On 4/12/02 janet schwartz [EMAIL PROTECTED] (flawed jai) wrote: > sidney ho!--outstanding initiative on your part, negotiating with the > upgrade engineers like that. my vast admiration and excitement at your > work. Thank you for the compliment and your enthusiasm. (But, AFAIK Janet, you have not previously indicated an interest in this upgrade??? :-) Fortunately previously monopolistic European phone rates have come down with market deregulation and surplus broadband capacity in the transatlantic pipe so my "dimes" don't have to be thrown out as fast ;-) > > have you forwarded to them the blue comet/black comet studies you posted > here? esp regarding the 'goolsbee'd' improvements to the heat sink, and > the results as measured? No, but I've long anticipated that heat will be the major limiting factor as to whether this upgrade can go forward or not. I'll expand upon this below, interspersed with your thoughtful suggestions. It's not useful and premature to barrage them with "unscientific" information. Any serious engineering effort will not rely at all on any third party data, however well-intentioned, and cannot, due to significant product responsibility and potential warranty issues, not to mention damage to reputation. I'll go into a bit of "anticipated" detail here, from the position of a (hopefully) knowledgeable layman, to dispel illusions/hopes that this might go forward painlessly--but I promise to be insistently promoting a serious effort.
Our "amateur/hobbyist" numbers are OK for indications but faulty for many reasons, for example: --no indication of ambient temperature --no ambient temperature control --indirect measurements via various algorithms arriving at inferred temperatures (which I've already established as wildly inaccurate). Even comparing between 2400s, it's impossible to know which software conflicts or otherwise influences measurements, and which other routines have priority in queueing for available processing power, but my own inference is that the significant variation between different software merely underlines a significant unreliability in temp measurements. --no controlled baseline for hardware/software configurations, repeatable. PC cards, especially modems, generate heat (each differently) and HDs are a huge heat source (sometimes hotter than CPUs!) so I've made a much bigger heat sink under the HD. Obviously different products generate different amounts of heat, and using both PC slots is different from using one--and there is even some difference between the upper/lower slots and whether you leave the PC slot plastic doors open/closed...etc. etc. Well...you get the idea. To assess heat transfer problems properly, a conditioned chamber equipped with temperature regulation for both ambient air and the platform should be utilized, with precision instrumentation and indexed software to drive the CPU at a certain (variably) constant rates of processor load. Then temperature probes and thermal imaging should be employed to assess heat transfer rates and dissipation patterns. The amount of electrical energy consumed to keep temperatures constant at different loads should be additionally converted to equivalent caloric values to calibrate CPU heat outputs. I'd presume laptop upgrade manufacturers should have even more sophisticated means to make their engineering determinations--as that would be the minimum--just to give (Mensa member(s)?) an idea ;-). Part of my discussions (including the "soft" part of leaving the proverbial door wide open to "unofficial" solutions) are to encourage a niche manufacturer for a "narrow niche" product to consider utilizing its possibilities together with a knowledgeable, enthusiastic, motivated user support network--which is naturally its best prospective customer base. This cannot be overtly stated because a manufacturer, irrespective of its size, cannot rely on such a support network for hard performance/warranty issues of which a CPU daughtercard is certainly the most critical component (bearing in mind the CPU itself can only made by a multi-billion $ corp.). So there are many issues to be anticipated and discreetly finessed in the coming months. > maybe they could make the upgrade a whole package, selling it with the > specifications that the card not be installed until chuck goolsbee's > heat sink maximizer steps be performed upon any machine it is going to > be used in. make it a requirement of warrantee, in big loud obvious > warnings on the package. An excellent suggestion, and already generically discussed--but more proffered as serious 3rd party support (for the time being). The issue of heatsinking/improved cooling can have many solutions of varying complexity. The expected (and undiscussed) problem is a manufacturer selling an improvement would normally want to avoid "admitting", however tacitly, that their upgrade needed additional 3rd party aftermarket heat reduction measures--definitely not a good selling point or portent for long life (perceptually), even if entirely reasonably and necessary. > > maybe chuck could 'goolsbee' them up some prototypes to run their > development tests on, with their 600MGz prototypes. > see what they say to that proposal. I love the creative verb! There are many possible approaches here. The method Chuck used, AFAIK, was an adaptation of a Japanese website which uses a thermal pad, copper, and thermal grease quite effectively. My own mods use folded, compressed aluminum foil and a thin black PVC substrate to cut any (remotely) possible electrical arcing for both CPU and HD (the latter is an intense heat producer). From an engineering standpoint, whenever possible, less is more--I see problems and sourcing hassles/costs to provide kits of precut copper, thermal pad and a tiny tube of thermal grease. Moreover, from a materials standpoint I don't like the idea of a semi-viscous material inside my laptop, especially because I dismantle it so often--but that's just me. Different installers will use different amounts of thermal grease differently, adding somewhat to variability between different units. The main problems are to eliminate air gaps and to ensure max heat transfer. My 2400 stays lukewarm, no matter what the internal measured temperature is. I've been considering where to drill strategically placed holes in a decent aesthetic pattern to improve passive airflow, should an upgrade be problematic for thermal reasons. However, I assume it is highly unlikely that an upgrade manufacturer will even contemplate an engineered heat sink solution of any kind, beyond supplying the thermal pad that came with Interware's G3-320 and 400. This will have to come in the form of official "unofficial" support by qualified installers--remembering even Apple severely restricted 3rd party repairs on 2400s. There is no "standardized" or "guaranteed" heat reduction solution which then raises the stakes for warranty responsibility issues. So the way to finesse this in a way compatible and comfortable with cost/warranty/ installation issues is an upcoming headache---assuming the G3-600 is technically feasible. Hopefully it will run cool enough so everything is vastly simplified. My fallback will be to offer a considered aluminum block for both the CPU and HD, possibly with an additional heat pipe for the CPU block, but this Rubicon can be crossed at a later date, if necessary. Then I'm of a mentality that I'd never buy a computer that needed to run an external add-on cooling pad of any kind. Just like turning down CPU speed in software, I think that's self-defeating as an "inadequately" engineered product. I have the same attitude towards too-early adaptation of OS X or paying for Apple's "beta field-testing" of the latest and greatest hardware teething problems (especially when compounded w/software incompatibilities) because I've been (expensively) down that road before and will leave it for the "adventurous". Hence, I'm trying to shepherd this 2400 upgrade project in a way that it can reach birth through its "natural" evolving possibilities--alternately softly, firmly and seriously encouraging the prospective vendor to invest preliminary R & D--and I'm very pleased they've put some (previously 2400 G3 expert) people to do this. I'm very used to industrial performance/product guarantee, manufacturing, testing issues since we've contracted/delivered many light industrial plants over the years, in exceptionally difficult and demanding environments (people, countries, conditions), and some with filtered, conditioned clean rooms. So, anyone/everyone seriously interested, please sign up and support by your numbers! Your collective enthusiasm is much appreciated! Now I'm going to be monitoring posts but go to mainly lurk for a while to do serious work. I may/may not reply, even to direct queries for a while--no offense intended but I have a major busy spell coming up... Sidney Ho ---------- Duo/2400 List, The friendliest place on the Net! A listserv for users and fans of Mac subportables. FAQ at <http://www.themacintoshguy.com/lists/DuoListFAQ.shtml> Be sure to visit Mac2400! <http://www.sineware.com/mac2400> To unsubscribe, E-mail to: <[EMAIL PROTECTED]> To switch to the DIGEST mode, E-mail to <[EMAIL PROTECTED]> Need help from a real person? Try. <[EMAIL PROTECTED]> ---------- Dr. Bott | 10/100 Ethernet for your 2400 is finally here! MPC-100 | <http://www.drbott.com/prod/mpc100.html> RoadTools $30 PodiumPad available at Apple retail stores, $20 Traveler CoolPad at Staples. Both in white for iBooks at <http://roadtools.com> Midwest Mac Parts ][ <http://www.midwestmac.com> After-market parts for Macs. ][ 888-356-1104 ][ MacResQ Specials: LaCie SCSI CDR From $99! PowerBook 3400/200 Only $879! Norton AntiVirus 6 Only $19! We Stock PARTS! <http://www.macresq.com>
