We've all encountered an analog IC that simply makes you stop and say “wow.” So how does an IC reach this status level?
Certainly, there's low price — but I don't really consider that the mark of a great IC, since it is not a technical virtue. After all, a vendor can decide to low-ball the price and make very little or no margin on it, or even chose to take a loss, perhaps to buy into a market (“We lose money on every one, but we plan to make it up in volume”).
Of course, sometimes the low price is the result of a radically new internal architecture or process enhancement, or even a production test technique that allows them to charge less and still make a decent margin — but I'll leave that angle for another time.
For me, there are three technical factors for which an analog IC can be especially noteworthy:
First, having a greatly improved and superior specification. Sometimes, a vendor dramatically improves a vital parameter, or manages to make a pretty good improvement in one spec yet without incurring the usual tradeoffs that such an improvement has traditionally required (for example, speed versus power). Or the part hits a “sweet spot” with just the right combination of specifications, and which wasn't previously achievable.
Second, providing an added function or feature which makes it much easier for the circuit designer. This is more than just packing more functional blocks onto the die. It means realizing that many designers are using the part to do “x” but they often also have to add an external something. By combining that extra feature with the basic function, the designer's project would be so much easier. It may be mostly just realizing that adding even the fairly mundane function to a high-performance part yields a case where 1 + 1 > 2.
Finally, there are the real head-turners, which do something that hasn't been done before. This doesn't necessarily mean integrating a lot of stuff onto an IC — it's when the IC design team uses the singular potential of monolithic technology to implement something truly innovative.
An example of simplicity is the AD590 monolithic temperature sensor from Analog Devices, introduced decades ago and still available for new designs. It exploits the temperature sensitivity of silicon to pass a current proportional to the temperature (1 μA/°K), and does so with a relatively simple IC comprised of just a few active devices. But what users really like is this: It's easy to use, small, rugged, and does what it is supposed to do, without complications (and yes, it is low cost).
A more recent example is the AS3935 Franklin Lightning Sensor IC from AMS (formerly Austriamicrosystems). When I first saw the announcement and background press package, I thought, “Whoa, this is something really different.” This small IC tackles a problem for which the only alternative solutions have been boards and boxes which may not even work as well, unless they are fairly complex and expensive instruments.
Have there been any ICs that really impressed you? In which one (or more) of the three above ways did they do so?