It’s a simple goal: build a waterproof box full of environmental sensors that can run continuously for the next century. OK, so maybe it’s not exactly “simple”. But whatever you want to call this epic quest to study and record the planet we call home, [sciencedude1990] has decided to make his mission part of the 2019 Hackaday Prize.
The end goal might be pretty lofty, but we think you’ll agree that the implementation keeps the complexity down to a minimum. Which is important if these solar-powered sensor nodes are to have any chance of going the distance. A number of design decisions have been made with longevity in mind, such as replacing lithium ion batteries that are only good for a few hundred recharge cycles with supercapacitors which should add a handful of zeros to that number.
At the most basic level, each node in the system consists of photovoltaic panels, the supercapacitors, and a “motherboard” based on the ATmega256RFR2. This single-chip solution provides not only an AVR microcontroller with ample processing power for the task at hand, but an integrated 2.4 GHz radio for uploading data to a local base station. [sciencedude1990] has added a LSM303 accelerometer and magnetometer to the board, but the real functionality comes from external “accessory” boards.
Along the side of the main board there’s a row of ports for external sensors, each connected to the ATmega through a UART multiplexer. To help control energy consumption, each external sensor has its own dedicated load switch; the firmware doesn’t power up the external sensors until they’re needed, and even then, only if there’s enough power in the supercapacitors to do so safely. Right now [sciencedude1990] only has a GPS module designed to plug into the main board, but we’re very interested in seeing what else he (and perhaps even the community) comes up with.
I think neither the plastic box nor the rubber seal is going to last 100 years. You’d have to use glass and probably something like plumber’s putty or stained glass putty to seal it.
I second that opinion. Perhaps finding a way to completely enclose it in glass in something like a light bulb. Think of a radiometer. The vanes are somehow inserted into the globe before the bottom is shrunk back and sealed. Guaranteed to be sealed for 100 years though you might want to put it in a second enclosure that has a tempered glass top on it so that hail/ falling branches, etc.. won’t break it. Also, keep in mind that solar cells in series only output as much current as the cell that is putting out the least current. If a bird poops in a place that causes a shadow over your mini panel, you lose data. It would likely be better to have a few panels in parallel which are spaced away from each other to minimize that sort of shading.
I’d be tempted to steer away from any packages with a fine pitch and stick with DIP as that what tends to survive for a long time just look at all those old Atari 2600s still around.
Wouldn’t electrolytic caps be the obvious bottleneck here though?
Had exactly the same thought!
Solder Creep ?
if you use lead-free, then yes
tin whiskers
I have been dealing with a similar proyect in my mind for collage but I have one concern: how do waterproof the sensors?
Really comes down to the type of sensor, if they can do the work despite being sealed (perhaps potted, covered in oil or just inside a sealed box) then it just comes down to characterizing what effect the sealing has on them. For example a spectra-graph can work behind anything that passes light, but its going to affect the reading and maybe write off certain bands of the spectrum outright.
For anything that needs to be exposed to air like a gas sensor of some kind you really can’t properly waterproof them, you can seal the entire board but the working area which will help. So then comes down to management – keep the sensor hot enough that water won’t condense on it, covered from direct rain/ spray etc.
Potentially you can get cunning with permeable membranes that will pass what you are interested in but keep water out, will no doubt lower the responsiveness of the sensor but it can be done. One interesting material for this is sintered glass beads – pick the right size balls and the spacing won’t let water in but its still permeable to (most) gases, and being a glass is very resilient.
GoreTex membrane across an opening on the bottom of the box. But I don’t think it would last 100 years.
Flash memory slowly loses its charge…
Does anyone think we will still be using GPS in 100 years though?
The GPS week number just did a rollover which confusedmany older GPS devices. The next one is in about 157 years, so no need to worry about that.
I thought the rollover occurs about every 19.x years.
Yes, however the standard changed and that is no longer the case.
The likelihood of GPS remaining backwards compatible for 100 years is low. SATNAV/TRANSIT only lived for 30 or so.
I would be surprised if GPS is replaced any time soon, like FM radio its too effective to easily kill off. Its possible that the frequencies etc of future ‘GPS’ could be different, but that is unlikely as the electromagnetic spectrum is so regulated it would take a great deal of effort to change it.
Other positioning systems are not likely to take over outright as using the wifi and mobile masts to position yourself etc only works when there are wifi/mobile signals. If the satellite internet clusters do get built for sufficient coverage maybe GPS could become redundant using the known orbits and gps style calculations, but it seems highly unlikely to outright replace GPS, as the millitary will still want the proven reliable gps system.
I know militaries move cautiously, but we’re talking about a hundred years here. I’m sure even the most cautious military would admit something’s reliable after thirty or fifty years of service. :-)
You would think so, but you only have to look at the stupidities of history to know it often doesn’t work that way. How long did warships get built with sail as their primary (or even only) propulsion even when steam engines had proven their effectiveness for long distances, reliability and performance for example. And as a substitute tech doesn’t even exist in theory yet much of the 30-50 years of service before it might render GPS obsolete might not even happen in this 100 year period.
Also hard to replace GPS with anything that isn’t functionally damn similar if not identical.
I do agree with lots of comments about potential backwards compatibility issues with new generations of the tech, but if your GPS device is a reprogramable number cruncher it shouldn’t be an issue to update the software to work with GPS version N. For me if GPS (or at least something that works in much the same way on the same frequencies etc) is not going to last it will be because we have been too busy knocking ourselves back into the stone age or rendering the planet lifeless to keep working satellites in orbit.
So in this situation if you intend to have accurate position without any software updates in the future it really needs to use inertial reckoning, a superb clock and a star map to plot its movement and location. Being entirely self-contained it then does not rely on anything that might change in the world. I wonder what you could use to get the accurate clock for 100+ years.. (I know we have some damn fine clocks built but none of them are exactly affordable or attainable as an individual)
If a localized conflict involving countries that have anti-satellite/GPS jamming technologies and the US in the next 100 years, they might be temped to disrupt GPS. Russia and China have their own alternatives GLONASS, BeiDou.
There is also the possibility of future versions of GPS as they’ll need to replace them at end of life. Who know if they will be backward compatible for 100 years.
“Who know if they will be backward compatible for 100 years.”
Pencil and Paper, but only if you print, no cursive.
Better learn to write in Chinese or some other ancient languages that have survived thousand years. Who knows if English would become Latin in a 100 years. People can’t even use the correct words these days, so chances what you write could be gibberish in a 100 years. e.g. then/than, effect/affect etc.
Thousand years and no one else uses your language. What does that tell you?
Agree with that. Local schools around me only teach cursive for one week. The kids practice writing their names and that’s it.
This was me. And I even forgot how to do my name like the following year. My signature is just two meaningless scribbly lines.
In similar concern I don’t think the caps are going to make it that long. With older radios the refurbishment kits are loaded with electrolytic capacitors for replacement. They tend to leak after a few decades and lose much of their capabilities in addition to damaging circuit boards. Mounting them upside down might help a little, but I wonder if you could go even lower power somehow. This would allow you to use non-electrolytic capacitors potentially trading the increased space requirement for component longevity. I found this PDF that might be helpful to you. In my quick review of it I was surprised how much of an effect temperature would have on the longevity of electrolytic capacitors. The information may be of some use to you.
https://www.illinoiscapacitor.com/pdf/Papers/Life%20expectancy%20of%20Aluminum%20electrolytic%20capacitors.pdf
I’d focus on the PCB — the most common cause of failure has to be mechanical/chemical. And the caps? Most amps need their electrolytics replaced after 10-20 years. Proper, rigorous engineering is the order of the day here. NASA knows about this stuff …
“NASA knows about this stuff …” its not that they know. Its that they have unlimited cash to buy the best of the best components. most people are not willing to spend $260 for a typical $0.45 cap. But hey its Nasa (Not Another Stupid Adventure)
Space qualification is not a matter of having unlimited cash to waste. It’s a 45 cent capacitor that’s been thoroughly tested to ensure that it will not fail and the paperwork (pedigree) to show it. Because it’s annoying to have expensive stuff fail because you bought capacitors from BangGood.
“Not Another Stupid Adventure” well those “stupid adventures” have done the following:
1. Put men on the moon.
2. Sent probes into deep space that are still working 40 years later.
3. Conducted huge amounts of experiments that have given us invaluable insight about the rock we live on.
4. Explored the surface of another planet.
5. Helped build a manned outpost in space.
6. Driven the development of metallurgy and material science resulting in many technologies we take for granted being feasible to produce.
7. Driven the development of semiconductors and computers.
8. Deployed telescopes that peer into the deepest parts of space we can see allowing us to understand our universe better.
WTF have you done that is even remotely as grand or beneficial to humanity?
And yeah NASA does know what their doing…they know when and why you need the $260 cap vs the $0.45 cap. The concept of high reliability clearly escapes you. They’re not building GD cat food dispensers…they’re building MF spacecraft operating in one of the most harsh environments know to mankind.
“its not that they know”? Of course it is.
Also watch out for tin whiskers if you are using RoHS solder. :P There are reasons why critical systems still use good old lead based solder.
And if he fails, who will be around to complain?
It’s a bold claim to say some piece of electronics will last 100 years… But who cares? Really? It doesn’t matter if it’s 100 or just 10 years. As long as it works long enough to get enough measurements to get some useful data it’s good enough. If it last longer you’ll only find out that the link has problems because the hf situation of the world around it changed or the protocol or receiver are no longer working or supported.
But it’s the effort that counts and it’s a nice puzzle to think about. Have fun, that’s the most important.
sometimes the journey it’s more important than the destination.
100 years? Clickbait…
Those connectors better be gold plated and the caps should be replaced with tantalum or other solid state types. Also dry nitrogen filled air tight container, would help a bit. That’s the kind of stuff one would use for mil spec that expects 10-20 years service life.
Extraordinary claim requires extraordinary proof. A spec that you cannot demonstrate/test is BS and would be consider a failure in implementation. Consider to include some accelerated ageing test if you want to be serious and not just some kind of BS.
Leave tantalums out of the equation as well. They fail spectacularly with age
Would it matter if they are gold plated? If it was filled with dry nitrogen (or whatever else), the contacts couldn’t oxidize, right? Even so, would 100 years be enough for the oxidation on contacts be enough to sever the signal?
Will a solar panel last 100 years? I heard 10-15 years, tops. It’s not that they completely fail, the output degrades over time. You still get power, just not what you are expecting, or need. Global Warming is going to speed that up quite a bit as well. Heat is an electronics killer, wonder what else we can expect to fail… Guess it won’t really matter, couldn’t afford the electricity to run anything but the most essential devices.
100 years is a lofty goal and I applaud the attempt, but there’s a fair bit more that needs to go into a design to get there.
Others have rightly pointed out the perils and pitfalls of connectors and electrolytic caps, the lifecycle of GPS and data communications, etc.
The other one that I’d be very concerned about is drift. Whatever the microcontroller is using for internal clock _IS_ going to drift. Which way and how much are unanswered questions. Any sensor that is read with time or frequency is going to have a new error term, that includes analog voltages in many microcontrollers. To some degree that could be mitigated by using GPS time to adjust for drift, though there are problems there too.
The sensors will have other drift terms as well. Age, temperature, barometric pressure, shock effects, UV exposure, contaminants, just to name a few.
Even if the electronics are technically functioning, accurate, useful data may not be possible without some plan for periodic calibration.
Don’t let accurate calibration put you off. Tree rings are good enough….
Well, there you go! A solar-powered sensor platform with a 100+ year lifespan – trees.
FLASH memory can fail too as they are typically only spec for 20-ish of data retention.
The topic of the longevity of electrical components and enclosures is a very interesting one, and what little I know of it leads me to conclude that this otherwise interesting project is as feasible as making a house out of bananas. Any useful pointers to more information on this topic would be most welcome.
This reply is meant to be constructive help to the project.
So what can go wrong with age … caps, sensors , connectors, solder joints, pcb, solar cell, crystals and let me add that list
The IC’s themselves will age also. Check the lifetime spec on commercial parts, it varies but is far from 100 years.
Sure you can do things that might help extend the life of the parts, use power management features of the chip, avoid temp cycling to reduce physical stress, operate at low temp , reduced voltage , lower frequency and so on.
For those curious and wanting to learn a bit more :-
DDG terms : EM Electromigration, NBTI Negative bias temperature instability, HCI Hot carrier injection , TDDB Time-dependent dielectric breakdown, a couple of articles on aging in general.
https://semiengineering.com/chip-aging-accelerates/
https://semiengineering.com/chip-aging-becomes-design-problem/ -some good links in this one.
I guess I would think about adding UV (box, solar cell), ACID rain, ESD , EMP and lightning protection too..
I think it was a miss print… ie instead of 100 it should have been 10….
Let’s pit the U.S. against this plastic box !
Who will survive the next 100 years ?
I think I put my money on that box ;)
That 2.4 GHz radio is unnecessarily costly in terms of power usage. Reporting sensor data is more suited to a low power sub-GHz radio like RFM69CW or RFM95 (LoRa). They have a much smaller bandwidth, which is suitable for periodically reporting sensor data, so they can operate at a much lower power and incidentally have a much greater range.
Too bad all the wackos who made lousy predictions won’t be around to make fun of. The maker should add a recording that does that to the build.
I’m not sure of the longevity of the crystal power cells, but I do know that they last a long time and could power this thing even in absolute darkness. Might be something to consider. You’d probably want to isolate the cell in some form though, as they tend to expand as they age. Anyway, just a passing thought, I’ll leave it up to others to point out the flaws.
https://www.youtube.com/watch?v=3cJFkIDImuM
100 years? you need some serious tech support. For starters are wlan standards this chip support even in use in 20 years, not to mention 100? Not to mention firmaware updates witch you need something that can actuallly talk to chip in language it can understand.
Id give this design 12 years… There is no redundancy, not great component selection, solar panels are not rated for anywhere near that amount of time, the memory in the micro will fail, water will ingress with that plastic, tin hairs will be an issue and none of those sensors are rated for long term use
Very long lasting design is an art few are good at but there are some tricks
Use a processor with a large transistor size, have a second system kick in if data stops being sent by the first, use properly raited sensors, use large footprints, zero tin solder, gold pads, ceramics for all capacitors preferabley in a 2 parallel 2 series configuration, don’t power it off solar, encase it in glass, don’t pretend that legacy GPS will be around in 2119 because it likely won’t be (they are not even 100 year capable systems)
Also an automatic reset system is key
I think that thermal cycling has to be taken into account if they want the electronics to last; the electronics themselves must be in a location that only changes temperature on a monthly, or yearly basis, rather than per-day. Using leaded IC packages can mitigate some of the dangers as compared to QFN or DFN packages, as the leads can take up the strain, but I think a big problem will be surface-mount resistors and capacitors.
considering that the technologies that we will have available in the next 100 years could be very, very different (indeed, we run into the problem with the lack of tape drives to read old NASA data), I’d probably want the system to record the data in a human-readable format, like printed/scored heavy paper or metal tape.
Agreed actually reading the data as technology moves on isn’t going to be trivial.
There is also problems of data density – to be directly readable by human senses you really can’t fit much data in a box of reasonable size. Where if you accept that its human readable under high magnification you get around that problem. But need a much more precise (so likely hard to make durable) data recording device.
Also worth asking the question of what data you are gathering. 100 years from now people may still call their language the same thing but its a sure bet it will have changed considerably. So for the data to be useful you may have to figure out how to save it in a way that means something to a future generation that has never heard of Monty Python.
It would be easier to grab all these good ideas to build a 40 years device, and put 3 of them in the enclosure.