Issue link: https://iconnect007.uberflip.com/i/1213413
60 PCB007 MAGAZINE I FEBRUARY 2020 digitized, the result would be plotted on white paper/mylar and then overlaid on the outer layer film to check for accuracy and/or missing holes. When approved, the drill file would be punched to tape. It would then be read into the drill machine, and the plate would be drilled. These fixtures were typically 1" tall and had a reusable 100 mil base plate. Once drilled, the plate would again be checked for accuracy, and along with stand- offs, it would be assembled. Once complete, the fixture would be placed on the fixture tester. A single board from the lot would be "learned" by the machine. Once learned, sub- sequent boards would be tested against the learn. If the majority of the lot "passed," the learn would be considered "gold" and could be saved for future lots. However, as one would expect, there would be times when a faulty board was learned, and the subsequent boards would all have what appeared to be the same "open" or "short" as compared to the learn. In these cases, the learn would have to be scrapped and a new board selected from the lot to be learned as the new master. This was known as the "learn-com- parison or self-learn" methodology. The risk was apparent. All one could certify was that all boards compared the same. It did not mean they were correct. If all boards had the same repeating defect from manufacturing, the self- learn test would not capture the defect. Expen- sive returns due to electrical defects were not uncommon during this era. Technology began to advance relatively quick- ly in the late '80s to the turn of the decade. All of a sudden, we had this thing called "Win- dows," which, before then, was just something to look out from. The floppy disk was now be- ing replaced by hard drives. This was great! You could purchase a 20 megabyte MFM hard drive for about $1,000 and store hundreds of files. Now, since we could save files without draw- ers of floppy disks, we could move them elec- tronically. Well, not so fast. There wasn't much yet regarding networks. Novell had a LAN topography working at the time, but this required add-on network interface cards (NICs). CAT5 didn't exist yet, so computers were linked by coaxial cables. It was cumber- some and expensive. One nice thing to come along is that we didn't have to "sneaker net" files from the customer any longer. Modems became the new thing, and we could use the phone lines to transfer data. You were really in the know when you had your 1200 baud modem online and auto- mated software running as a 24-hour data ser- vice. The 2400, 9600, 14400, and 19200 mo- dems came later, which made life easier. Time warp to 2020, ask anyone of a younger generation, and they will give you a blank stare regarding any of the technology I just men- tioned. Today, we can do pretty much anything from the comfort of our couch, I mean, office chair. To put it in perspective, remember the IBM PC? It had a 4.77-MHz processor and 64K of RAM if you were lucky. You were smoking if you had 128K or the unbelievable 512K. The Apollo 13 mission utilized many computers at mission control. These took up full rooms. Did you know that the computing power of the Apple 4S iPhone could have managed the en- tire Apollo 13 mission? That's mind-blowing indeed. Today, you have to go to eBay to find retro floppy disks for your antique if, like me, you have a place in your heart for these old ma- chines. We are in the mainstream of the digi- tal age. "Plugged in" is just the way of life. We are plugged in at home and work, and even while playing. Due to the digital age, we have automation that was just unheard of 30 years ago. From 1986 to the present day, electrical test- ing has made momentous advances. From the self-learn of the '80s to today's automated fly- ing probe and fixture testers, the escape risks have gone to basically zero. These advances include the ability to perform buried passive Back then, it was all self-learn and compare.