Fat.Tag@SU-SIERRA.ARPA (09/06/84)
From: Tim Gonsalves <Fat.Tag@SU-SIERRA.ARPA> In a recently pcized office on campus the question has arisen as to whether it is better for the machines to turn them off when not in use or to leave them on the whole day (or whole week). Does turning a machine off and on reduce its lifetime? When does this become a problem compared to the ageing due to leaving it on when not in use? Is the answer different for the monitor and the system unit? Some of the machines are used sporadically, say, for 1/2 hour now and again, others more continuously. Any experiences in this matter would be appreciated. The pcs in question are of genus IBM with floppy drives. However, I suspect that there should be general principles applicable over a variety of machines. There is a minor conflict of opinion between the head of the office, who has a fan-less Macintosh, and the plebes who object to unnecessary background noise. The office, located in Stanford U., is not air-conditioned. If there is sufficient interest, I will summarize the replies. Thanks in advance, Tim Gonsalves Gonsalves@SU-Sierra.Arpa -------
mlsmith@nadc.ARPA (09/10/84)
Frequent >3 times/hr) cycling of power on IBM-PPC's is definitely bad for the power supplies health. We have 3 PC's in the office and one power supply failed after three weeks. Due to power conservation we turn off the PC's at least once a day. CTRL ALT DEL sequence solved most of our power cycling excesses (software reset). good luck with your PC's mlsmith@nadc.ARPA
eve@ssc-bee.UUCP (Michael Eve) (09/11/84)
I tried to find the answer to this question myself, but couldn't find any reliablitity data which made sense to me. What I did find indicated that once a chip was pass the infant mortality stage (100 hours or so), the chip became virtually immortal if run continuously. Data was scanty but I got the impression (reinforced by my preconceived ideas?) that thermal cycling (on and off) was a major factor contributing to lead breakage within the chip. I became interested in this when my Apple started dying once a week after 3 years of heavy use and being turned on and off several times daily. It rarely died while in use, but would often refuse to boot up when restarted. After replacing several chips, I now leave my Apple on all the time. Sometimes I will now go two months between uses and the machine would still be up and running. caveat: Within the past 2 weeks, the machine has again become flakey. This time I found just moving RAM chips around (not replacing them) will make it work for another day or two. I wonder if oxidation of the pins from continuous use is now catching up to me. (The older Apples run very warm.) mike eve ssc-vax!eve@ssc-bee p.s: I originally submitted this as a reply, but it was returned undelivered so I have now made it a followup--maybe somebody is interested. -- Mike Eve ssc-vax!eve
julian@ecsvax.UUCP (09/21/84)
I never imagined such voluminous reponse to a simple question, but there must be some controversy about powering down. Bad air conditioning and no-reset-key on the IBM PCs favor frequently power-downs. But, the harm in power-downs seems to be the spikes and voltage transients originating from or harming the power supply. But what if the electrical lines are conditioned and free of spikes and transients? Then would it be safe(r) to power down often? Many sites have conditioned lines, and many micros are equipped with electronic filters. The implied culprit is a trashy power supply (is there no such thing as clean power?). Now, we have time to talk about questions like power-down sequences. Which one do you turn off/on first -- CRT, disk, computer, modem, ... -- and what effect does it have on the equipment? Phil Julian mcnc!ecsvax!julian (continue)
schimpf@utah-cs.UUCP (James Schimpf) (09/22/84)
In this discussion the way light bulbs fail (i.e. when turned on)
is mentioned as a justification for leaving equipment on. An
incandescent bulb failure ,assuming no mechanical shocks, is usually due
to high current through a thin spot on the filament melting the
tungsten. The thin spot is due to the migration of the tungsten atoms
from the filament to the glass bulb (Edison Effect). The failure occurs
at initial turn on because the metal filament has a positive temperature
coefficient of resistance change. Thus the tungsten has its lowest
resistance (i.e. allowing higher current) when it is cold. In IC's
there is an analogous effect (electromigration) which affects the metal
(particularly aluminum) traces on the silicon. IC manufacturers are
very aware of this effect so the designs compensate as much as possible.
But if this occurs and when turned on the chip is cold then the current
surge due to the lower resistance of the metal leads could be the final
blow.
On the original subject most newer terminals are designed to shut
down the screen when not in use. Floppy drives now stop spinning when
not in use. Hard disks I agree are another problem but with sealed
winchesters it seems that bearing failure is a much smaller threat than
track loss due to bouncing heads.
Finally my personal experience with a half dozen micros which
were left on continuously for years has been quite good. In that time
we lost only two dynamic ram chips and a disk controller chip. Over 90%
of the down time was for periodic adjustment of floppy drives.
Jim Schimpf
CS Department University of Utah