Greetings! I just got my Apple 1 build up and running with help from this forum, Mike Willegal's great guide, and help from Uncle Bernie. Thanks to all for your help!
I was testing the temperatures on my chips to see if any were getting too hot. In general, these original TTL chips definitely run hotter than LS TTL that's for sure. Anyway, I couldn't find any general information about what temperature range to expect for various ICs when working properly in an Apple 1 clone. Since my clone is working, I decided to measure each chip and the voltage regulators with a laser temperature sensor and record the results. I put the information in a spreadsheet if anyone is interested in comparing. I was running Uncle Bernie's diag tests he includes with his PROMs at the time for about 15 minutes before taking any readings and left it going throughout my tests.
Good morning MajorMar,Do you also have a photo of your power supply structure?Did you also measure the exact voltage (maybe even the current) of the transformers output..What heat sink do you have on the LM323k
I myself (with my fingers) do not notice a large temperature development on the ICs.
(I also protected the small voltage regulators with a small heat sink).
If the replica is assembled correctly, you can use it for hours without turning it off and nothing but the LM323 and the heatsink is particularly warm. For the last two weeks I turn on Uncle Bernie's MadHammer or Apple30th in the morning and only turn it off in the evening.
heat is one of the biggest enemy of electronics and considerably shortens their operational life, so I wouldn't take any risk...
As a personal rule of thumb, Replicas and Originals should always be adequately cooled by at least one fan.
Also consider that you are measuring the outer case, the inner junctions will surely be much warmer.
Furthermore, the high heat in the LM323K area can, in the mid-long run, deform the PCB and alter its colour.
The temperatures you measured seem pretty fair to me, except the LM323K which seems a bit too hot.
Ok, it should be self-protected and the internal junctions should work up to 125C, but still your 84C seems a bit too much.
Maybe the input voltage is a bit too high? Or the heat-sink is too small, perhaps?
Could you please post a picture of it?
I remember I measured 65-68C quite a long time ago. I do not remember env temperature but for sure it was between 20 and 25C.
Unfortunately I do not own the laser temp meter anymore, so I can't check again... :-/
Hope someone else can take some measurements...
Claudio - P-LAB
... in the Apple-1 everything runs well within the so-called "commercial" temperature range, 0-70 degC, for which even the cheapest and lowest grade ICs are specified. One exception is the 7812 regulator which may run at about 100 degC if you use a 28V CT transformer and keep the DRAM busy (such as running a RAM test). But in any case there is no need for forced air cooling (fans). If you put the Apple-1 into an enclosure, you need to design it with adequate cooling openings such that the hot air from the regulators can get out in their vicinity with not much resistance to air flow and the "cool air feed" slots should be on the other side of the enclosure such that this "chimney" action draws an air flow over all ICs. If this is properly done, no fans are needed even if in an enclosure. But if you have anything above the LM323K heatsink which blocks the convection from that heatsink, you will have trouble.
A professional way to assess the design of such enclosures is to make them out of corrugated cardboard, maybe with some glued-in reinforcements (wood) along the edges, precisely cut cooling slots, and then put temperature sensors at strategic places and run wires out to instrumentation. After a few iterations you will have a design that works (thermally) and then you can proceed to make actual drawings for the $50000 injection molds ;-)
The Apple-II enclosure is a good example of how fans can be avoided - and the Apple-II draws much more power than the Apple-1.
About these contactless IR thermometers - I use a MicroTemp MT100 which I bought for somewhere below $15 (my search a few minutes ago on Ebay showed me that I paid too much, as there is a vendor who sells them for $12.97 with free shipping in the USA). It works perfectly fine for me, but your milage may vary - all products made in China are suspect, and there may be bad ones. There is no word for "Quality" in Mandarin, or so I was told. Which might explain why some people wrote scathing reviews on Amazon (where it is not sold anymore). But it could be these people just did not understand that if you aim such an instrument at a shiny surface (like the LM323K case) you may get measurements with a large error. These IR based instruments work best with matte black surfaces. Like ICs or the LM323K heatsink (hint, hint !) The mine works fine for me and I am very satisfied. Except that I paid $2 too much !
It is a myth that you can't run ICs "hot" because they would die like flies. Even the military grade ones which are specified for -55 deg C to +125 deg C die temperature come from the same wafer process as the cheap commercial ones (except for more inspections and much, much more paperwork ! --- this is how the $600 hammer happens). So as far as the silicon is concerned, there is no undue risk when running at 70 deg C or so. There are some effects like formation of intermetallic compounds at the gold / aluminum bond wire interface and electromigration in the aluminum interconnect layers which accelerate at hot temperatures, but these are far too slow to cause concern for typical Apple-1 use patterns. We are talking about several 10,000s of hours running at 125 deg C before these effects may kill an IC (if it was properly designed ... some designers don't understand electromigration and some ICs may have been designed intentionally to fail from electromigration after X amount of device hours - this is a very well studied and understood effect, much like the 1000h designed-in life limit for incandescent light bulbs, an international conspiracy by the light bulb cartel dating back to the 1920s --- only communist made light bulbs lasted longer as their manufacturers did not need to follow the cartel rules).
So don't be afraid of ICs running hot ! I'm writing this very long post because just recently some concerned builder using my IC kits (he had full success at 1st power on as expected !) noticed that the 74S161 and 74H04 I substituted for the standard TTLs do run hotter than standard TTL (as they are expected) and he started to search for shorts before asking me. Duh !
My "golden" rule from 40 years of professional experience: if you can touch your ICs with your fingers and it won't hurt (does not trigger the retraction reflex) and it does not draw a blister, the IC is fine. Some savvy engineers and lab technicians use fingers wetted with spit and listen for the sizzle it makes if the IC is too hot and could hurt. I would not recommend that "finger test" for power components like regulators, though, with one exception: if you keep your fingers on those regulators while turning on the power from the cold condition, and one of them gets too hot too fast, you have a chance to quickly turn off the power again before some bigger damage happens elsewhere. In most cases, this is an IC, TTL or DRAM, which was put into the socket backwards (turned 180 degrees) and these, if not caught early, tend to get so hot in their death throes that they melt the plastic of their IC socket. No good ! Hence, my "feel the regulators" trick !
Happy Apple-1 building and no worries !
Just to be clear, the spreadsheet is just for reference. I am not having any issues that I know of so far. But I thought it might help someone else who is unsure if their temps are way out of whack or not.
I did test the transformers before using them and they seem to be fine. The heat sync on the LM323K is a little thinner than the ones pictured here http://www.rlvintage.tech/lm323k.html#a1lm323k, but it is running well within the guaranteed tolerances of 125 deg C.
Thanks for confirming those temps are OK.
I will take more temps running other programs to see if they are drastically different when not running Uncle Bernies self test page.
Note that as Uncle Bernie stated, the IR temperature meter is calibrated to a specific material emissivity (ε), which is how perfectly the object emits radiant energy. If the expected ε is different from the material being measured, the reading will be inaccurate. Polished stainless steel has ε=0.20, much lower than the ideal "black body emitter" which has ε=1.00. One way to ameliorate this is to stick a small piece of black electrical tape on the object, which has ε=0.96. FLIR even has an application note on this technique:https://www.flir.com/discover/rd-science/use-low-cost-materials-to-increase-target-emissivity/
These concerns don't apply to contact temperature sensors (thermocouples, thermistors, and RTDs), although they have other issues related to thermal bonding to the object under measurement.In either case, the external surface temperature of the package will also be different than the junction temperature Tj, but there is a simple formula to correct that. Tj = Pd · θjc + Tc where Pd is the power dissipated, Tc is the measured case temperature, and θjc is the junction-to-case thermal resistance which will be written in the device datasheet.
An infrared temperature meter is not a particularly accurate measuring device. I used to use it to measure the temperature of pipelines to power plants, but I gave it up over time. The readings change a lot in the areas where the pipes are painted. Much more accurate readings will be taken from an ordinary alcohol or mercury thermometer.
Update on LM323K: Uncle Bernie was kind enough to send me a thicker heat sync for the LM323K, however after a similar amount of time I was still seeing a maximum temp of 75.4C. So, a slight improvement, but nothing very dramatic. YMMV
Uncle Bernie answers:
This was expected. I did not send you the thicker heat sink because it would make the LM323K run dramatically cooler, I just wanted to help you to get the "looks" of our Apple-1 right.
The Physics behind this boil down to surface area. The cast aluminum thicker heat sink does not have a much larger surface area than the thin stamped sheet metal one. But it has more thermal mass so it would work better with pulsed loads. The Apple-1 however has no such pulsed load. Hence, the fat and heavy heat sink has almost no benefit other than more correct "looks".
Just for the benefit of builders out there, I found a cheaper substitute for the Wakefield 680-125 which costs ~$12.
The substitute is Aavid, Thermal Division of Boyd Corporation, part #568303B00000G, see Digikey part #HS531-ND.
This one costs only $6.21 (about half of the Wakefield) and is almost identical except the top surfaces of the fins are somewhat irregular (as if the casting form was open there). You can spot the difference only if you compare the two.
A casual observer would not spot this small imperfection, I think. (you get what you pay for, remember ?)
It also has been brought to my attention that Digikey has the Sprague 39D 2400uF/25V "big fat blue" filter capacitors back in stock. Digikey part # 39D248G025JL6-ND. Price $16.01 each. (Ouch !)