the device.” J.A. 40819. ELE does not argue that the claimed “load terminals” do not include user load terminals such as the receptacle outlets shown in the '340 patent.3
*1319Contrary to this unanimity of experts, my colleagues dispose of the experts with the statement that “an expert’s subjective understanding of a patent term is irrelevant.” However, an expert’s testimony concerning the definition of a technological term in his field of expertise is highly relevant. An objective definition of a term of electrical art is not “subjective understanding.”
The Commission accepted the uniform extrinsic and intrinsic evidence, and proceeded on the premise that “user load terminals” are a form of “load terminal.” Yet the panel majority now reverses the Commission and rules that “load terminals” excludes “user load terminals.” This ruling is contrary to the patent specification and contrary to the entire record. The panel majority, seeking support, simply argues that the experts were “fastidious” in referring to “user load terminals” separately from “load terminals,” Maj. Op. at 1311, and indeed the experts differentiated the broader general term “load terminal” from the more specific “user load terminal” such as receptacle outlets. No one except this court proposes that “load terminal” by definition excludes user load terminals.
The Commission discussed Trimone’s attorney argument that, despite this uncon-troverted understanding and usage, the '340 patent claims exclude receptacle outlets from the scope of “load terminal.” The Commission heard this argument, and found no distinction between the meaning of “load terminals” in this electrical technology and the “load terminals” described and claimed in the '340 patent. The Commission rejected Trimone’s argument, and declined to remove “user load terminals” from the scope of “load terminals.” The Commission explained that “the testimony of multiple witnesses confirms that outlets/receptacles have load terminals. The fact that one would be unlikely to miswire a GFCI by wiring AC power to the outlets/receptacles does not preclude them from being considered ‘load terminals’ in the context of miswiring protection.” I.D. at 89 (citations omitted).
It is simply incorrect for this court now to hold that “load terminals” exclude user load terminals. The Commission’s finding of infringement by the user load terminals is supported by substantial evidence; the court’s contrary finding is totally devoid of support. I must again dissent from my colleagues’ refusal to apply the proper standard of review and failure to credit the Commission’s correct rulings and findings.
C
The '398 patent: “latching means”
ELE does not dispute that its devices embody this element of the '398 patent. The Commission found, applying 35 U.S.C. § 112 ¶ 6, that GPG’s 2006 Interrupter includes latching structure equivalent to that described in the '398 patent. Although this finding is supported by substantial evidence, the panel majority rejects it.
Claim 1 of the '398 patent includes the following clauses:
f) latching means releasably retaining said conducting member in said first position; and
g) actuating means for releasing said latching means to permit said biasing means to move said conducting member to said second position in response to a predetermined fault condition in said electrical circuit.
'398 patent col.13 11.41-46. The “first position” is a “circuit-making” position in which the conducting member completes the circuit in the Ground Fault Circuit Interrupter, allowing power to be delivered from line to load. The “second position” is the circuit-breaking position. The *1320'398 patent describes a mechanical latching structure wherein the shoulder of a pin catches on a latch connected to the conducting member, and a spring biases the latch member and conducting member into the designated first position. If the latch is disengaged, the spring will bias the latch member and the conducting member into the circuit-breaking position.
The GPG 2006 Interrupter uses a permanent magnet to hold the conducting member in the first position, against the bias of a spring. The Commission found infringement in terms of § 112 ¶ 6. The Commission received evidence of the known interchangeability of mechanical and magnetic latches, including expert testimony that prior art patents “speak about a magnetic latch in a GFCI,” J.A. 40524, and that “[t]he use of a magnet as a latch is fairly common in other devices,” id. The Commission found the fact that “at the filing of the '398 patent, a person of ordinary skill in the art would have considered the magnetic latching structure in the GPG 2006 GFCIs to be interchangeable with the ‘latching means’ recited in claim 1.” I.D. at 67. The Commission explained that in these devices the mechanical and magnetic latches work in substantially the same way, for in both cases the force of the pin or the magnet overcomes the biasing force of a spring to hold the conducting members in the circuit-making position. I.D. at 66-67.
The court rejects the Commission’s finding of equivalency, without discussing the evidence for or against this finding. Instead, the court rules that there can never be equivalence between a mechanical latch and a magnetic latch, misciting Toro Co. v. Deere & Co., 355 F.3d 1313 (Fed.Cir.2004). Toro did not so hold. My colleagues remove from its context Toro’s quotation of the argument of the defendant Deere, viz. “Deere contends that this difference means the two systems accomplish clause (e)’s function in fundamentally different ways.” Id. at 1324. This was not a pronouncement of law for all mechanical-magnetic substitutions. On the facts of that case, the mechanical camfollower that travels a slope, used in the Toro patent to lift a valve stem, was found to operate in a substantially different way from the magnetic solenoid used by Deere to create a magnetic force that pulls open a liquid valve. Id. This court in Toro affirmed the district court’s finding of non-equivalency for that system; the court did not establish a universal law of non-equivalence for all systems and all mechanical/magnetic facts.
Equivalence is a question of fact to be determined “against the context of the patent, the prior art, and the particular circumstances of the case.” Graver Tank & Mfg. Co. v. Linde Air Prods. Co., 339 U.S. 605, 609, 70 S.Ct. 854, 94 L.Ed. 1097 (1950). As explained in Warner-Jenkinson Co. v. Hilton Davis Chemical Co., 520 U.S. 17, 36, 117 S.Ct. 1040, 137 L.Ed.2d 146 (1997), “known interchangeability of substitutes for an element of a patent is one of the express objective factors noted by Graver Tank as bearing upon whether the accused device is substantially the same as the patented invention.” Substantial evidence supports the Commission’s finding that GPG’s 2006 Interrupter contains structure equivalent to the latching means described in the '398 patent. There was no contrary evidence.
My colleagues on this panel disregard the rulings and findings of the Commission, and render de novo rules and findings on new theories to which the parties have had no opportunity to respond. This is an inappropriate appellate process.
I respectfully dissent.
. Thomas Harman of the University of Houston is a special expert to the committee of the National Electrical Code dealing with house wiring and safety devices. J.A. 40385. Dr. Harman’s Ph.D. is in electrical engineering, and he is a master electrician who wrote the “Guide to the National Electrical Code.” Id. at 40385-86.
. The panel majority at footnote 3 disputes whether ELE raised the issue, citing ELE’s statement in its brief that "[b]y treating the 'predetermined signal’ as including whatever AC power happens to arrive at a building from the local power plant, the Commission has broadened this limitation beyond all recognition and effectively reads it out of the claim.” The context of this statement reveals that ELE was concerned only with the "predetermined” nature of the signal and whether it is used to differentiate between correct or incorrect wiring. The sentence from ELE’s brief quoted by the majority is preceded by the statement that "[i]n order to find the 'predetermined signal' in ELE's 2006 GFCIs, the Commission merely disregards this perspective and disregards the teachings of the '340 patent's specification about the meaning of "predetermined.” ELE Br. 42. The majority apparently does not credit any of ELE's arguments regarding the "predetermined” nature of the signal. In any event, the issue is not waiver of the argument, but rather the complete lack of any disagreement among any of the parties regarding the term "generate,” contrary to this court's sua sponte definition.
. The majority at footnote 4 states that ELE indeed "raises the load terminal issue” by stating in its brief that the '340 patent does not show separation of "the downstream load terminals from the surface ('face') load or receptacle terminals,” ELE Br. 44. Notably absent from the ELE briefs is any suggestion that "load terminals” do not include receptacle outlets. Indeed, the above-quoted portion of the ELE brief accepts that receptacles are surface load terminals, as ELE's expert conceded, consistent with the testimony of other experts.