More on trial
posted on
Sep 29, 2013 09:07PM
First, in my previous post, I said: There was a similar aged guy sitting at HTC table. I guess he may be Mr. Fish.”, I found out later my guess was wrong; he is one of the lawyers from HTC side.
Second, the formula TI expert used for calculating PPL output frequency -
DPLL output frequency = ckref (frequency of reference clock - the external clock) * N (a integer 1-32) / N (a integer in 1-4).
is just a mathematical formula, not actually happen on the chip, according to Dr. O. On the chip, there is a frequency divider that can derive a slower frequency from a high frequency generated from faster clock (ro), not another way around. No circuit can multiply frequency. When asked about frequency multiplier, Dr. O said it cannot be done on the chip. (He did not explain further why, but this maybe just a common sense if one think about it – if a clock tick 100k times a second, a divider can pick up every 100th of the tick (a tic toc) to produce a 1k/second clock. If a clock tick 1k times a second, to produce a 100K/per second clock, you need to break each tick to 100 slices and catch each slice to generate a 100K/per second clock. This is impossible to do, since you don't have 100 tic toc in a single tick.) So in actuality, it's another way around -
ckref (frequency of reference clock - the external clock) = r/o output frequency / N
He said there is no role for external crystal clock in generate frequency to clock cpu. If you disconnect the crystal clock, RO and cpu can still run.
Another thing Dr. O who corrected what TI expert said is that it's RO produce the clock signal, to say PLL produce the signal is misleading. It's ro directly connected to CPU, clock cpu and every thing else (memory, registers, etc.) on the chip. PLL is doing adjusting or control using the reference frequency of crystal to keep ro running in a range, just like a cruise control of the car (engin is the one who generate speed, not the cruise control who adjusting the speed.)
In the Varying together language, both TI expert and HTC lawyers tried to tie PVT (process, voltage, temperature) using “AND”. Dr. O immediately corrected them using the claim language of 336 – it is one or more, not requiring all of them at same time. Since process means the fab process when the chip is made, It is a fabrication related parameter (done when a chip is made), not a operation related parameter like voltage or temperature that vary at time of operation. Dr. O further illustrated the binning practice by CPU makers using picture slides. (HTC lawyer later said that binning was done by chip maker, not HTC when they use the chip. Seems a weak counter argument to me.) Binning is the result of Process variation, according to Dr. O.
Dr. O also clarify about the odd number of inverters used in RO, since there are RO used two rows of four inverters arranged in special way to produce similar oscillation as a odd numbered RO. He said even the # of inverters is not odd (4 * 2 = 8) the inversions is still a 5 stage inversions, not outside of 366 claim.
A sidebar event (it was pubic info now if one read the pacer doc posted) - When asked the quality of 336 patent, Dr. O said nearly all high speed microprocessor use 336 patent (not exact word). HTC lawyer objected, saying this has conflict with ITC ID. The objection was sustained after the sidebar event. My guess is that judge does not like to raise issue related to ITC ruling, even it was an ID and got appealed by our side.
Dr. O went detail about elements HTC infringe 336, 4 elements for claim 6 and 13 - cpu clock on the chip, varying together, on chip I/o interface, off chip second clock (note that for 6 and 13, cpu clock on the chip is not limited to RO), 5 elements for 7 and 14 – add async operation as the 5th (note that for 7 and 14, cpu clock on the chip is limited to RO).
Ok, the dilemma I have is that its very hard to cover everything in these four days. I thought about using bullet points to let the board know what I know, but summarize so much complex things in short words will distort the information. Also, since I'm a PTSC shareholder, my pro MMP stand is obvious and my wish to win this battle is the same as most of you. This bias may affect my posting. So do not let this kind of posts as your reason of action oneway or another. As Ron said, in a trial, you don't know which side will win until the verdict is announced. The good thing is that we are counting hours now instead weeks, months or years.
To cut things short, here is my impressions after Friday -
We heard damage experts of both side, our expert did a good analysis first. His final # is about $10 million, about 0.125% of HTC's 8 Billion revenue. He used different models or ways to calculate damages. He tried to counter HTC's expert's very low # ($475K) by underplay APPLE deal ($960K), saying it was special deal when TPL is in financial distress. (APPLE license was done 04/2010. Public disclosure of the license details can only be revealed 6 years after patents expiration (2021). (In my opinion, this deal will hunt us each time in court, since infringes will use this low ball # in the average license calculation to lower the final #.)
Our damage experts also said that his final # is conservative for two reasons, 1)Validity and infringement of patent not captured; 2) include lower rates because discounts that are not available to HTC.
HTC's damage experts disclosed some license paid: Apple 960K, PanTeck 460K, RIM 1.5M, Nokia 20M, Motorola 6.16M.
He used many average numbers, one of the numbers is license related to phone companies like above. The lowest number from him is 0.003% and the highest # is 0.238%. Obviously, he is not using average of these two numbers, he will put the relative revenue onto the weight.
He also argued that this trail is about just one patent, not entire MMP patents, to counter our expert's calculation (who put 336 as the dominate value of MMP) .
(My impression on damage is not as good as infringement part. On infringement, our lawyers and witnesses did good job. TI, Qcom and HTC's engineers did poor job to defend their non-infringement case. They avoid answer question directly; many times the Qcom guy go round about just try to avoid a direct answer; the HTC guy even got nailed by saying contradictory things between his previous testimony and Friday's testimony. On the other hand, on damage, our expert is confined by those low ball deals (Apple is the worst). I think their expert did explore the weakness of our license history to score some points. Another observation is that since damage is argued after the infringement towards the close, our side need to do more to score points towards the finish line.)
(So if I expect big numbers (or even reasonable #) from this trial, the chance is not great. The willfulness part, Head of Alliance did do a good job on exposing negotiation history with HTC. We saw those email exchanges before the trail. I hope juries will largely take his testimony to judge the willfulness.)
From questions asked by Juries, one can sense that they are sophisticated, at least some of them are. During HTC engineer' testimony, a technical schematic was show to jury with detailed PLL structure. After he finished, juries gave questions to judge. They asked very technical questions about how RO is connected to CPU? which component inside PLL provide clock signal to CPU? Is power supplied to RO the same as to CPU at start up? The clock rate of cpu at start up? Obviously, HTC side avoid answer them directly. Some of these questions can be answered by Dr. O's testimony. But this was good opportunity to nail them hard while they score points on damage calculations. Our side did not followthrough it well, I don't know if it is because we have limited time left, or if it is our reliance more on binning (process/fab parameter) for convincing jury, since binning is much more easy to illustrate and understand by a layman.
Mr. Fish's pre-recorded video (under 1 hour) will be viewed Monday. Final jury instruction likely by Monday evening.
On Monday, HTC will continue witness testimony. I guess they have a more time left than us.
Judge said he will give final jury instruction Monday evening.
(About Apple license, I like to ask lawyers of this board: it is plain and simple that TPL and Apple deal harmed all PTSC shareholders, the 50% holder of MMP patent. This trail revealed the under $960K number. HTC expert used this deal to lower damage calculated. Other companies will do the same in future case or deals. Based on Apple's revenue, if a reasonable license is $80 million, we are harmed a lot more than 79 million (80M – 1M) divided by two, since every infringer will use the low number to lower the reasonable royalty. This trial is a solid prove of that. It translate to 10 cents per share of PTSC stock. The current stock price. So is there anyway to we can go after Apple for this harm? Did Apple do it not in good faith or even possibly broke the law?)
(I hope we get the infringement and get a reasonable (lowered by the Apple deal) damage reward. For people (I was one before attending the trail) hoping to get a sizable one to release our long suffering seems unrealistic (need to also include the loss as one possibility, and a settlement as another possibility). If we win with disappoint amount, it's ok for me now if it is a few times bigger than Apple #. Just hope TPL will stop sweet deals from this point going forward. We need to build up reasonable damage in the future to compensate sweet deals of the past (Sony, Apple, etc).)
I hope above answered questions of some board members' private messages. Now lets counting the hours.
Good lock to all!