OPTIONS TRADER

You can see from even this limited description of what a filter does that there is a virtually unlimited number of combinations of passband, stopband, attenuation, ripple, phase delay, and so on. Making the situation more complicated, there are different implementation options, many with overlapping capabilities, to choose from.

As their name implies, the simplest passive filters can be constructed from a mere resistor and capacitor, with perhaps an inductor thrown in for good measure. The obvious advantage is low cost. But, when it comes to filters, you generally don’t get what you don’t pay for, and a simple RC filter is no exception.

Moving up the ladder, there’s the active filter, which relies on our old friend the op-amp. A single op-amp and few Rs and Cs can do a pretty good job. And stringing them together for even better performance is an option. Just remember that adding more components is a primrose path upon which you might twist your ankle dealing with noise pickup and component tolerance and drift issues. Manufacturers have responded with integrated filter chips that are easier to design in and hide messy details under the hood.

Note that all of these solutions perform their filtering function in the analog domain (i.e., prior to digitizing by the ADC). The other option is to filter after the signal has been converted to ones and zeros. Enter digital signal processing (DSP), which offers a number of rather compelling advantages.

First, the DSP approach completely eliminates the aforementioned tolerance, drift, and noise issues that plague analog components. A 1 is a 1 and a 0 is a 0, and they stay that way no matter what. That means DSP solutions are inherently more producible and repeatable than their analog counterparts.

Second, a DSP (i.e., a computer) can perform signal-processing functions of virtually unlimited complexity. Yes, the top speed of the sampled time approach may be limited compared to a continuous time analog solution, but thanks to Moore’s law, the gap is always shrinking.

Finally, digital solutions are easy to design, debug, and reconfigure. A couple of mouse clicks and a quick download are much easier than tweaking a rat’s nest of analog parts, or worse, having to spin a new PCB for a design that has already gone into production. The ability to dynamically reprogram the DSP in-system offers a degree of adaptive flexibility that’s difficult to obtain with an analog solution.

It isn’t my intention to declare a winner in the analog-versus-digital signal-processing debate. As far as I’m concerned, both have their place—they always have, and they always will. The point is to help focus the discussion by recognizing that the QF4A512 is a digital solution, a mini-me DSP if you will.