My Electronics Blog Article: Why I Loathe TTL | Hacker News Reader

My Electronics Blog Article: Why I Loathe TTL

mecrisp-stellaris-folkdoc.sourceforge.io

12 pointsoldguy1013 comments

Author had bad experience with TTL (the integrated circuit logic family introduced in 1965) - their company was trying to design industrial equipment using TTL-based controllers, and apparently none of their engineers managed to properly protect the circuits from the electrical interference.

Sorry, I'd blame the combination of harsh industrial environments and engineers at author's company who did not quite know how to manage those. Because there has been quite a number of very successful devices built from 7400 series TTL chips - for example famous Xerox Alto [0] is full of them.

(But yeah, 7400 are not a good chips by modern standards. Large static current consumption, and they will happily emit pulses on power bus while switching. But CD4001 is 60nS propagation, while 74S00 is 2nS propagation, so if you want high speed, you don't really have a choice. Just put lots of capacitors - for max robustness, you want one capacitor per each chip)

[0] https://bitsavers.trailing-edge.com/pdf/xerox/alto/schematic...

The Xerox Alto works great sitting in a nice air conditioned Silicon Valley office. Put one in a steel mill with motors and welders wreaking havoc on the electrical supply, and it wouldn't last a few minutes.

Well yeah, 7400 needs clean power, so you need to have a good power supply. And if you are in a steel mill, you probably add an extra input capacitor or ten, few chokes, and even shield the whole thing. This is somewhat complex, but not impossible. After all, even the worst "super fast runt pulse" won't survive a simple L-C filter, as long as the filter is done correctly (say with bulkhead-type capacitor and metal shielding).

Why then author's company failed to do this? I don't know.. Maybe they didn't have anyone with the right kind of knowledge, maybe they were cash strapped and weren't ready to pay for huge metal boxes and hundreds of EMI filters. (The latter does actually sound plausible - after all the engineer was fired for messing up a single revision of 12"x12" board.. and a good oscilloscope that can see the high-frequency pulses is much much more expensive that such board.)

An example is somewhat famous AN/UGC-74 terminal, which is built on 5400 series (a higher cost version of 7400). I could not find a lot of good pictures, but there are some at [0]. Note:

- Electronics are in many individual sealed metal boxes

- One separate shielded section just for input power filtering

- Dozens of power supply filtering capacitors (small black squares, the "cpu board" seems to have ~1 per chip)

- Weights 100 pounds from all that metal, and not cheap!

I bet that one would work in the steel mill just fine.

[0] http://www.vintagevolts.com/the-an-ugc-74/

Reply on news.ycombinator.com