Replies

That was indeed an awesome project. It seems you had most of the kinds of stuff I've done all in a single system. It sure looks like that rail car system was a wonderful project. You put everything in there everything but the kitchen sink. Few people know that these things are "labors of love".

I used a PC104 project on a battery change balancer for Ma Bell. It is a delightful package. We went with VRTX on that project.

I know how you feel about project cancellation. I did a life safety system for the East Area Rocket Engine Test Facility in Huntsville for the NASA boys. We hooked to the countdown clock, and made an enormous field of sensors and actuators for doors, gates, railroad tracks, infrared sensors. We could detect cars, people, trains and control gate crossings, locks, sirens, horns, and gobs of warning lights. The network was carried on RS485, custom message protocol. The processor was a 68040 running on a multi-bus platform. Runtime was VRTX, developed under Idris, a UNIX clone that predated Linux. It even had sectionalized voice announcements. The labor union forced them to tear it all out so that a glob of signalmen and operators could keep their jobs. This was somewhere around 1984.

I remember one NASA guy told us, "If you ever hear that klaxon when you are down on a test stand, it means you are about to die.

I never saw a PIC die on the job. AVR devices were not nearly as rugged.

That safety of life project sounds pretty interesting. Lots of sensors. I never had a project that used VRTX, but some of my co-workers did. While I did a fair amount of embedded systems and UNIX kernel work for multiple manufacturers, that wasn't really common at my place of employment. During my years at PacBell, I taught embedded systems in the evening. 6800 and 8085 trainers were available for my students. On weekends, I was building code blue/nurse call systems on 6805 platforms. The income derived from my embedded work was sufficient to finance my fixed wing, private pilot license.

The PIC processors were very stable and the toolchain including the C compiler was well supported. I still wrote a fair bit of assembler where I had time critical interrupt service routines. Most of those were signal sampling ahead of doing some DSP to assess ride quality. The next gen hardware that never happened was intended to have a hardware ADC with 48 KSPS sampling. It would have saved significant expense over the Diamond Systems PC104 ADC.

Today is my last day of "work" with a charge number for my labor. I'll be retiring on June 8th after 33 1/2 years. I'll miss the work.