Recently I have been thinking more of PSOC and Anadigm. Both offer programmable analog components on a chip. At first look they are really cool technology. You can drag and drop program interesting analog circuitry. But recently I have been thinking about how I would use them. I am now confused. Let me explain why.
Typical analog components this might replace are interfaces to sensors. We might need to measure voltages, currents, temperatures, pressures... We might need to stimulate inputs. What we do now is to process these with discrete analog components. Op amps, comparators, resistor divider and low pass filters. So my first reaction is part of this decision is more of a management decision rather then a technical one. For instance it just may be more convenient for us to do it this way rather then some more highly integrated choice. The choices may be available manpower, observable testable points, proven circuitry. On the other hand, I am always willing to at least consider something new, and maybe, if it makes sense, try something new.
So now I come to my other epiphany. How many of you know the bilinear transform? It says s=2/T(z+1)/(z-1). There are many transforms that relate the s domain to the z domain (don't tune me out, I will explain this all). The S domain is the analog domain. The Z domain is the digital domain.
Now back to English. With this simple equation I can realize all my analog circuitry in a digital processor! Why do I need analog at all? What does PSOC buy me that I can't already do with with a DSP and that equation? Perhaps the drag and drop nature of the PSOC tool makes it easy to realize these functions then the math involved with implementing a digital filter using the bilinear transform. Well if the advantage of PSOC is that its easier for an engineer to implement what about just using op amps, comparators and capacitors. Isn't that easier still? Wouldn't this be a tool issue?
I think I need to evaluate PSOC on some specific circuit or goal. I am just not sure.