From owner-freebsd-hackers Thu Jan 2 16:35:57 1997 Return-Path: Received: (from root@localhost) by freefall.freebsd.org (8.8.4/8.8.4) id QAA17360 for hackers-outgoing; Thu, 2 Jan 1997 16:35:57 -0800 (PST) Received: from whizzo.transsys.com (whizzo.TransSys.COM [144.202.42.10]) by freefall.freebsd.org (8.8.4/8.8.4) with ESMTP id QAA17333 for ; Thu, 2 Jan 1997 16:35:49 -0800 (PST) Received: from localhost.transsys.com (localhost.transsys.com [127.0.0.1]) by whizzo.transsys.com (8.8.3/8.7.3) with SMTP id TAA25368; Thu, 2 Jan 1997 19:34:43 -0500 (EST) Message-Id: <199701030034.TAA25368@whizzo.transsys.com> X-Mailer: exmh version 2.0alpha 12/3/96 To: spork cc: Christoph Kukulies , freebsd-hackers@freefall.freebsd.org From: "Louis A. Mamakos" Subject: Re: advice sought - Quantum 2GB Atlas broken References: In-reply-to: Your message of "Thu, 02 Jan 1997 18:47:08 EST." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 02 Jan 1997 19:34:41 -0500 Sender: owner-hackers@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > . Now I just have to find > a way to detect when the CPU fan dies... If you run xntp on your box and your clock is synchronized over the net to a reference clock, you could monitor the contents of the /etc/ntp.drift file. This contains the computed drift rate for the system's clock, which is derived from a crystal oscillator which the interval timer is clocked from. The frequency of the quartz crystal varies with temperature; this cause the drift rate (which is essentially the frequencey error) to vary. If it changes by a lot, then some significant thermal event has occured. (This is why your cheapo plastic wristwatch keeps good time - it's got it's own crystal oven attached to it, keeping it heated to a relatively constant temperature). I once calibrated the computed drift values with temperature for a workstation in my office - you could tell the temperature of the room within about 4 or 5 degrees by logging into the system and looking at the computed drift rate. Of course, this isn't a great bit of help for the CPU fan. Perhaps you could do a bunch of divide operations and check to see if you got the right answer :-) There's a company I saw on the net with an RS-232 attached temperature sensor. You could actually attach up to 16 Dallas Semiconductor DS1802 temperature sensors to it, and query their temp via the RS232 interface. Something like this might make sense, if you could figure out how to attach to temperature probe (which I think is in a TO-18 "can" package) to the heat sink on the CPU. Check http://www.spiderplant.com for details. louie (By the way, Happy (no leap-second this time) New Year!)