Job Description

Party pooper.

It goes way back. Remember this story from the '80s? Originally it was supposed to shoot down attacking jet fighters.

That doesn't make sense for analog hardware testing. "Regression" in software has to do with revision management and quality control.

The ongoing revision of software to the degree that it occurs is rarely seen in the hardware (particularly the analog) world.

On further contemplation, I can see it as returning to your models to see where they don't match the physical realization... I do it all the time... refining models and macros.

But calling it "regression" makes me ill... irrespective of what Bloggs wants to claim ;-)

...Jim Thompson

Unfortunately it's not that strange. We have a complex analog front end designed without specification feeding complicated information into a reasonably complicated algorithm. The particular application is Flow Cytometry. Since the circuit was not built to a specification (just improved until the results agreed with a "gold standard'), any modification has to be fully qualified/regression tested over a large number animals and illness types. Our customers are intelligent and do cross checking. If the our new instruments started giving substantially different results we would have a problem. I have always recommended moving the A/D conversion "up stream" to eliminate a lot of specialized circuits, but there is a regression hurdle.

Ray