From owner-freebsd-hackers Sat Dec 2 15:13:13 1995 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.6.12/8.6.6) id PAA12511 for hackers-outgoing; Sat, 2 Dec 1995 15:13:13 -0800 Received: from clem.systemsix.com (clem.systemsix.com [198.99.86.131]) by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id PAA12426 for ; Sat, 2 Dec 1995 15:12:29 -0800 Received: from localhost (localhost [127.0.0.1]) by clem.systemsix.com (8.6.8/8.6.6) with SMTP id QAA05944; Sat, 2 Dec 1995 16:10:42 -0700 Message-Id: <199512022310.QAA05944@clem.systemsix.com> X-Authentication-Warning: clem.systemsix.com: Host localhost didn't use HELO protocol X-Mailer: exmh version 1.6.2 7/18/95 From: Steve Passe To: Don Yuniskis cc: freebsd-hackers@freefall.FreeBSD.org (FreeBSD hackers) Subject: Re: No Thumbs??? (lack of concensus) Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sat, 02 Dec 1995 16:10:41 -0700 Sender: owner-hackers@FreeBSD.ORG Precedence: bulk Hi, >"third"? is this a full height case with three third-height knockouts? yes, I guess I misspoke b4, it's a "full height 3.5 inch" formfactor which means one MC3243 will fit in it. It comes with a 2/3 knockout (snapout, actually) and a 1/3 knockout panel. I leave the 2/3 in, replace the 1/3. The resulting open area is immediately in front of and aligned with the top of the disk drive top surface. >I'll have to see if I can get half or third height knockouts with vent holes >(the full height panel only has one set of holes along the *bottom* (tho' >I could flip it over...) unless it looks like swiss cheese something like that will only allow a fraction of the airflow of a removed panel with foam insert. >Hmmm... the fan in this box is 25x80mm. Claims to be 250mA (i.e. 3W). >Just gauging from size, similar fans (ball or sleeve bearing) in DigiKey >catalog seem to be 25 - 40 CFM. Of course, it's thermostatically controlled >so I'll have to hot-wire that... thats a step up in size from the ones used in the box I mentioned. They are 40mm square by 20mm thick. The thermostatic variety fan is another potential for trouble. I replaced a powersupply with one of these in a cabinet where the supply sat in the bottom of a tower cabinet. The thermistor never saw the heat generated by the cards, which rose to the top of the cabinet. I happened to notice one day that the cabinet top was almost too hot to touch, yet the fan was barely moving. Needless to say, I never use these anymore. >> I recommend 2 3243's in 2 half-height boxes. > ... >So far, my only disagreement is 2 drives in a *single* full height >enclosure. :-/ as I said above, I misspoke, I meant to say: I recommend 2 3243's in 2 full-height 3.5" boxes. (as oppossed to one double height box) >> combinations >> of external boxes that became unhappy with mixed flat and round cables. >... >> ... there are a lot of impedance changes happening. ... >... >Ah, well, I'll hopefully buy good enough cables ;-) cable quality is not the problem, its the different impedance characteristics of ribbon vs. shielded round cables. The ANSI SCSI-I spec says: " ... characteristic impedance of 100 ohms +- 10% is recommended for unshieled flat or twisted pair ribbon cable. ... greater than 90 ohms is preferred for shielded cables; however, most available cables have a somewhat lower characteristic impedance. To minimize discontinuities and signal reflections, cables of different impedances should not be used in the same bus." >... >Right. My point was that if the yellow is ignored, don't count on >someone "observing" (i.e. reacting to) the red, either! I feel that yellow my widget doesn't count on anyone observing either condition. there is a separate hook for a user provided callback for each condition. I expect the user to put code in place to 'do the good thing' without human intervention. The visuals are there to make it look pretty. I suppose they would also be useful while one was 'tuning the airflow' of a cabinet, etc. I also expect to add a stripchart widget once I actually build the hardware. One could then get a history of temp. changes over time. With this you could compare the occurance of transient system errors (parity, core dumps, etc.) with the temp. of the machine at the point they occurred. >How long has this been in service and what kind of (ab)use does it see? >I'd like to order 4 of them next week (or comparable). I've been using it since installing the 2.1.0-092295-SNAP, so about 2+ months now. It sees moderate usage as a personnal workstation. never seen a disk related error yet. There has been alot of mail lately about the quality of the micropolis drives going downhill, which bothers me a bit. My own experience with micropolis has been such that I have used nothing else for 4-5 years now. I have to admit that I have always seen a high 'infant mortality rate' with them, maybe 1 in 10 in the last 5 years, but have never had any problem getting a replacement. And when they make it past that first week, I have never had one die in the last five years. Since I run mostly unix, that means 5 years of round the clock usage for the older ones (330 Mb SCSI I). My advice would be to buy them from a reputable dealer, then burn them in well, perhaps run 'locate' every half hour from cron the first few weeks. If that doesn't kill them they should last for awhile. Also let me qualify everything I have said above as to being based on my own limited personnel experience as an end user. I buy maybe 2 drives a year, so my statistics are fairly narrow. -- Steve Passe | powered by smp@csn.net | FreeBSD