The Speed Of Light Is Actually Really Slow, And This Is A Pain In The Neck.

The speed of light is actually really slow, and this is a pain in the neck.

All you aerospace engineers and economists understand my pain, but I'm not talking about signal delays from Earth to the asteroid belt, or between Chicago and the New York stock exchange. There are local problems too!

I was once working on a project that involved a big honkin’ high voltage power supply.  It was powering a bunch of klystrons, devices that produce high-energy radio pulses.  In the event that something went wrong in the experiment, the power supply needed to shut down within 20 microseconds. Any longer than 20 and our very expensive klystrons would break, to the tune of a quarter million bucks a pop.

I'm simplifying things a lot, but here’s the basic timeline involved:

Something goes wrong in the experiment (the experiment was a fusion reactor btw, but that’s not relevant here).  The clock starts ticking down from 20 microseconds.

The fault detection circuits take 8 microseconds to realize something has gone wrong. 12 microseconds left.

The power supply does its emergency shutdown procedure, which takes ten microseconds. 2 microseconds left.

Everything is shut down and safe. We breath a sigh of relief, and replace about $1K worth of parts.

There was just one problem:  When I say “big honkin’ power supply,” I mean big. It was the size of two shipping containers stacked atop each other. The only place with enough room for it was about 1000 yards away.

Hm. 3 micro light-seconds.

So now we have to add an extra step to our timeline:

Something goes wrong. The clock starts ticking down at 20 microseconds.

The fault detection circuits trip. 12 microseconds left.

The fault signal travels via fiber optic to the power supply, 3 micro light-seconds away.  9 microseconds left.

The power supply takes 10 microseconds to shut down

Everything is finished… 1 microsecond too late.  We rend our clothing and spend a million dollars replacing fried klystrons.

The solution? Spend a metric buttload of cash on fancier fault detection circuits that will trip much much faster.

But we could have saved a LOT of money if the speed of light were just a tiny bit faster instead.