Let’s continue with another story of Ben and the CADHouse who designed his Printed Circuit Boards (PCBs).
The CADHouse reported that they had a new problem that had stopped all layouts on their primary minicomputer based Computer Aided Design (CAD) system.
The minicomputer included a device called a digitizer. This was a really early version of a computer mouse or maybe a little more like one of the later “digitizer tablets” that let you enter precise physical locations by moving a puck around on a special surface. This particular digitizer required no special surface, but the movable puck was attached to two mechanical arms that translated X and Y position into digital values the computer could read.
The problem the CADHouse team reported was that if they moved the puck any direction, the position of the cursor always moved along a diagonal line. The digitizer interface was a single PCB in the minicomputer, and they had documentation for the computer bus interface. Unfortunately, they did not have complete documentation for the interface board.
The CADHouse manager was planning to travel near Ben’s company, so he said he could bring the interface board with him for Ben to study.
While waiting for the board, he thought about the description of the problem. A “diagonal” line sounded like a line with a 45 degree slope. There was something magical in that description. Suddenly, it hit him. A 45 degree sloped line means that the values of X and Y are the same. Something in that interface board was causing it to read the X value when it wanted the Y value (or maybe it was the other way around). But Ben knew he needed to look for some place on the interface PCB where trying to read the X value could give you the Y value; or trying to read the Y value could give you the X value.
When the CADHouse manager arrived, Ben and his team nearly tore the PCB from his hands. They found some tristate buffers connected to the computer data bus and near those, they found some 74LS157 quad 2-input multiplexers. Aha! The multiplexers were an obvious place that the X and Y values could come together before being driven onto the computer data bus through the tristate buffers. It seemed unlikely that multiple bits might be shorted together in logical order, so they focused on the multiplexer SELECT pin. An ohmmeter revealed that the SELECT pins of the multiplexers were all showing a short-to-ground.
Ben traced the connection to the SELECT pin across the PCB and studied it under the brightest light and magnifier he had available. The trace appeared to run only across the component side of the PCB. There were no obvious shorts anywhere to be found. The lowest resistance value between the SELECT trace and ground appeared to be slightly closer to where the trace ran under an Integrated Circuit (IC) in a Dual Inline Package (DIP).
A replacement PCB would take several days and cost some serious money to obtain. At this point, Ben and his team needed to have confidence in their skills. They de-soldered the apparently bad IC. The short remained between the SELECT trace and ground. They again studied the PCB, now that the IC was not covering that portion of the top of the PCB. Under bright light and magnification they could see a slight discoloration—really barely a shadow—between the SELECT trace and a nearby ground pin. An Exacto-brand razor-knife with a very sharp blade let them clear that shadow. The ohmmeter no longer read a short-to-ground. The original device was re-soldered to the board and the CADHouse manager returned to his business.
A short while later (that felt like forever to Ben) the word came in: the digitizer was happily moving the cursor to all locations and not locked to a 45 degree line.
Once again, if Ben could see it, he could fix it. In this case, the repair was accomplished because Ben could envision in his mind that a diagonal line implied X=Y. That led to the step-by-step analysis of where in the circuit such a value replacement could happen. But the team also needed to be able to physically see a microscopic short circuit on the PCB. And they needed the hand skills to successfully remove and replace the IC device.
Ben had previously read about and also had encountered “dendritic growths” on PCBs caused by minor contaminants and moisture. These defects allow a PCB to leave a factory in working condition and then to fail later in the field.