Fish Finders and Analog Front Ends
Have you ever seen a fish finder? Its kind of a crude SONAR you mount on a boat. A rotating disk triggers a sound pulse and the return pulse lights a light on the disk. The point on the disk's rotation where the light blinks tells you the depth of either the bottom or the fish.
I couldn't find it on the Internet shockingly enough, but for awhile one of the vendors (National, I think) produced an IC that was pretty much all you needed for a fish finder SONAR except for the transducer. You can only assume they could use more components on the IC die than you could use in a practical design and -- assuming they planned to sell volumes to multiple firms making fish finders -- that they could spend more money on development than you could.
This isn't really a new idea. There was a time when you had to struggle to get your own op amps working, or build A/D or D/A converters from discrete components. Now you just buy it in a package.
Naturally, there was a time when people build CPUs out of parts too but now only die hard hobbyists do that. And the trend there has clearly been to cram more and more functions on the chip. Memory, clock generators, I/O devices, and more have all migrated on the chip.
But analog interface components tend to still be building blocks. But recently several vendors have introduced AFE or Analog Front End ICs. These chips try to go further than just provide basic analog or digital conversion. They attempt to put all the features you need, often for a specific application, into a single chip. So, in theory, you hook your AFE to your CPU and that's all you need; program away.
Applications range from digital cameras, to power meters, to power line networking. Like any other integrated solution, you lose some flexibility, but you presumably get a better engineered circuit for a lot less money.
AFE's are especially interesting if you don't have the expertise to develop the circuit to start with. A lot of sensors and special applications have little quirks that aren't apparent. For example, the fish finder seems simple enough right? Kick a pulse into a transducer and then wait for a response back from the transducer. But if you ever really try doing that you find you need to kill the transducer right after you send the pulse or you keep hearing what you just sent immediately. Another twist that most systems do is to run at a relatively low receiver gain for awhile and then, if no pulse is heard, the gain is increased to hear the echo coming from (presumably) further away (and thus weaker). If you don't want to develop that expertise in house, its easier to just buy a ready-to-go circuit.
I wish I could find the data book that had that fish finder IC in it. I'm still looking. And, yes, TI makes a two chip solution but that's not the one I'm thinking about. Of course, it wasn't an AFE anyway since it had no digital interface. But I still think of that chip every time I see an AFE.
Are you using AFEs? Do you plan to? If not, what are you doing about the analog signal path in your designs? Leave a comment and share your thoughts.