Some opinions on leaving your computer on 24 hours a day and mounting your machine sideways. =============================================================================== This is the version that was included in the Frequently Asked Questions file until May 1990: There's a lot of stress placed on the system when it first comes up. Generally, you should leave your system running if you plan to return within the next few hours or so; if you don't expect to be using it for a day or so, whether or not you turn it off is a personal judgement call. Monitors aren't the same, though, since images can get "burned into" the phosphor. Use a screen blanker, or turn it off if you'll be away for more than five or ten minutes. As for mounting it sideways: The most common myth about this is that it will make the disk drive bearings wear unevenly. If you look at the manufacturer's information when you buy a disk drive, you will see that it is warrantied in any position *except*upside-down* (with the defect label down). However, the disk may hang slightly differently in the bearings after being placed on its side, so if you plan on mounting your system sideways, you should back up your hard disk while it's level, then mount your system sideways and do a low-level reformat of your hard disk and restore it from the backup you made. If you do this you shouldn't have problems. =============================================================================== From: chuck@eng.umd.edu (Chuck Harris) Subject: Computer on 24hrs.day? Date: 30 Mar 90 I can provide some info on the phenomonon. Semiconductors have failure curves that look like: || || Failures|| | \ /... | \ / |_____\______________ ... ___/__ 0 100 1E6 = ~100 years Hours of use The failures in the first 100 hours are some times called Infant mortality. I'm not sure at all about the 1E6 hours (Lets say it's a lot of hours!) The first 100 hours failures are the reason the most reputable manufacturers will "Burn-in" new systems. Statistics shows that there is little gain in confidence by burn-ins much longer than 100 hours. NASA (I think) found that semiconductors that had been left on for a long, long time were very likely to fail if their power was cycled off then on again. However, statistics showed that the same semiconductor if just left on would continue to work almost indefinitly. As I recall the failures due to cycling power occured while the power was off. it had something to do with the transistor's junctions migrating too close together while power was off, then when power was turned on, the transistor failed because the junctions were shorted. (power being on continuously apparently prevents this) Anyone know more about this? These are the tests that everyone alludes to when they tell you to leave computers and other electronics on all the time for greater life. They only mean something if you have a statistically large number of transistors (the transistors need to be from different lots not just a large number on an IC) in your system, and you are using the system for a LARGE number of hours. Your individual PC does NOT have a statistically large number of semiconductors in it. The entire country's PC's do. 100 years is a long time. Conclusions that I think you should draw from this diatribe: 1) Big computers that have millions of IC's in them perhaps should be left on. 2) Small computers (PC's etc) It just won't matter. Note: Failures of mechanical parts are nothing like that of semiconductors! (You know, disk drives, switches, keyboards, fans, etc.) I wish I could provide references to all of what I have stated, but I can't easily. This is stuff that I have gleaned from years as an Engineer, and many many hours of college course work. So some settling may have occurred :-) Chuck Harris C.F. Harris - Consulting =============================================================================== From: nicholso@hpcuha.HP.COM (Ron Nicholson) Subject: Re: Dusty Dorms. WAS: Re: Computer on 24hrs.day? (yes or no) Message-ID: <10350001@hpcuha.HP.COM> Date: 30 Mar 90 23:29:35 GMT gregk@ubvax.UB.Com (Greg Kendall) / 6:59 pm Mar 28, 1990 / writes: >I've heard a lot of claims about how it's "harder" on the PC to power >up than to leave it on. I have yet to hear of any real data on failure >rates. ... ---------- Long long ago, in a far away place, when I worked for a high volume computer manufacturer, I ran across some real statistics. Some experiments had been done on the difference in failure rates between continuous burn-in and power cycling. My dim recollection is that there was a significant increase in the infant mortality rate of the group that underwent power cycling. The sample size was large enough to be statistically convincing. The primary cause of the failures was due to thermal shock on solder joints, IC bonds, sockets and connectors. Alas I no longer have access to the details of that experiment. What I now do is to frequently power cycle new equipment (while it's still under warranty of course) to shake out the lemons, and to minimize power cycles thereafter. I still have seen no good data on electromechanical equipment, like disk drives. --- Ronald H. Nicholson, Jr. Hewlett Packard uucp: nicholso@hpda.HP.COM Cupertino, CA =============================================================================== From: dan@tinton.tinton.ccur.com Subject: Computer on 24hrs a day? Date: 5 Apr 90 Power conditioning must play a part in the equation. If one's AC supply is relatively dirty and one has limited power conditioning equipment, then leaving one's system on constantly leaves it open to large power glitches wreaking havoc. =============================================================================== From: uunet!tiamat!quintro!bpdsun1!rmf (Rob Finley) Subject: Re: Computer on 24hrs.day? (yes or no) Date: 15 Apr 90 06:38:02 GMT My floppy drive fails to read some disks when cold. So, I leave it on. After three years, the only thing that died is the cheap 12V fan in my UL listed power supply. $15 and a trip to Radio Shmuck did the trick. It's still ticking. Just be sure to use a good screen blanker program. Most of our machines at work are never turned off. But. My 386 at work ate two hard drives and three motherboards before they replaced the power supply. The +5v line ran at 5.2v while the other voltages were acceptable with low electrical noise. New supply and it works great. Before deciding whether to leave it on or not. Consider these points: Does your system draw air into the box through a vent on the back? If you feel air being blown out the vent on the power supply, then it is most likely sucking it in through the biggest hole: your floppy drive. When air comes in through the drive door, it drops the dust it was carrying all over your machine, mostly on your disk drive. Remedy: replace the power supply with one that has the fan going in the right direction. Swapping the Red and Black wires on it won't do. You will cause irrepairable damage to the solid-state controller (they don't have brushes like conventional DC motors to control noise). I had to open the power supply box (after removing it from my machine) unscrew the fan mounting hardware and turn it over so that it draws air in from the vent that sticks out the back of the cabinet when installed. With your fan now drawing air from one place, you can tape a piece of foam or air filter material over that vent to catch a large percentage of the dust before it gets inside. But, you must check the filter regularly. The entire system may be at risk with a blocked air filter. Or, if anything, your expansion boards will attract the dust before it reaches your floppy drives. Unfortunately, too much dust on the motherboard or expansion boards will insulate the chips and prevent them from keeping cool. If you have large expansion boards or full height hard drives, look into adding additional fans. Some cabinets allow you to have a second one on the front end of the expansion card cage. If you don't want to open your machine and it overheats or attracts dust in all the wrong places, turn it off if possible. Machines today are durable. Safety warning. If you don't feel comfortable opening your machine, find someone who is. If your dealer doesn't feel comfortable, find another dealer. The one you have now probably can't fix it if it dies. The power supply circuit can still hold a charge when it is unplugged. You shouldn't have to touch any of the circuits on the power supply board. You aren't rewiring it, you're flipping the fan over and putting the screws back in. That wasn't hard. Was it? ----- quintro!bpdsun1!rmf@lll-winken.llnl.gov uunet!tiamat!quintro!bpdsun1!rmf =============================================================================== From: kabra437@pallas.athenanet.com (Ken Abrams) Subject: leaving PC on Date: 8 Aug 90 In article hart@blackjack.dt.navy.mil (Michael Hart) writes: >I refer you to the "light bulb law". That law being: When do light bulbs >burn out?? _when you turn them on_ There is a large surge of current through Even though most of the components in a PC are NOT light bulbs ( and don't behave like one either), I would agree with you that leaving it on is best if the ONLY consideration was hardware reliability (and it is connected to a stable power source). As I see it, however, hardware reliability is not the only issue and a case can be made for the theory that a properly designed disk drive will wear out from heat and friction and dirt before it will suffer any electrical damage from being started and stopped once a day. I think that all the energy wasted by millions of PCs left on 128 hours a week when they are not being used is a bigger and more important issue than whether or not it will extend your repair cycle from 3 years to 5 years. There is also a small extra potential for a fire in a running device. I have been in the computer industry for close to 25 years, mostly as a technician. I have weighed all the arguments and I have decided to turn MY machines off when they will not be needed for 6 hours or more. I even turn the Xenix box off over the weekends. Like a lot of other things in modern life, this is not strictly a technical call but it has some moral undertones too. Make your own call but don't overlook part of the factors in the process. =============================================================================== From: linderd@merrimack.edu (Doug Linder) Newsgroups: comp.sys.ibm.pc Subject: Re: Horizontal cards (was Re: Standing your box upright) Message-ID: <18991.262c5b5c@merrimack.edu> Date: 18 Apr 90 12:19:40 GMT In article <1990Apr16.181035.3017@seri.gov>, marshall@wind55.seri.gov (Marshall L. Buhl) writes: > Another thing to consider when standing a PC on end. One of my secretaries > stood hers on end for a year or so. When she left, I tore the machine > apart and found that the AST 6 Pack was really warped. If you laid it on > a table, one end would be 1" in the air. It still works, but... Maybe the solution is to flip the thing from end to end every 6 months to even out warpage? ;-) But seriously, folks, when I was a PC tech I did see some problems with this. Essentially, it seems that the larger the number of cards you have, how big the cards themselves are (height/thickness) and how good your fan is are the big factors. Remember, these cards are designed to stand on edge and let the heat float to the top of the box where it can be sucked away by the fan. A PC on its side screws up this system and the heat, trying to rise, gets caught by the cards and what you end up with are little heat sandwiches between the cards - the hot air has nowhere to go. I have seen machines with lots of big cards in them that you could fry an egg on. The biggest danger with cards "melting" is not only that the card itself may malfunction, but that it wil come in contact with something it shouldn't (usually another card, the one below it) and short out the whole works. I have seen motherboards die this way. This can be alleviated by some things such as: 1) a fan (a standard room fan pointing at the machine), 2) room A/C, 3) simply leaving the cover off the machine and covering it with a cloth supported by a wire frame (to hold the cloth away from the machine a few inches) 4) If the cards are long enough, mounting brackets at the front of the machine for the ends of long cards to rest in will prevent "drooping" at the ends but alas, cards still droop in the middle. My best advice: Unless you have only a monochrome card and a serial/parallel card, or some other very low heat/high circulation setup inside the machine, place it the way it was designed to be placed - "power users" take note. The equipment is too expensive and your time too valuable to waste with breakdowns. Would you want your PC to die in the middle of a presentation because the Video card just warped enough to touch the drive controller and short out the whole works? BTW, though, I have found that smart terminals like Novell network PCs work OK on end - as long as the novell card is about the only thing in the machine. Hope this helps. -- Douglas D. Linder linderd@merrimack.edu Merrimack College, N. Andover, MA {uunet,wang,ulowell}!samsung!hubdub!linderd =============================================================================== From: alz@tc.fluke.COM (Al Weiss) Subject: Is it [orientation] harmful to disk drives? Date: 4 May 90 Dunno about floppies. For hard drives it depends upon the manufacturer. All of the Seagate manuals I've seen, for instance, are VERY specific in their manuals about not elevating the front or back more than 5 degrees from horizontal. They can however be turned on their sides up to 90 degrees, but no further (ie not upside down). They should also be formatted in their permanent orientation. My understanding is that the head positioning mechanism gets worn out on an angle, and the motor bearings can't hack being upside down. On the phone, Seagate told me that they would not honor any warrantee if they know that it has exceeded those limits. I've heard some newer Seagates don't have limits, but don't know for sure, nor do I know about the CDC's. Conner, Quantum, Miniscribe, Maxtor(?), on the other hand, (of the ones I've seen) specifically say "any orientation". ===============================================================================  |-THiS FiLE PASSED THR0UGH --- /\ ---.------ /\ ---*--.- FiDONET 2:200/612 --| | . * . // \ . // \ . FUJiNeT 7:102/102 | | I.C.S Swedish HQ // \ + // \ . MeGANeT 66:666/1 | | + // / \ // \ + NeST 90:1101/112 | | Sync World HQ /\\ \\ / . // \\ / | | . // \ \/ // /\/ . 16800 DUAL STANDARD | | +46-451-91002 \\ / / \\ \/ + | | * \\ / + . \\ \ . . . | | . \\ / \\ / | |- SysOp: Troed ------------ \/ARCASTIC -- \/XISTENCE --- CoSysOp: Zaphod B -| < Advertisment added using -=Bad Ad=- 1.92 by Troed/Sync. BBS: +46-451-91002 >