Monday, May 10, 2010

Claims with ambiguous relationships between elements

Takeaway: Many claims include entities with more than one instance ("a plurality of widgets"). When the claim involves relationships between multiple entities with multiple instances, getting the claim language right can be challenging. For example, suppose the claim includes a plurality of widgets, a plurality of wikis, and links between the two. Without careful thought, the claim is likely to be ambiguous about whether a link exists between each widget-wiki pair, or between one particular widget and multiple wikis. In fact, such claims might even read on one link between the two groups, rather than links between particular widget-wiki pairs. Today's district court case shows how this sort of ambiguity gives parties something to argue about in a claim construction dispute.

Power Integrations, Inc. v. Fairchild Semiconductor Int’l, Inc., et al.,
C.A. No. 08-309-JJF-LPS
(D. Del. Dec. 18, 2009)

The claims at issue were directed to power converters. One of the limitations at issue was a controller that generated a discharge-time signal by "multisampling a voltage signal ... during an off-time of said switching signal." On its face, the limitation is ambiguous:
  • multiple samples during each off-time?
  • one sample during each off-time, with multiple off-times resulting in multiple samples?
The patentee argued for the first interpretation, while the accused infringer advocated for the second. [Because the patentee's is a narrower construction, I'm guessing that the broad construction leaves the claim vulnerable to an invalidity challenge.]

In support of the broad interpretation, the accused infringer relied on the basic claim construction principle "A means one or more." Thus, by using "AN off-time" the claim covers taking samples during more than one off time, which results in multiple samples. The accused infringer also relied on portions of the spec which show a comparator that uses only one sample per cycle to determine when the transformer has been discharged.

The patentee argued that "the claimed circuit could not perform its desired function if it operated in the manner required by [the accused infringer]". Specifically, the claimed discharge time signal results from taking multiple samples until the last one is different enough from the previous one to indicate the end of the discharge time. The patentee countered the accused infringer's comparator argument too. While agreeing that the comparator holds one sample per cycle, the patentee explained that precisely which sample is held by the comparator is determined by taking multiple samples during a single cycle.

The court adopted the patentee's construction. "While the claim language itself is ambiguous, the specification demonstrates that the voltage signal is sampled at least two times for each of the switching signal's off-times in order fo the correct sample to be generated ... There is no teaching as to how the circuit would operate if the voltage signal and discharge time were sampled only once during an off-time."

My two cents: I think the judge reached the right answer here. But I also think that the wording of the claim gave the losing side ammunition to argue. When possible, I'd rather have a claim that more clearly expresses the nature of the sample-cycle relationship so that going to the spec isn't necessary.

How about:
multisampling a voltage signal within each of a plurality of off-time cycles of said switching signal.
OK, "each" is a very limiting word, so I'll try to avoid it, while still reducing the ambiguity about multiple-across-all-cycles vs. multiple-in-a-single cycle.

How about:
multisampling a voltage signal within at least some of a plurality of off-time cycles of said witching signal.
I doubt my claim is bullet-proof, but I think it improves on capturing the feature asserted by the patentee.

No comments:

Post a Comment