Smoke on the Horizon... oh s**t!

From: Tom Jennings <>
Date: Wed Feb 18 13:10:45 2004

On Tue, 2004-02-17 at 15:45, Mark Tapley wrote:

> >Seriously, I think you just have to change them often, eg. decadely.
> >(That's a perfectly cromulent word.)

> "Decadely" I accept, it makes sense by analogy with "weekly".
> But "cromulent"? Communication is not being served. Got a synonym I
> can look up?

Sheesh, don't all you people watch THE SIMPSONS? How do you make it
through the week?! cctalk content: it's > 10 yrs old now.

> >I think spacecraft programmers know about this sort of thing, but you
> >could probably design a power supply that presupposed possible failures
> >like this, but basically no one but us lunatics care about electronic
> >objects lasting more than five, never mind 10 or 20 years.
> It's not guaranteed, even there. There are spacecraft instruments in
> build with single capacitors in a circuit where fail-short means the
> instrument can't turn on.

(But I bet they derate that single cap :-)

> Same job could have been done with two
> capacitors in series. Then if either fails short, the circuit still
> works. (It might exceed its timing spec, but it'll work.) In fact I
> have caused that addition to be made in some instruments and power
> supplies, but the design engineers don't seem to learn easily. And
> there are some I didn't get to review until "too late" in the design
> cycle. Sigh.

Good trick. Design-for-failure isn't a popular subject in
cost-sensitive, quick-to-market business, but it's fascinating work.

I'm working with 5th-year EE students who don't even know how to make
real-world input protection networks, or do real bypassing, but boy can
they crank out SPICE simulations of CMOS gates. You should see their
solenoid driver designs -- >shudder<. Never heard of inductive power
resistors. Reading the fine print in capacitor specs? Fat chance. We're
all doomed.
Received on Wed Feb 18 2004 - 13:10:45 GMT

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:36:42 BST