6
\$\begingroup\$

I have read that in vacuum tube computers, a tube would fail every couple of days. How did the technicians know that a tube had failed?

Wouldn't a broken tube silently cause the calculations to sometimes produce incorrect results? How would the technicians know that the results of running a program could be trusted?

\$\endgroup\$
6
  • 8
    \$\begingroup\$ By far, the most common failure is an open heater. Just look for the tube(s) that no longer light up! \$\endgroup\$
    – Dave Tweed
    Jul 19 at 4:03
  • \$\begingroup\$ @ David Tweed.Are the computer valves different? When looking at American Valve car radios a valve can be bad with heater fine .If heater is open then valve is bad but most bad valves have good heater.The valves that I deal with are 6.3 Volt indirectly heated with active material on the cathodes. \$\endgroup\$
    – Autistic
    Jul 19 at 6:12
  • 2
    \$\begingroup\$ ftp.arl.army.mil/~mike/comphist/48eniac-coding/sec5.html \$\endgroup\$ Jul 19 at 8:34
  • \$\begingroup\$ @BruceAbbott: I don't know whether Eniac had such a facility, but some systems allowed the supply voltage to be adjusted within a certain range. If a system ran a test correctly e.g. a supply voltage set to 90V and also with it set to 110V, that would imply that nothing in the system had degraded in such a way as to be unreliable at 100V. \$\endgroup\$
    – supercat
    Jul 19 at 17:08
  • \$\begingroup\$ @BruceAbbott Your link is dead. Could you please post a fresh one? \$\endgroup\$ Jul 21 at 5:25

2 Answers 2

9
\$\begingroup\$

Same way you trust any other instrument. Run pre and post calibration style checks and assume that things didn't magically fail and fix themselves in between.

The power of a computer is not in doing complex computation, but in doing lots of simple and verifiable computations that then produce the results of the complex one. You only have to run a handful of checks that require every tube/transistor to be operational. Doesn't cover cross-wiring cases, but you weigh that against how rare it would be, given people cared not to screw up back then...

Now, the practice is assuming some acceptable amount of screw up, throwing more data at it and praying something will catch bugs or they'll be forever buried in the noise level of big data.

\$\endgroup\$
2
  • \$\begingroup\$ Warning: rhetorical questions follow. Do you have a pacemaker? Have you ever been in an airplane? Did a nuclear power plant provide the electricity you used to post your answer? Not everything is big data. Not everywhere being a bit wrong sometimes is acceptable. Your view of the industry is too narrow. Up-voted anyway \$\endgroup\$ Jul 21 at 5:35
  • 1
    \$\begingroup\$ As rhetorical as they may be, both airplanes and nuclear powerplants use redundant systems as well as periodic re-calibration. Next time you spot the activation of a redundant system (or failure thereof), check whether someone knew or could have guessed about a failure from the re-calibration type test that preceeded it. \$\endgroup\$
    – Abel
    Jul 21 at 9:58
9
\$\begingroup\$

From the heat and current, aging occurs on gain and leakage.

By measuring the plate leakage current and mu that affects voltage gain they could test the loss of output swing and thus loss of function of an inverter Triode.

How long does it take for you to see if one tube is going or gone?

enter image description here

The 1st ENIAC has over 7000 tubes so if it replaced 3 tubes every day then that is an MTBF = operational hours ÷ # of failures = 24 * 7000/3 = 56,000 h which is the lifespan of a modern high-quality tri-phosphor 48" fluorescent tube running 24/7.

By the end of its operation in 1956, ENIAC contained 18,000 hi-rel vacuum tubes, 7,200 crystal diodes, 1,500 relays, 70,000 resistors, 10,000 capacitors, and approximately 5,000,000 hand-soldered joints. It weighed more than 30 short tons (27 t), was roughly 8 ft × 3 ft × 100 ft (2 m × 1 m × 30 m) in size, occupied 1,800 sq ft (170 m2) and consumed 150 kW of electricity.

In the 70's I had my fave Bogan stereo tube amp and I would go to the drug store every year or two to test the tubes and perhaps replace 1 if it was glowing blue plasma or weak.

\$\endgroup\$
2
  • 2
    \$\begingroup\$ What is a drug store doing testing tubes? That's like going to the drug store to get your photos developed. Completely anachronistic :P \$\endgroup\$
    – DKNguyen
    Jul 20 at 2:01
  • \$\begingroup\$ That drug store also had nails and hardware in the basement. Now they do makeup and perfume so alot has changed in 50 yrs. \$\endgroup\$ Jul 20 at 4:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.