Test Talk
|
Each issue of the DFT Newsletter brings another hilarious look at DFT by Alfred Crouch, author of the renowned book "DFT for Digital IC's and Embedded Core Systems." October 2007 The Design Princess and the DFM Pea… Hello there, Noah Bawdy here again. I’m not going to regale you with stories about my boss and his late-night TV management initiatives, but I have a story about my daughter and my grandson (which oddly enough, relates to my job). As most of you should know, when children reach a certain age, their mothers like to sit down and start reading them fairy stories. What I didn’t realize was that daughters that are going through this phase view both the child and their parents in the same light. In other words, my daughter treats both my grandson and me, the identified grandpa in question, the same way (well, except for that diaper thing…she expects me to change my own diaper). This means that I have had to re-learn baby-talk just to understand what is going on (it is a much more sophisticated language than I remember). I misapplied the verb goo-goo the other day and had to sit in the corner on a time-out for an hour. Anyway, the story this week was my daughter’s homemade and personalized variation on the Princess and the Pea. Given the passion she threw into the story, I couldn’t tell whether the prince represented my grandson and the story was for him, or the princess was supposed to represent my daughter and the story was about her. This is a story whereby a young handsome (and successful in any industry but engineering) prince grows up and brings what looks like a common girl (on the surface) home to the royal family. The royal family, of course, doesn’t want anything to do with the girl and they tell the prince he can’t marry her because she isn’t royalty. So, they propose to give her a royalty test to prove that she is a commoner. They put a single (soft and squishy) pea under a stack of very expensive silk mattresses to see if she had the sensitivity and attitude of a princess. The thought process from the royal family is that the girl is a commoner and is accustomed to sleeping on a wad of straw stuffed into a burlap sack…she surely won’t find fault with sleeping on an expensive silk mattress with a squishy pea under it. She should come down happy and smiling from a good night's sleep. The next morning, the girl awakes and comes down to breakfast clutching her back and moaning and groaning about what a horrid mattress. The royal family is shocked and thinks maybe that someone tipped her off, so they start asking questions to see if they can figure out the exact nature of what the girl is complaining about. So the queen, having the current most-sensitive back in the castle, asks just where in the mattress the grumpy princess found fault and what the nature of the “bump” was. The princess states that there was a squishy boulder under her back in just the place where the pea was…but that wasn’t the main problem. The main problem was that the fine silk stitching on the mattress wasn’t straight and consistent – proving that she had a finer royalty metric by the queen’s self-imposed metric than even the queen had. During the whole story, my grandson kept whispering in my daughter’s ear and related the whole story as it unfolded to his day-care class. He would whisper something un-intelligable in my daughter’s ear and my daughter would say, “no, I don’t think little Brittany is a princess.”; a few moments later, another whisper and, “I’m sure Craig didn’t bite her because he doesn’t like peas.”; then another whisper and, “well, yes, I would paint her green too if she did that.”; and eventually, “yes, yes, naptime is on reedy mats, not silk mattresses.” Of course, I could glean a little information from this conversation, but mostly I was left with the feeling of being an outsider. This whole story and the interaction between my daughter and my grandson put me in mind of a meeting I had had with the yield folks in the FAB. A group of design folks went to talk to the FAB folks about the new DFM issues, the alarming drop in our yield, and what needed to be done coupled with what information needed to be shared. We gave a presentation that had pictures of gates and nets and had models, timing analysis results, simulation results, and so on. When we were done, one of the FAB folks walked up to me and said, “that was very nice, colorful slides, about the right length, um, what’s a netlist? You see, we care about X-Y locations, metal layers, mask layers, and process steps – we don’t have any idea about this other stuff – we would probably have to hire an interpreter to explain it to us. And, from our fabless customers, they go through great lengths to make sure we don’t ever see this stuff – something about proprietary interest.” I might as well have been talking baby talk to them (goo-goo dada DFM). After thinking about my daughter’s story for a while, I realized that the pea represented the old defects in the Deep Submicron Technologies: particulate or dirt. I also realized that the very fine, almost invisible, yet non-consistent, stitching on the silk mattresses represented the new DFM based defectivity in the sub-100 nanometer Technologies: lithographic, planarization, and chemical-mechanical-polishing problems that come from using certain standard cells, or not understanding the layout structures that are non-compatible with processing limitations. So, I’m going to go back in there and try to have a meeting more like an interaction with my daughter and grandson. Now, if I can just find a princess that will sleep on a tray of 65nm low-power parts… |