Date: Tue, 14 Apr 2009 23:55:43 -0400 From: Michael Powell <nightrecon@verizon.net> To: freebsd-questions@freebsd.org Subject: Re: questions about Fatal Trap 12 Message-ID: <gs3lq3$ktt$1@ger.gmane.org> References: <200904031144.31198.ray@stilltech.net> <02c401c9bb0e$d2b3e130$6900a8c0@NBD41VJ> <200904141523.31164.ray@stilltech.net> <4ad871310904141529r5dbe9517ub29bf56a87045e62@mail.gmail.com> <49E51B1E.8070400@ibctech.ca>
next in thread | previous in thread | raw e-mail | index | archive | help
Steve Bertrand wrote: > Glen Barber wrote: >> On Tue, Apr 14, 2009 at 5:23 PM, Ray <ray@stilltech.net> wrote: >>> I Just had the power supply die on this machine. Could a failing power >>> supply cause this type of issues? >> >> Absolutely. > > Seconded. Power supply issues have caused me this kind of grief more > times than memory has over the years. > Yes, "me too". With the advent of the digital age there is less and less cognizance of analog electronics these days. Simply checking the output voltages with a Radio Shack VOM will make things seem to be OK, as in the measurements seem to be within the range you'd expect, but this is not an entirely accurate assessment. It can be the case where a power supply has aged to the point that it can support only some fraction of the load it was rated at when new. As long as the load stays below this derated value the box seems all right, but plug in another drive or some other hardware that pushes it over the edge and you get another story. What you will get is non-pure DC with huge amounts of unfiltered ripple. The VOM will not show this. You would have to be using an oscilloscope to see it. But even an old power supply that is not loaded down to the point of total failure can begin to show out of spec ripple measurement as load increases. This ripple can be the source of seemingly intermittent hardware problems such as hangs, mysterious automagic rebooting, lock ups, etc, that seem to have no rhyme or reason to them. Many times in the distant past I replaced one by one each subsystem with known good ones to the point where the power supply was the last thing I tried. These days if it is an older box with a lot of hours of MTBF on it I do it first, using a known good. Probably 70%+ of the time it has turned out to save lots of time. Rather than try every thing else first I have learned to eliminate the power supply first, rather than the other way around. But I also have a 100MHz dual trace 'scope too. -Mike
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?gs3lq3$ktt$1>