On Wed, 1 Oct 2008 10:35:21 +0100, "N_Cook" wrote:

As far as heaters are concerned. I went to a lecture by Tony Sale of
Colossus rebuild fame
http://www.codesandciphers.org.uk/lorenz/mk2.htm
About 2,500 valves in this computer.
They power up Colossus each day via a motorised variac.

How many valve failures a year, due to failed heaters in those 2,500, would
you expect ?
Please reply here

http://www.mosweb.com/knowledgebase/tp5.htm
http://www.john-a-harper.com/FilamentHeating
About 100,000 hrs lifetime. No clue if that's mean, median, or
average. Ignoring the bell curve and assuming burned-in tubes without
infant mortality, that would be one failure every 40 hours for 2,500
tubes. With 8760 hrs/year, that's 219 failures per year.

However, that's not really true because the failures are not uniform.
With new tubes, there's a peak for early failures (infant mortality),
a long time with a very low failure rate, and then another big peak at
end-o-life, where filament thinning turns them into fuses.

I recall stats from the original Eniac, that was initially
experiencing a tube failure every 30 minutes of operation. Someone
finally decided to reduce the filament voltage from 6.3VAC to about
5.9VAC(?), which extended the lifetime dramatically by reducing metal
migration/thinning. Reduced filament voltage and temperature are not
a great idea for power tubes, but are well tolerated by digital tube
(flip-flop) circuits. I suspect Colossus uses a similar trick in
addition to inrush current control.

Answer tomorrow

Predicted or actual tube life?

--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558