The art direction and set design are truly wonderful, though: look how they’ve captured the essence of gentrifying Williamsburg in the mid-2000s in all its resplendent glory!
BRING BACK KOKIE’S OR WE’LL TURN THIS ENTIRE BLOCK INTO CONDOS you can fairly hear them say…
6800, actually. And it was designed for testing purposes.
That was the “official” HCF instruction; the CPU actually neither halted nor caught fire. The bug in the 68K I referred to caused it to do the first and, given a low enough 5V rail source resistance, to do the latter as well. Which is why it was such a problem at the time.
Other annoying features of easrly CPUs: the 9900 microprocessor inconveniently had FFFF as an instruction that affected registers, rather than being a no-op, so that for safety reasons in embedded systems all unused Eprom addresses had to be set to 0000 just in case of a bad jump. A surprising number of bugs were traced to this, including one which was my fault, I now reveal, and caused an expensive tank of fluorinert to boil dry in an experimental setup.
Ouuuuuuch! How did that happen, may I wonder?
As others said, I just couldn’t get into the first season of this at all. It just screamed of a panicked meeting at AMC where someone said, “Oh crap, Mad Men is ending. What do we do now? WHAT DO WE DO NOW? Wait, let’s do Mad Men in the 80s! With computers! Mad Men with computers!”
It was necessary to keep the fluorinert at a temperature of 170C +/- 0.025, which is fairly tight. To do this we had a big impeller and a heating element. The heating element was controlledby a PCM DC amp, and the feedback PID loop was driven by a microprocessor. Off the shelf PID controllers weren’t good enough so it was custom (one thing was that when a sample was inserted, a calculated power boost was added sufficient to compensate for the energy needed to warm the sample).
As I recall, after 30 years, there was at one point a call to a library function which was in Eprom. Unfortunately the call was to the location one before the function, which was FFFF. This caused a register to increment which held the most significant 16 bits of one of the control terms. As a result, power to the amplifier didn’t stabilise around the correct temperature.
A repetition of this expensive embarrassment was prevented with a 50c thermal trip which the hardware guys should have thought of in the first place.
Also, later on, I reviewed the theory of the whole setup and discovered that the bath didn’t need to be anything like that accurate; the sample could be tested at anything from 169-171C and a small compensation made, due to its characteristics.
This topic was automatically closed after 5 days. New replies are no longer allowed.
