Tube filiment voltage ???

[Magnus Danielson]

From: KA4HJH <ka4hjh at gte.net>
Subject: Re: Tube filiment voltage ???
Date: Mon, 3 Jan 2000 01:47:51 -0500

> >      I remember (well I read that... I'm not THAT old...) that ENIAC
> >used a reduced filiment voltage to prolong the life of the tubes so that
> >the MTBF would be acceptable.
> >Not that the tubes were unreliable... but with that many tubes in one
> >place its a done deal that one is (statistically) sure to die at any
> >time.
> >
> >    Other hand I've heard it said that this was done just to reduce HEAT.
> >Thousands of tubes would do that also...
> 
> I just had to look this up.
> 
> 18,000 tubes. Weighed 30 tons. Cost $486,840 in 1946. Of the 1847.8 
> hours spent on "unscheduled engineering", 90% of that time was spent 
> finding and replacing bad tubes. 19,000 were replaced in 1952 alone. 
> I guess that HAD to be acceptable.

As a historic note, back in Germany Konrad Suze did things in relays at this
time and mainly because he considered tubes too unreliable at the time. He
later turned to tubes and later transistors (like in the Z23).

The probability of a single device to fail may be low, but the probability
rates (usually measured in FITs) add for each component/connection you add so
if you have 18000 tubes, the tube's probability rate is multiplied by 18000
and this would take the average failure of a single tube to be a real
contributor to the failure of the machine. A average lifetime of tubes of
say 36000 hours would only give 2 hours of average uptime of the machine, so
the trouble is real.

Note that while adding components to a cursuit may decrease its lifetime, you
may actually create an environment which makes critical components fail less
and thus addition of components may increase you lifetime. This fact is often
not modeled in normal type of analysis.

So, saying this one has to recall the number of tubes and how that relates to
tube constructions of usually used in audio. Many amps are in the range of
4-10 tubes and has much longer lifetime than 2-10 hours.

Of interest here is naturally the failure mechanisms of tubes, by what
mechanisms will tubes wear out, what accelerate/retardate these mechanisms etc.
As being a novice to tubes at least I would love to hear about the experience
that exist on this list on these issues.

Especially:

What failures do I get?
How does the wearing mechanisms work?
What can I do to reduce the failure probability?
How can I measure the condition of a tube?

There are naturally related issues on caps etc. Electrolytic caps are also one
of the things I'd be looking at.

So, tube-people, you know who you are... enligthen me!

Cheers,
Magnus