i agree that the machine should be fully self-describing, like darius bacon's 'book in itself'. of course, that won't help you if it won't turn on...
rather than just conformal coating and seals, i'm thinking i'll not just conformally coat the boards, but also pot the whole assembly in something like silicone or toilet-ring wax, so that the potting material can be removed, but otherwise it's trivially ip68. except for the keyswitches, but the keyswitches can be a magnetic type that doesn't have any exposed electrical contacts or any springs
solar-powered devices generally don't have to worry too much about cooling, though. or if they do it's from the heat they passively absorb from the sun, not from anything they dissipate internally. if your solar panels are 10% efficient (the best you can get out of amorphous silicon, which is the only kind that works under indoor lighting) and cover 30% of your device, they only harness 3% of the illuminance, which is later converted to heat in the circuitry. the other 97% is either reflected or converted into heat immediately. so the heat produced internally is only about 5% of the heat it has to deal with. now think about how much heat you feel on your skin from indoor illumination; we're talking about 5% of that
socket-mounted chips tend to induce a lot of unreliability, so i don't favor them. desoldering surface-mount chips is a bit of a pain but still done routinely in cellphone repair shops around the world. but i want it to make it to its design lifetime without replacing components, and one reason for that is that i don't want to depend on the economy. (consider what components were commonplace in 01971 and how many of them are hard to obtain nowadays — and that was 53 years of relative peacetime. imagine trying to get parts today in moscow.)
unlike batteries, chips do have a long shelf life, so you could conceivably stockpile chips today for future repairs. but if you're going to do that, probably the best place to stockpile them is on circuit boards inside the device, so you don't have to solder anything to fail over to them. that's not applicable for every device, but fortunately things like voltage regulators have very low failure rates, and you can still probably build in redundancy at a higher level. that's what we did when i worked on satellite systems: of n identical systems onboard, we only needed one to survive to run the satellite. there were even multiple power buses, so that even a failure that shorted a voltage rail to ground would only disable part of the satellite
as i understand it (not having taken the measurements yet myself) over-rating is especially important for bypass capacitors in this context — not because they're prone to failure (unless you're using tantalums) but because capacitors have enormously higher leakage near their rated voltage
if you want to build such a thing and have a mechanical charging option, i think a pullstring like 20th-century children's windup toys is the best option. it's compact, the spring constrains how much force you can apply to the generator and runs it at a predictable speed, and the pullstring grommet constrains the direction of force applied to the mechanism so that kofi annan can't break it by applying side-loading to your crank handle like he did with olpc. but i don't think electromechanical generators are likely to fit into the reliability budget
for screws, i think protruding screws would catch on my pockets. but there are lots of ways to fasten things together in reopenable ways, especially if you don't need seals. if you did want to use conventional milled-head screws, well, i filed a flathead screwdriver out of a bolt a few months ago. i used a file, a vise, and a steel bolt. but exposed screws would tend to rust unless you made them out of titanium or something, which isn't an option for me
a metal case has advantages and disadvantages. being able to take input power inductively (qi style) or communicate over rf could be useful, which you can't do with an all-metal case. metals are not very chemically stable, except for gold, silver, platinum, iridium, palladium, lead, nickel, aluminum, titanium, tin, and chromium. most of these have major disadvantages of their own, though titanium would be a pretty good option. many plastics are very chemically stable, including polypropylene, polyester terephthalate, silicones, and epoxy resins, and the epoxies in particular can be glass-fiber reinforced up to steel-like strengths and stiffnesses. so that was my plan
Pocketable wasn't in my mental spec, agree this is a good idea. You can also inset them.
Marketing a portable as Kofi-proof is a neat marketing angle ;)
Depending on your radio frequency perhaps you could put your RF behind your screen to get signal out, or use a slot as the antenna, or have a fold-out or telescopic antenna, or have a screw-on antenna port exposed when you open the thing, or some combination thereof.
Aluminium's fine for cases, cheap, light, castable, extrudable, cheaply machined, and readily anodized.
Steel won't rust for ages if it's of a decent chemistry and either treated and/or oil is used when the thread is mated. Even steel that appears heavily rusted after many decades of total abuse can be restored quite easily to a good state in many cases. I mean people had multi hundred year old wooden structures ... I don't think an embedded steel thread with an oil surface coating is going to disappear overnight. As a learning project I recently restored a 1980s drill press, that's basically 40 years old which is ~most of your target length, it had apparently been left with exposure to moisture for 20+ years, and was mostly intact. The net result of the learning was it's not worth restoring old drill presses, but I sure learned a lot!
For maximum chip shelf life, you want to store them in nitrogen or some similar inert gas.
With Wikipedia and survivalist content modules you could market this to the prepper crowd. Of course long distance radio would be an add-on. As would the radio direction finding hardware, encryption, etc.
I still like the pneumatic idea. 100% of existing bicycle pump hardware becomes your viable charging interface. Speaking of which, a low profile wheel based generator would be a good add-on module.
i hadn't actually considered machining it out of an aluminum billet. you might be right that it would be adequately stiff, even if it's much less stiff than glass-fiber-reinforced epoxy. it's much heavier than epoxy, but only slightly heavier than glass, and of course most of its grades are enormously less brittle
oh, i don't know why i didn't answer the stuff about steel and preppers and pumps; maybe it was a brain fart or maybe you edited it in later. i'll try to answer later
Yeah it was a few edits. I often stay on a theme awhile once consciously captured, resulting in a re-read and new ideas/tweaks. Ninja skill: trainable psycho-inertia, AKA "focus", now perhaps ostracized as a 'spectrum member' activity, I believe it was merely associated with clarity of thinking and general education in the 19th and early 20th centuries.
so, with respect to steel, i think a steel case in a sweaty pocket will rust rather quickly and comprehensively, even if it's a good steel. often alloying elements that improve steel's mechanical properties make it more prone to rust rather than less so
i'm guessing the drill press was not in working condition after the 20 years of moisture, and i'm assuming you mean something like 'in a basement with relative humidity that reached 100%' or 'with the bottom resting on moist soil' and not 'under dripping water for 20 years' or 'underwater for 20 years'
i don't really have a good way to machine steel anyway, and it's inconveniently heavy. aluminum is amenable to wood rasps and drills
i think probably embedded in silicone or paraffin wax or something would be better than just stored in nitrogen. gottta be careful of static buildup tho
while i sympathize with preppers in some ways (i, too, value autonomy) i think they might be a hard group to market to. a friend of mine occasionally reads survivalist board forum and there's a lot of toxic ultra-right-wing stuff there; i don't want to be lynched for not supporting trump
you can of course do secure encryption on any general-purpose computer unless you're trying to defend against side-channel attacks. i ran pgp on a 286, and we have more efficient algorithms now
long-distance radio is potentially interesting but i agree that it's probably a different piece of hardware; short wavelengths can only go short distances unless they're bouncing off the moon or something, and you need a long antenna to transmit long wavelengths efficiently, although efficiency isn't a significant concern for reception (at transcontinental-capable wavelengths radio noise is a bigger concern than noise generated inside your amplifier). so for long-distance radio you probably need a large, fixed antenna installation rather than a pocket computer
my concern with pneumatic-to-electric power is that it's probably hard to make reliable. but i don't really know
For the radio, ask a HAM. I'm sure they can workshop something out of nothing and turn a clothes line in to a global command center... anyway yes, the size of long distance antennae will prohibit having them internal. One could include an SDR module but why not make it an add-on?
Since you only need a very small amount of current, something like a small compressed air canister could then presumably store many months worth of power. It would need to activate the air source for recharging only sporadically. This could be done using physical input from the operator and an NC-type valve in order to sidestep the flat-start problem, with a parallel option for electronic actuation.
Perhaps steampunk social networking would be a better market. One imagines darkly clad figures sneaking around urban environments clandestinely sneaking power top-ups from from random vehicle tyres...
i don't think it matters whether the efficiency is 100% or 1%; as it happens, i just pumped up my bike tire from dead flat earlier tonight, and i did on the order of 2 kilojoules of work with a hand pump to do that. 1% of that would be 20 joules, which would be 20000 seconds of full-speed 1-milliwatt zorzpad usage
and conceivably the pneumatic storage option could help to solve the problem of 'where do you store those joules until you're ready to use them?' this is a problem at the electrical level because batteries and electrolytic capacitors are not long-lived, and non-electrolytic capacitors max out at about 1 microfarad for ferroelectric mlccs. 10 volts at a microfarad is 50 microjoules, so storing an entire joule in long-lived capacitors would require about 20000 capacitors
but you still have the problem of how to convert the pneumatically stored energy into electrical energy at milliwatt rates (plus or minus a few microjoules, anyway) with decades of mtbf
if you're willing to require manual labor to pump up the palmtop with energy, though, a pullstring can store the energy in a steel spring, which can easily hold several joules and parcel them out at milliwatt rates to an electromechanical generator, with efficiencies in the neighborhood of 90% if you care about that. maybe you could even make it reliable, but it isn't obvious how. (grossly overspecced gears on bronze oilite bushings, maybe, driving a grossly overspecced generator?) a single heavy pull on a bowstring routinely stores around 100 joules, so i think a pullstring could be competitive
a manually powered computer like this could harness more energy at night than the purely-solar-powered zorzpad can. in the daytime, though, the sun provides more energy than you can comfortably provide by hand. consider a 150×90 pocket-sized notebook; in full sunlight it's collecting 20 watts, the equivalent of that heavy bowstring pull every 5 seconds. if only 30% of the notebook is covered in solar panels and they're only 10% efficient, you're still collecting 600 milliwatts, the equivalent of that heavy bowstring pull every 2½ minutes
that all makes sense but i think if you're aiming for longevity then separating the active mechanical bits (mtbf low) from the electronic bits (mtbf high) is a good idea, just because physical effort in proximity to breakable things is likely to break them faster. this would be a case for not building in the pullstring and instead using an external pump. but you could have both, as well. supercaps are good for capturing fleeting charge but don't store for too long. they do store longer than regular caps, but they leak current over time. an air tank can be 'on board' with high longevity and low mtbf. the connections/seals the probable locus of problems.
yeah, i guess i don't really know how long-lived supercaps are. just as a rule of thumb i figure that they'll probably fail soon just because they're full of liquid, but that's not an infallible rule; maybe they're long-lived
a 10-farad supercap at 2.7 volts could hold 36 joules and would still fit in your pocket, so you could totally charge it up with a pullstring. https://www.digikey.com/en/products/detail/tecate-group/TPL-... is 10mm diameter, 31.5mm long, and it's rated for 1000 hours of life at 85°. the datasheet projects that at 25° it will last 10 years with ±30% capacitance change and no more than 2× increase in esr. 53 years is 465000 hours but maybe at a low temperature and low voltage it would last that long? maybe it'll be 32× increase in esr but not a problem?
i agree that the machine should be fully self-describing, like darius bacon's 'book in itself'. of course, that won't help you if it won't turn on...
rather than just conformal coating and seals, i'm thinking i'll not just conformally coat the boards, but also pot the whole assembly in something like silicone or toilet-ring wax, so that the potting material can be removed, but otherwise it's trivially ip68. except for the keyswitches, but the keyswitches can be a magnetic type that doesn't have any exposed electrical contacts or any springs
solar-powered devices generally don't have to worry too much about cooling, though. or if they do it's from the heat they passively absorb from the sun, not from anything they dissipate internally. if your solar panels are 10% efficient (the best you can get out of amorphous silicon, which is the only kind that works under indoor lighting) and cover 30% of your device, they only harness 3% of the illuminance, which is later converted to heat in the circuitry. the other 97% is either reflected or converted into heat immediately. so the heat produced internally is only about 5% of the heat it has to deal with. now think about how much heat you feel on your skin from indoor illumination; we're talking about 5% of that
socket-mounted chips tend to induce a lot of unreliability, so i don't favor them. desoldering surface-mount chips is a bit of a pain but still done routinely in cellphone repair shops around the world. but i want it to make it to its design lifetime without replacing components, and one reason for that is that i don't want to depend on the economy. (consider what components were commonplace in 01971 and how many of them are hard to obtain nowadays — and that was 53 years of relative peacetime. imagine trying to get parts today in moscow.)
unlike batteries, chips do have a long shelf life, so you could conceivably stockpile chips today for future repairs. but if you're going to do that, probably the best place to stockpile them is on circuit boards inside the device, so you don't have to solder anything to fail over to them. that's not applicable for every device, but fortunately things like voltage regulators have very low failure rates, and you can still probably build in redundancy at a higher level. that's what we did when i worked on satellite systems: of n identical systems onboard, we only needed one to survive to run the satellite. there were even multiple power buses, so that even a failure that shorted a voltage rail to ground would only disable part of the satellite
as i understand it (not having taken the measurements yet myself) over-rating is especially important for bypass capacitors in this context — not because they're prone to failure (unless you're using tantalums) but because capacitors have enormously higher leakage near their rated voltage
if you want to build such a thing and have a mechanical charging option, i think a pullstring like 20th-century children's windup toys is the best option. it's compact, the spring constrains how much force you can apply to the generator and runs it at a predictable speed, and the pullstring grommet constrains the direction of force applied to the mechanism so that kofi annan can't break it by applying side-loading to your crank handle like he did with olpc. but i don't think electromechanical generators are likely to fit into the reliability budget
for screws, i think protruding screws would catch on my pockets. but there are lots of ways to fasten things together in reopenable ways, especially if you don't need seals. if you did want to use conventional milled-head screws, well, i filed a flathead screwdriver out of a bolt a few months ago. i used a file, a vise, and a steel bolt. but exposed screws would tend to rust unless you made them out of titanium or something, which isn't an option for me
a metal case has advantages and disadvantages. being able to take input power inductively (qi style) or communicate over rf could be useful, which you can't do with an all-metal case. metals are not very chemically stable, except for gold, silver, platinum, iridium, palladium, lead, nickel, aluminum, titanium, tin, and chromium. most of these have major disadvantages of their own, though titanium would be a pretty good option. many plastics are very chemically stable, including polypropylene, polyester terephthalate, silicones, and epoxy resins, and the epoxies in particular can be glass-fiber reinforced up to steel-like strengths and stiffnesses. so that was my plan
why 53? i just like 53