I view this as a very good announcement. Here are my thoughts for what they are worth:
Firstly the announcement is exactly in line with the last announcement's timing, so at least they are keeping us informed a bit better than previously.
"The Company advises that the vendor has supplied the initial wafer from the batch that has undergone various optimisation techniques in the final steps of fabrication. The vendor will continue to supply the balance of the wafers to the Company in regular intervals to allow for measurement and analysis of pixel yield and performance under “real playing” conditions. This will enable the vendor to properly assess the relative contribution each process adjustment has on the overall fabrication process."
Firstly I do not imagine this initial wafer is literally #1 off the production line. I expect that TJ has been producing a stream of replicates from an adjusted process. The wafer received by AP is one that they consider representative of the stream. I assume that the production process is amenable to a number of potentially helpful tweaks that will affect particular structures in particular ways. There will also be some interaction between each of these tweaks, some of which will offset or work against the others. That makes it important to see how to apply one or more of them beneficially. Salt and soy, chilli and fish sauce, if you will.
Each tweak will affect more than just yield specifically. The net result of optimisation of a process will result in shifts from say 85% yield through 87%, 83%, 89%, 88%, 94%, 97%, 95%, 98% etc. Some increases, some reversals. Some for instance will allow a reduction in process time by a few percent, others increase it by 10%, but with a higher yield. Which do you choose?
In this I completely agree with "Where am I", "Sins" and others that this is all understandable stuff.
I have no doubt that whatever they are doing, the commentary from Fred has attempted to cover it under a generalised statement that tells the truth without giving any details they do not want published, or which are just too detailed to outline briefly.
The new comment about "real playing" conditions is, I think, very promising! I also think that this means playing sound - e.g. audio signals, maybe even music! Unlike Vlad (who must have worn out quite a few shovels digging holes in his basement floor by now) I agree that this implies not only that the integration process is still considered to be unlikely to give problems, but also that it is practicable in short order, and can be done on small numbers of singulated devices as required. Clearly they are not expecting to send a wafer off and wait 6 weeks for a finished set of 100 chips to come back so one can be measured. I am assuming - and I certainly hope - that they plan to turn wafers into chips as part of the optimisation process so that process can occur promptly. Otherwise we are likely talking weeks between optimisation steps.
I am currently working on developing new methodology for an analysis in the chemistry lab I consult for. I have been working on it for maybe a year now. There has been a lot of frustration and reversals as I "circle the mountain" to find the best way to the top, i.e. to get the highest sensitivity and best reproducibility etc. In other words I am going through exactly the same kind of process I describe above. But just over the past couple of weeks I have crossed a few significant milestones that allow me to finalise a few details of the procedure that I am confident will work and will form a part of the final method. They are "set in concrete", as you might say. I can see parallels in the announcements AP has been issuing over the past year that are exactly the kinds of things I have been saying to my client. So I am getting increasingly confident that the same process will lead to the same results at AP that I expect to deliver in a similar time frame.
(20min delay)
