It's no secret that much of the innovation in ICs comes through the strategy of providing one or more of “faster, better (more integrated, more accurate, lower power), cheaper” virtues. The driver for this approach has been the semiconductor road map of Moore's Law, made real through new processes and geometries.
While there's nothing wrong with this strategy for success, there are also worthy innovation paths that are outside of this trio. Sometimes, solving a well-defined problem with a basic building block can be just as impressive, and perhaps even more profitable.
A recent product from Linear Technology Corp. clearly illustrates this situation. The LT3092 is “just” a two-terminal current source, period. It not only replaces tricky, lower-performance current-source designs based on FETs and resistors, but also can operate “in-line” without any connection to ground, a requirement in many current-loop applications. Even better, it knows nothing about your software, operating system, or processors.
I spoke with Robert Dobkin, LTC's Engineering Vice President and Chief Technical Officer and the IC's designer, and he noted he had been thinking about the need for this function and how to implement it for almost 30 years. What kept him from realizing it was, he said, a combination of lack of knowledge and understanding on his part, as well as some process limitations.
For example, one of his challenges was how to eliminate the need for external bypass capacitors to handle sudden load changes. Any such capacitors would have to be ground-referenced, but Dobkin wanted a true, floating current source which could be inserted in a current loop.
Although modest in function and scope, this two-terminal current source is a problem-solving building block for a wide variety of applications and circuit configurations. It doesn't force the designer to match the design to the IC; instead, it frees the designer to more easily use current-sources in a wide number of applications that either require such a source, or would benefit from one.
This is not Dobkin's first such innovation on a “small” scale with large implications. About a year ago, he developed a significant twist on the classic three-terminal, low-dropout regulator (LDO) by adding capabilities while simplifying design-in, an especially impressive feat considering that the LDO is among the most basic and simple-to-use components available. Dobkin said one of his other goals with both this current-source IC and that improved LDO was to develop products that would not incite a lot of calls from users needing help in getting them to work.
Devising a simple solution to such a fundamental building-block problem may not get the attention of the media as much as a faster processor, denser memory, or cheaper IC may get. Nor is it the kind of innovation that pops to the top when you ask engineers what they would like to see in the next IC: their answer is usually along the lines of faster, lower power, more integration, and, of course, cheaper.
But especially in this difficult period, sometimes a basic functional block, executed with a spin that adds real value to the circuit designer, may be a smart strategy for getting on the OEM's BOM as irreplaceable, and thus support long-term success. ♦