In the mid-90s, I was working for an OEM of large industrial machinery. I was about nine months out of engineering school and had just completed my first major project. This particular project included the installation of the first PC-based HMI that my company (and the customer) had ever used. This HMI was a very complex system for its day.
The system also had extreme environmental challenges due to the shock, vibration, and dirt. We ended up with the Allen-Bradley T60 computer for hardware, which was about the only system on the market that would survive more than a week. I was also pushing the envelope of DOS and the available hardware, so much so that I needed a third-party memory manager to manually allocate each program to make it work.
One of the key features of our new PC-based HMI was that the customer could now extract fault information from the machine to analyze it in the office, and even store long fault and maintenance histories. Due to the equipment’s location, network cabling to the office was not possible and wireless connectivity was still a number of years off. However, "Sneaker-Net" (or in this case, "Steel-Toe-Boot-Net") was a breakthrough solution. The ability to save the fault data to a floppy disk and load it back to the computer in the office was a major development for our customers.
After a lengthy onsite commissioning process, I returned to the office and everything seemed to be going well for the customer. But that changed. I started getting phone calls from the electrical maintenance supervisor saying that the information downloading process was no longer working. This started an in-depth troubleshooting process. Was the raw data good on the HMI? Yes, shutting down the graphics still allowed the raw data to be read on the hard drive.
Was it copying correctly? Yes, the files could be reread after the download. It appeared to work every time, but the files were totally unreadable in the maintenance office. We even swapped out the floppy drives to see if the vibration on the machine misaligned the drive heads.
After going back and forth like this for some time with no solution, the only thing left was to walk through the process and see if there was anything anywhere that could be corrupting the data. I asked the electrical supervisor to tell me step-by-step what he did. He verified a freshly formatted disk -- good. He put the disk in the HMI drive -- good. He went through the proper download process -- good. He waited for the drive light to go out -- good. He removed the disk and put it in his shirt pocket -- oops.