Date: Tue, 26 Nov 1996 16:31:02 -0700 (MST) From: Terry Lambert <terry@lambert.org> To: msmith@atrad.adelaide.edu.au (Michael Smith) Cc: terry@lambert.org, chat@freebsd.org Subject: Re: SCSI A/V drives Message-ID: <199611262331.QAA25852@phaeton.artisoft.com> In-Reply-To: <199611262302.JAA04660@genesis.atrad.adelaide.edu.au> from "Michael Smith" at Nov 27, 96 09:32:39 am
next in thread | previous in thread | raw e-mail | index | archive | help
[ ... thermal expansion ... ] > > It *could* be predicted, if they stored a thermister value at format > > time. > > Presuming you could get an accurate enough sensor close enough to the > platters, and presuming that their expansion is linear across the > disk surface. > > Only you can't, and it isn't. And it's "thermistor", or in the case > where you're trying to be accurate it's a PT100 or a thermocouple. 8) Actually, the coefficient of radial expansion for various temperates could be calculated from knowledge of the materials inside the box to which you attach your electronics. This is something a manufacturer should already know. The problem is knowing location along an expotential curve. The slope of the curve is relative to the baseline, which you would establish at the time you formatted. So it *is* possible to predict where a radial line will be given a known coefficient, the current temperature, where it was, and the temperate at the time you determined where it was. The problem in prediction comes from not knowing the temperature at the time of format. Even if you didn't know the temperature at the current time, if you knew the temperature at the time of format, you could locate a post expansion known track, and use it to interpolate the curve slope, with knowledge that the expotential is radially bound. This would be a one track thermal recal instead of a whole disk thermal recal, since you could get your temperature from knowing where you thought you were vs. where you actually were. In theory, you could look at the first track, the last track, and some track in between (hopefully close to the first track) and interpolate the curve (and thus the location of the other tracks) from there. This would require a great deal of "embedded" knowledge about the properties of the physical disk, and the calculation would probably take too long without a known baseline. With a known baseline, any track you read could, in theory, give you enough information for a full recal. As to putting the thing close enough to the platters: ugh. That would be a problem, assuming the actual and ambient temperatures were inequal. Probably you'd have to be so exacting about thermal emissibity, and therefore manufacturing tolerances, that it wouldn't be worth it. Never mind that somm weeine out there will end up reformatting the thing at some unknown temperature, right out of the box, so the damn thing heats up as the format progresses, drastically skewing the baseline value. 8-(. Better to just do the recal, if your equipment isn't going to be used in known conditions. If you are going to engage in preemption, and it lasts long enough to actually matter, then you would be well off to format the drive at operational temperature, and maintain the temperature at the same value as you use it. So even if it matters in theory, there is no longer any practical consequence. Anyway, too much on this subject. 8-). Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199611262331.QAA25852>