In article om,
says...

I understand there are a gazillion systems, switches, sensors, valves
in them fancy rockets. But...

What percentage of scrubs are due to hardware problems and not software
problems ?

Seems to me like there is enough experience with "rocket science",
liquid fuels and very cold temperatures to be able to design hardware
that would be reliable for sufficient number of system activations to
last the lifetime of the rocket ? (testing during assembly, testing at
pad, and launch).

Helium leaks are not uncommon because it's quite difficult to contain
gases with such low atomic weights. H2 is even worse. Over the life of
the space shuttle, hydrogen leaks were fairly common and proved quite
difficult to completely prevent.

Seems to me like a lot of scrubs appear releated to sensor problems.
Since rockets spend far more time at the pad than actually flying,
shouldn't reliability of those systems be increased for operation at the
pad (such as changing temperature, metal contraction when cold fuel is
loaded etc) ?

Seems to me like engineers spend a lot of time ensuring components work
during flight, but less time to ensure they work on the pad before
launch. Or is that a false impression I have ?

I think that's largely a false impression. This is why "hot fire" tests
of stages are performed. Potential problems involving leaks and/or
sensors are likely to happen during this sort of test. But problems do
sneak through and when they're found, they are dealt with. In this
case, it's better to fix the helium leak on the ground than take the
chance that it will get worse and cause problems during flight.

Jeff
--
"the perennial claim that hypersonic airbreathing propulsion would
magically make space launch cheaper is nonsense -- LOX is much cheaper
than advanced airbreathing engines, and so are the tanks to put it in
and the extra thrust to carry it." - Henry Spencer