Should an audio system be left on at all times?

semiconductors, resistors, capacitors etc don't have MTBF (not in our lifetime). They operate practically forever (except electrolytic caps).

Not true at all, unless you are referring to a single part operating by itself, which you are not. The combined mtbf of the hundreds or thousands of semiconductor and passive devices (even excluding electrolytic capacitors) in a typical sophisticated audio system will be measured in decades at most, and quite conceivably in years (less than one decade). That is especially true when you consider the sharp increase in failure rate that occurs with older components, according to accepted reliability models, and is even more true when you consider that many amplifier designs intentionally run semiconductors at high junction temperatures in the interests of improving the sound quality.

Please take a quick look at some of the 205 pages of MIL-HDBK-217F, linked to below, which is the guideline document for reliability calculations for military systems. Note that it addresses just about every conceivable type of electronic component. And keep in mind also that these are typically components that are manufactured, burned in, tested, and screened vastly more rigorously than anything in most consumer audio systems.

http://assist.daps.dla.mil/docimages/A/0000/0005/3939/000000041349_000000020839_BTJRBUHXWM.PDF?CFID=18958726&CFTOKEN=fdbcdd5971174382-FEB22389-1372-548A-D368EF4CD631E514&jsessionid=063010dbe387801aa415

Regards,
-- Al

Kijanki
All 'lectronics is subject to failure. The metal interconnects used in ICs and discrete devices 'moves' under the pressure of the flow of electricity. The break will usually occur when the line goes over an edge, where the metal is thinnest and current density is highest.
When we (my company) introduces a part, especially a new technology/revision, part of the specification includes Lifetime. Parts are put thru a rapid aging cycle while under electrical stress. 1000 hours is typical and an industry standard. There is an accepted rate of failure, usually highest at lower hours.....so-called infant mortality.
That being said, I have never had a SS failure...cap, resistor, I/C, discrete device. My old amp was on as continuously as possible for 20 years. Other stuff switched as needed.
My company makes a line of 'HiRel' and some of our stuff even finds its way into output and power supplies.....and makes full-on class 'd' modules.
Everything has a MTBF, even if it is a HUGE number.
I personally count on Infant Mortality of new equipment. Do a run in for a couple weeks and call it good.

Doesn't current surge, the first 1/2 cycle to saturate a transformer count as stress?

"Doesn't current surge, the first 1/2 cycle to saturate a transformer count as stress?"

Yes it does create rush current - but there is no evidence that it shortens life of electrolytic caps. It does nothing to transformer or rectifier.

You won't fint MTBF on any datasheet of any transitor, diode, IC (digital or analog) etc. - probably because it is in order od few hundred years. The question was if switching on/off is shortening life compare to keeping it constantly on - it doesn't. It will shorten life of switches, relays, tubes but not the SS stuff.

You won't fint MTBF on any datasheet of any transitor, diode, IC (digital or analog) etc. - probably because it is in order od few hundred years.

Yes, but when you combine hundreds or thousands of devices into a system, the several hundred year mtbf of an individual device essentially gets divided down by the number of devices (with each device receiving greater or lesser weight in the overall calculation depending on its individual mtbf).

There is an entire branch of engineering that deals with this, known as Reliability Engineering. MIL-HDBK-217F, that I linked to above, provides an inkling of how involved it can be.

Regards,
-- Al

Almarg - when you have hundreds of thousand od devices you increase chance that one, or connection between them, might be faulty. If number of devices should decide alone on reliability then Pentium processor that contains milions of transistors should fail every day. What about system that contain thousand of Pentiums - it should fail constantly.

It is not even relevent to audio gear that contains small amount of components. Stress done to wires inside of IC caused by current flow cannot be that bad since we have some of ICs working (and often in bad environment) for about 50 years (and first ones were poorly made).

If you believe that SS amp is more prone to failure from switching on and off - please tell me what fails!!!
(Don't count cases when you turn system off to modify something - like shorting speaker wires and then turn amp on again with failure.)