The fifth/sixth area of the specification that Samsung attacked in the reexamination requests is the language that calls for “sampling data on the transition of an external clock signal” and similarly that a first portion of data is sampled “in response to a rising edge” of such signal while a second portion of data is sampled “in response to a falling edge” of such signal. Since the issues are intertwined I am discussing them both together here.
Of all the arguments raised to date in the reexam, this one has to give Rambus the most heart burn. At its core, this attack is directed to Rambus’ claims to double data rate (DDR) DRAM, and Samsung has clearly convinced the Examiner (to date anyway) that there is no support in the original specification for these claims.
The stress is multiplied because obviously if Rambus loses this point on this patent, it will lose the same point for at least another critical patents that it has purporting to cover DDR. As those of you who follow these cases surely know, DDR is mainstream technology, and Rambus has been successful at securing higher rates for such products on the platform that it had more claims covering such products. The claims at peril in the ‘020 include claims 5, 15, 26, 31, 33 and 41. Clearly however if the PTO kills these DDR related claims, it is apparent Rambus’ leverage would be severely compromised in all lawsuits involving claims of similar scope.
Here is what the Examiner said about Rambus’ support in the specification for the data sampling question:
The Examiner notes that this figure (13) discloses sampling the data at the even and odd inputs of two internal clocks, but as shown in the figure, data is sampled at one of the transitions.
So there is the double whammy; first, he concludes the transitions are based on an internal clock, and second, he maintains the data is only sampled at one transition. Ergo, no double rate data is disclosed.
But will the Examiner maintain this position? Both parties had an opportunity to comment on his findings, and my review suggests that there may be sliver of hope after all for Rambus to turn this around. This conclusion is surprising because at first glance the Examiner’s argument seems quite ….unassailable. First, one has to start with the claim; here is what claim 15 actually recites:
15. The controller device of claim 1 wherein the input receiver circuitry samples:
a first portion of the amount of data in response to a rising edge of an external clock; and
a second portion of the amount of data in response to a falling edge of the external clock signal.
I highlighted these portions because they are critical to the correct analysis of these claims.
Thus far the Examiner has said that the data sampling is not done “…on the transition of an external clock signal…” – but this statement reveals that he is not paying attention to the precise language of the claim. The author of the claim was extremely clever and careful, and used very broad language to denote a cause/effect relationship between the external clock signal and the data sampling. Notably, it says nothing about the sampling occurring on the transition; the claim simply states that the sampling occurs…. in response to the rising edge of the external clock signal. There is no question, from reviewing the timing diagrams, that this is exactly what is disclosed in the ‘020 specification. IMO, Rambus clearly points out (on page 14 of their brief) that indeed, the internal clock signal is used for the sampling as well, but that signal has its origin from the external clock signal. Thus the claim is clearly supported by the specification (even if it is only shown in the drawings and not mentioned well in the text description) since there is an explicit cause and effect relationship:
As seen here, the rising edge of external clock 53 allows for an input sample at time 127; the falling edge of this same clock allows for an input sample at time 125. Again the Rambus papers appear to have a solid rebuttal, but it takes some time and effort to get to the correct answer. If the Examiner considers the points noted above, it is my prediction that he will reverse his reasoning for this claim. The type of language used in the claim “… in response to…” is extremely broad and should be entitled to a very liberal interpretation to cover a number of scenarios, including the one shown in the specification.
As an easy analogy, this situation is no different than stacking a set of dominoes in close proximity, and then nudging the first one at the end of the line to cause them all to fall. While the last one falls because the next to last domino knocks it over, there is no question but that the first one set the action in motion. As in the clock example, the last one falls “…in response to” the first one being knocked over.
This particular point is extremely crucial to Rambus’ success in its ongoing battles. Look for this issue to be debated very heavily in the papers to come. It is my belief that the PTO should eventually come around to Rambus’ thinking on this, even if it takes some time. To a large extent their current thinking appears to be heavily colored based on their misreading of the claim to say that the data is sampled on the external clock transition, instead of simply in response to.
If Rambus succeeds - as I believe it will on this point as well - the written description score will stand 6 – 0, and more importantly, their claims on DDR will be brought back from the precipice and put back into play.