标签: kilby labs

The age of open-ended, "investigate what you want" charter characterized by Bell Labs is over, no doubt about it. We may miss their vaunted RD effort which spawned so many advances, and quite a few Nobel Prizes, but it's not coming back, so there's no point in lamenting it. The corporate "model" that supported Bell Labs, and others like it, is gone; too much has changed to allow it to remain in place. But maybe we shouldn't worry. There's plenty of RD being done, but it is being done in different ways and with different imperatives. Today's corporate labs are much more driven and focused on visible and viable goals, with shorter timelines, and more tangible outcomes. I saw this in a recent visit to the Kilby Labs at Texas Instruments (Dallas, TX) and a meeting with Venu Menon, Kilby's vice president and manager of the Analog Technology Development, and Kelly Krantz, lab manager. The operation is named for TI fellow Jack Kilby, who was a co-winner of the Nobel Prize in 2000, "for his part in the invention of the integrated circuit". The approach that Kilby Labs uses is very different than the open-ended Bell Labs approach. Anyone at TI can propose a project that may be worth exploring, by explaining why it may be worthwhile, what the outcome might be, and what the measure of success is. If the project is approved by the Kilby management committee, a small team (typically four people) is pulled together from various TI groups and business units to investigate it over a period of between 12 to 18 months. Except for lab administration, there is no permanent staff. Furthermore, the team may not even include the person who suggested the project, if he or she doesn't have a particular technical expertise that the project needs. The team members know they are on a special assignment, and will be going back to their home units once their project is over. The objective is to determine if the idea has a future from a technical and market perspective, and if so, what is needed to make it happen—and if TI can do it, or find partners who can help. In many ways, the projects are enhanced feasibility studies, but they go much further than just simulations and "paper" analysis; there is real hardware and software involved. If it looks promising, TI may then put additional resources into the appropriate business unit or product line to develop ICs (or other products) to make the lab project into reality. Because the Kilby Labs goals and investments are relatively modest in terms of people and time, there's an acceptance of failure. In fact, the people at Kilby told me they expect that only 20% of the projects will be deemed "successful" at their conclusion, and worthy of further investment by the business units—and that's OK. The other 80% will become part of the corporate learning process, and sometimes knowing where not to go is as important as knowing where to go. The Kilby Labs model is not right for everything, of course, and TI acknowledges this. That's why they (and other corporations) feel that longer-range projects that have more "R" than "D" are a better fit for universities, often working in conjunction with a corporate partner. Does the Kilby Labs approach work? It's too early to say, since the lab has been using this structure only since 2009. But it already played a role in one significant product, in May 2011: the TMP006, a passive infrared (IR) temperature sensor with on-chip MEMS thermopile sensor, signal-conditioning channel, 16bit analog/digital converter, local temperature and voltage references, and digital interface—all in a 1.6 × 1.6 mm monolithic device. Does our industry need more of this type of lab model, situated some between the "R" of research and the "D" of development? Or is it a cheat, an attempt to get something big, without too much up-front long-term commitment? Is it better take a lot of smaller, more manageable shots than a few big ones? Can we rely on universities, alone or with corporate involvement, to do more of the open-ended research? What do you think? I think these are all good questions—and that I should apply for my own grant, to study them further.

The era of open-ended, "investigate what you want" charter characterized by Bell Labs has ended, no doubt about it. We may miss their vaunted RD effort which spawned so many advances, and quite a few Nobel Prizes, but it's not coming back, so there's no point in lamenting it. The corporate "model" that supported Bell Labs, and others like it, is gone; too much has changed to allow it to remain in place. But maybe we shouldn't worry. There's plenty of RD being done, but it is being done in different ways and with different imperatives. Today's corporate labs are much more driven and focused on visible and viable goals, with shorter timelines, and more tangible outcomes. I saw this in a recent visit to the Kilby Labs at Texas Instruments (Dallas, TX) and a meeting with Venu Menon, Kilby's vice president and manager of the Analog Technology Development, and Kelly Krantz, lab manager. The operation is named for TI fellow Jack Kilby, who was a co-winner of the Nobel Prize in 2000, "for his part in the invention of the integrated circuit". The approach that Kilby Labs uses is very different than the open-ended Bell Labs approach. Anyone at TI can propose a project that may be worth exploring, by explaining why it may be worthwhile, what the outcome might be, and what the measure of success is. If the project is approved by the Kilby management committee, a small team (typically four people) is pulled together from various TI groups and business units to investigate it over a period of between 12 to 18 months. Except for lab administration, there is no permanent staff. Furthermore, the team may not even include the person who suggested the project, if he or she doesn't have a particular technical expertise that the project needs. The team members know they are on a special assignment, and will be going back to their home units once their project is over. The objective is to determine if the idea has a future from a technical and market perspective, and if so, what is needed to make it happen—and if TI can do it, or find partners who can help. In many ways, the projects are enhanced feasibility studies, but they go much further than just simulations and "paper" analysis; there is real hardware and software involved. If it looks promising, TI may then put additional resources into the appropriate business unit or product line to develop ICs (or other products) to make the lab project into reality. Because the Kilby Labs goals and investments are relatively modest in terms of people and time, there's an acceptance of failure. In fact, the people at Kilby told me they expect that only 20% of the projects will be deemed "successful" at their conclusion, and worthy of further investment by the business units—and that's OK. The other 80% will become part of the corporate learning process, and sometimes knowing where not to go is as important as knowing where to go. The Kilby Labs model is not right for everything, of course, and TI acknowledges this. That's why they (and other corporations) feel that longer-range projects that have more "R" than "D" are a better fit for universities, often working in conjunction with a corporate partner. Does the Kilby Labs approach work? It's too early to say, since the lab has been using this structure only since 2009. But it already played a role in one significant product, in May 2011: the TMP006, a passive infrared (IR) temperature sensor with on-chip MEMS thermopile sensor, signal-conditioning channel, 16bit analog/digital converter, local temperature and voltage references, and digital interface—all in a 1.6 × 1.6 mm monolithic device. Does our industry need more of this type of lab model, situated some between the "R" of research and the "D" of development? Or is it a cheat, an attempt to get something big, without too much up-front long-term commitment? Is it better take a lot of smaller, more manageable shots than a few big ones? Can we rely on universities, alone or with corporate involvement, to do more of the open-ended research? What do you think? I think these are all good questions—and that I should apply for my own grant, to study them further.