At what temperature does an Apple III "overheat?"

I don't know of a exact temp. All I remember is over time the chips start working themselves out of the socket.

So you just make sure they are all secure.

TTFN,

Josh

The ones that were the worst about overheating were the ones with the 12V memory boards.  The design where the memory board was over the top of the motherboard trapped a lot of heat.  Although the /// had a huge heat sink in the aluminum rear part of the frame if didn't have a lot of way to conduct heat from the chips to that.  The /// really needed a cooling fan.  But it was famous that Steve Jobs HATED fans.  The Apple /// was a very ambitious design in the 1979-1980 time frame.  Few computers of the day had the amount of memory (up to 256K from Apple, 512K from 3rd party), built in floppy drive, and as many other standard things...  clock, ports for periperals, etc.  If the /// had been built with a fan, and the memory card had been vertical, and possibly some of the periperal I/O ports been on a vertical card instead of packed onto the motherboard it might not have had so many overheating issues.

In a way it is too bad that the /// was considered such a flop that Apple didn't really ever design a replacement/successor to it.  If they had re-designed it around the time they did the //e using ASICs to reduce the chip count and with the higher density RAM chips available at that time along with an internal 3.5" drive or at least a high density 5.25" instead of the old low density 5.25" it could have been a pretty amazing machine.  But by then it was considered by most too late for the ///, and Steve Jobs was only interested in pushing the Mac at that time.  And the rest of Apple was being supported primarily by Apple II sales.

Anyway, a /// with the 5V memory board and an added fan probably won't have too many overheating issues.

Since I don't have an Apple III, I'm not able to tell what is the proper temperature range where it works.

But I have a few thoughts on the topic which might be of interest for the Apple III scene.

The question in the OP (post #1) is somehow  misformulated IMHO. This is not how electronic engineers think or address overheating problems. It's not about room temperature, it's about die temperature, the temperature of the actual silicon die within the IC packages.

Measuring the package temperature may be deceiving, but typically is a good approximation for ICs which do not consume much power. For those the temperature gradient between the silicon die and the package surface is small, maybe in the range of 5-10 Kelvin. For the power hogs this does not apply. There is a trick to measure the die temperature by using a PN junction in the IC which on NMOS/CMOS always is found in the ESD protection structures. So you could get exact results if you want to know the gradient. But this accuracy is not needed in most cases.

Semiconductors are offered in several temperature grades, Commercial, Industrial, and Military. The timing specs are only guaranteed within the respective temperature range. The wafer fabs can't always keep all process parameters centered, so there will be border cases which won't meet spec for the higher temperature ranges, but will work for the "Commercial" range. The trap that lurks here is that most "Commercial" grade ICs still may work fine in the "Industrial" or even the "Military"  temperature range, because they came from a well centered wafer lot. Having these "better than spec"  ICs in the development phase may delude designers into believing that their design is robust. Then, later, after production ramp-up, with massive amounts of ICs being consumed, the occasional borderline lot of ICs may come along, and despite these may meet their datasheet spec limits, they might cause trouble, if the application's design is marginal / botched.

For the Apple III, they most likely used commercial grade ICs (cheaper). Commercial temperature range normally is 0 - 70 deg  C (some manufacturers specify 0-75 deg C to look better). Industrial grade would be -25 deg C to +85 deg C, again, some manufacturers extend this, but not much is gained at the hot end.

I 'd bet that Apple did not use the heinously expensive miltary grade ICs, -55 deg C to +125 deg C.

Now here is the rub: proper digital electronics design is done such that a) the timing parameters are never exceeded and b) the ICs always work within their specified temperature range. Among other things, like proper supply voltage range, observing input and output loads, etc., but these are easier to get right, as they are obvious and mistakes there are easy to avoid and easy to spot in a design review.

Less obvious is the effect of temperature on the timing margins of the digital logic. Most digital ICs slow down when hot, but TTL logic has some surprises, like getting faster at hot for high to low transition on outputs. The absolute differences are small (according to TI datasheets, ~2.5ns from +25 deg C room temperature to +70 dec C), which is ~ 25% of the gate's propagation delay.  This adds up unless chains of logic are designed such that low-to-high transitions (which get slower towards hot) and high-to-low transitions (which get faster towards hot) over a chain of gates compensate the temperature effects to some extent.

Now, based on this technical background, the prevalent questions are:

a) was the Apple III circuit design sound, timing wise ? No datasheet timing specs of any ICs ever exceeded ?

b) do these ICs run within their specified temperature range ?

If the answer to a) is "No", it is a botched digital circuit design, and usually can't be salvaged.

If the answer to a) is "Yes", but the answer to b) is "No", it's a botched passive cooling system design. Which could be salvaged by adding forced air flow - the fans which Steve Jobs hated so much (as I do, their noise disturbs thinking processes - no problem though for computer users who don't think deeply, like paper pushers).

If the answer to a) and b) are "Yes", then there is no reason for the machine to fail except for some mechanical ailment, like poor solder joints, or bad IC sockets / connectors.

So, is the Apple III a botched design or not ? You can find out. Measure the IC temperatures and how they rise when the machine is running, until it crashes, you can systematically go through it with small squirts of cold spray on the suspect ICs to see which one is the one causing the crash. Cold spray will also unveil bad contacts, PCB trace hairline cracks, and bad solder joints in most cases.

But be careful - the good inflammable cold spray we had back in the day was banned because of the ozone hole scare. The ones they sell nowadays are flammable and hence, a hazard for  your  health and property instead of being just an alleged hazard for the ozone layer which might just have been a lie, like the climate change scam. And the "fossile fuel" aka "recycled dinosaurs" lies. But I digress.

So if you use modern cold spray, keep open flames away (i.e. don't smoke when working with it). Once you find out where a problematic IC is, note its location, type, manufacturer, and date code down, and proceed until you found all the culprits. These findings could go into a database for Apple III users and if a pattern emerges, it might be possible to substitute some of the culprit(s) with faster logic families or faster RAMs / EPROMs.

As a final note, if you do collect vintage computers, aim for getting them into fully functional condition. Otherwise your heirs will treat them as defective electronic junk and to the landfill they would go. Would be a pity.

- Uncle Bernie

I don't own an Apple ///, have never used one, and have no experience with it. I am therefore completely unqualified, unequipped, and uninterested in defending this observation that is not my own. However, I must point out that at KansasFest this year (speaker videos not yet released to the public, but presumably sometime soonish), there was a session (can't remember which, unfortunately, and the year included several Apple /// talks, it being the primary focus this year) in which someone provided a fair bit of heat data and infrared images, and came to the conclusion that, in fact, the Apple /// has no "overheating" issues, nor do the ICs themselves become unseated. However, the official Apple solution to the "overheating" problem, dropping it from an inch or so above the desk, does indeed fix a common issue with the Apple ///, for the reason that the memory board that is connected above the Apple /// mainboard, tended to have connection issues. IIRC (I maay not) the issue was oxidation on the connector, and the act of dropping it tended to resettle the connector and scrape some of the oxidation.

The "oxidation" bit may be a memory fabrication on my part, my memory on that particular is fuzzy. But it did have to do with the connectors between the mainboard and the daughterboard, and not any issue with "IC reseating", nor with overheating in particular.

As I say, I have no personal knowledge or experience either way with this. But it certainly seems worth considering, to me, that just because the solution to the widely-known problem does indeed fix things, doesn't mean that the problem it solves is necessarily the one everyone believes it to be. At the very least, proving that your own machine does in fact have demonstrable heat issues (with an IR imager or some such), in various "room" temperatures, should probably be prerequisite to attempting detailed solutions to safeguard from those heat issues.

Very good background on thermal design and process limits.

A brief note about cold spray: the chemicals banned under the Montreal Protocol for depleting the ozone layer were mainly in three categories: (1) Halons, that is, bromo-chloro-fluoro-carbons used in fire extinguishers;

(2) Freons, or chloro-fluoro-carbons such as the refrigerants R-12 and R-22 and the precision cleaners CFC-113 and AK-225;

and (3) trichloroethane, also called methylchloroform, used as a solvent and propellant in aerosol cans.

Halon and Freon are trademarks but have been genericized and used to describe classes of chemicals.

The freeze sprays sold in the 1980s used R-22, which has a boiling point of around -40 degrees (C or F). After the Montreal Protocol's phase-out of CFCs, freeze sprays switched to R-134A, with a boiling point of -15 °F (-26 °C). It is non-flammable at 1 bar ambient pressure, but it is possible to ignite it under higher pressures, which is why you can sometimes ignite the gas sprayed directly from "duster" cans.

More recently a fluoro-olefin refrigerant called HFO-1234ze has been introduced and used in cans of freeze spray. It is also non-flammable at atmospheric pressure.

Because of super-cooling, manufacturers claim that these sprays can chill components down to -60 °F.

The short version is that all diagnostic freeze sprays you can buy from reputable suppliers are non-flammable when used as directed. There would be no purpose to a flammable freeze spray since it would be unsafe to use on energized electronics. Just don't get one of the "freeze-off" penetrating oils that is supposed to loosen stuck bolts by freezing them!

Chemtronics has a fairly good summary: https://www.chemtronics.com/ultimate-guide-to-diagnostic-freeze-spray