This is continuation of https://hackaday.io/project/146880-power-for-a-century, towards one particular incarnation of a household gadget - a solar-powered clock - capable to run unattended for a century or more. The challenge is in the restricted choice of parts used for the construction, so that they have a chance to survive and function properly after 100 years.
The basic premise of the project is, that during daytime, there's relatively plentiful source of the primary energy (solar panels can be bought in larger sizes if necessary), but the key issue is to provide enough energy to sustain rudimentary functionality during night. Even if the time is not been indicated during night, storing energy for timekeeping alone may be challenging enough.
The "root" project concluded, that for the given goal it is necessary to omit parts based on "chemical" principles. This largely disqualifies the conventional energy harvesting models, which rely on accumulators and/or supercaps.; even electrolytic capacitors are off. Omitting exotic principles such as mechanical storage, we are left with a short list of "solid-state" capacitors. This project will try to find a way to make the best use of them.
While the discussion in the "root" project identified also other potential issues (e.g. enclosure), for sake of simplicity, we are not going to be bothered by these here.
Buy 100uF X5R 4V capacitors. There are larger capacitances around, but it's not necessarily good to push things to the edge - larger capacitors in the same package means thinner dielectrics which may mean higher leakage. Try several manufacturers, maybe also different dielectrics.
Measure their leakage (and dielectric absorption) - after hours of charging.
After finding out that the project could be accomplished in a "humanly" simple way, unfortunately, there was not enough time left to acquire the parts - a bunch of 100uF+ ceramic capacitors and a ultralow-consumption RTC. (Btw., turns out, theadvertised 40nA RTC module is not yet available from the "usual outlets", but another module consuming only 60nA is, at a reasonable price, and it even features a calibrated-RC-mode when the consumption drops to mere 17nA!) So, using a couple of 10uF 0805 ceramic capacitors at hand, and the built-in RTC of STM32L476 on the L4-DISCO board, I proceeded at least with a basic proof-of-concept project.
The VBAT pin, supplying the built-in RTC in mcu, on the DISCO board is not brought out separately but is connected to VDD - luckily through a resistor, R90. This has to be replaced by a 1N4148 diode, and the 10uF capacitors have to be added in parallel to the existing 1uF C79:
1N4148 has a leakage of a few nA at 3V reverse voltage. While charging (i.e. mcu is supplied through VDD), mcu supplies the RTC internally, thus no current flows through VBAT, so - after several hours of sunlight/charging even the dielectric-absorption-capacitance of the storage capacitor is fully charged, so only leakage flows through the diode, which we hope will be only a few nA. At that current, the forward voltage drops towards tens of mV, which is something we probably may sacrifice - it's not worth to experiment with exotic low-leakage transistors, or even relays, in place of a simple diode.
A good thing about the ST devboards such as the DISCOs, is, that ST provides extensive documentation, including schematics, Gerber files and even a detailed BOM - so modding them is relatively easy.
While we assume enough power for the "normal" run during daytime, there's no reason to waste energy if it can be avoided. So the firmware first initializes the internal RC-oscillator-based clock (MSI) and switches it into 400kHz mode to reduce consumption, which enables to switch from the standard internal regulator to a low-power one. These few simple steps brings basic runtime consumption down to cca 250uA. This can be conveniently measured by removing the jumper from JP5 and routing an ammeter from any of the outer pins to the middle pin - the middle pin of this jumper connects directly to the mcu's VDD; the ON position pin goes from the onboard current measurement and the OFF position pin is directly the power supply input bypassing the current measurement (using the onboard extra microcontroller for measuring current involves talking it through I2C in a protocol which is not very well described and has to be mostly extracted from the demo sources, so I left it for another time).
Then the firmware checks, if the 32.768kHz LSE oscillator is already running and if RTC is enabled - if yes, everything is left as it is; if no, the oscillator is started, RTC enabled and initialized. There is no provision to adjust time, yet; the joystick on the DISCO board is to be used for this purpose.
Then the LCD is initialized, and then the main loop is entered. In it, the current time is read out of RTC and displayed, and then the mcu goes to sleep until the RTC ticks and wakes it up - this further reduces consumption to some 50uA. Even lower levels of consumption should be possible to be achieved, down to maybe some 10uA (including the LCD's consumption), but this should be enough for the first experiments.
I then took a solar cell salvaged from an commonplace solar lantern, with moderate illumination this one easily achieves 2-3V, sufficient to run the module. As the RTC consumes around 300-500nA (depending on voltage, but also on the particular crystal and surrounding circuitry), the 3x10uF capacitors, after being fully charged, should be able to supply it for cca 100 seconds or roughly 2 minutes (the RTC works down to 1.5V, and the current consumption with voltage is nonlinear, so endurance is slightly longer)
A couple of days ago, two of my friends - independently - discussed projects involving relatively large but precise (low leakage, low temperature coefficient, etc.) capacitors. Both expressed concern about dielectric absorption (DA) being potentially detrimental for their applications.
DA is one of those pesky "secondary effects" which make our parts so different from the ideal ones. The most dramatic way how it demonstrates itself is when charged capacitors are discharged momentarily, and then without outside source they "regain" part of the voltage they were charged into. DA is caused by charges trapped in quantum states with relatively long release time - contrary to "elastic" dipoles which are behind the "classical" permeability of the dielectric and which act "instantaneously" with applied electric field, making up a "perfect" capacitor (up to a certain frequency of course). DA is modeled by several series RC, in parallel to the "root" capacitance (plus any "conventional" parasitics, e.g. the resistance representing leakage), with time constants in minutes to hours, and capacitances up to several percents of the "root" capacitance.
So, to my friends, I recommended foil capacitors off top of my head, but then remembering Bob Pease's articles on "how soakage in Teflon is way better than Mylar", immediately went to search for solid data. And they are not that easy to find. First of all, there are several types of foil capacitors with different dielectrics, which obviously implies different amount of DA - and long are gone Pease's times with dozens of different foil capacitors; today there are basically only three materials used: polypropylene (PP), a family of polyamides (polyesther terephtalate PET, basically your plastic bottle, that's Pease's Mylar; and polyesther naphthalate, PEN), and polyphenylene sulfide (PPS). There's no polystyrene anymore (probably because of its low melting point, making it unsuitable for modern SMD and RoHS processes), no polycarbonates (maybe the same reason), and no PTFE (Teflon) touted as the best of bests, at least not in the common commercial offerings - they were said to be notoriously hard to process, mainly metallize.
Okay, but how ceramics perform with regard to DA? They were only in nascent state in Pease's times - a plain ceramic disc with two metal foils glued on - and technology made giant leaps in this area in the past few years, with intense research both in dielectrics, and their processing into the MLCC chip capacitors which are commonplace today. Manufacturers understandably tend not to publish more than absolutely minimum data on their parts, and DA is rarely if at all specified. From what I've found, the absorption capacitance for C0G/NP0 ceramics were said to have be around 0.5%, whereas PPS foil was said to have absorption capacitance 0.02-0.05% of the nominal capacitance.
(The second conclusion from Mr. Teverovsky's research is, that the absorption capacitance in X5R and similar high-permittivity capacitors is surprisingly high, tens of percents, typically around 25-30%; again, the difference to the much lower value in other sources - often around a few percents - can again be explained by the "unpatient" standard methodology of measuring DA, with the "discharge momentarily and then let "recover" method", where the "recovery voltage" is measured after 15 minutes, which again is probably one or...
The key issue in maintaining RTC operation during "no power period" (night in case of using solar power) is storage of sufficient amount of energy. We excluded parts based substantially on chemical processes (accumulators, supercaps, electrolytics) because of their inherent lifetime limitations, and ceramic capacitors on grounds of their relatively high leakage, preventing them to keep charge beyond a hour or so. So what we're left with is the finest kind of foil capacitors, namely the polypropylene ones, which exhibit the lowest leakage of all commercially available capacitors.
However, PP capacitors are expensive per microfarad, and they are bulky, too. From the charge equation:
assuming that a CMOS RTC will work in a cca 2V-3V supply power span (i.e. a 1V voltage drop), the required time from dusk to dawn is around 10 hours i.e. 36ks, and the current consumption is 40nA, and applying a snippet of optimism (i.e. that there are no other significant leakages) the required capacitance is around 1.5mF. In commercially available PP foil capacitors, available in capacitances up to few tens of uF, that converts to hundreds of euros/dollars, which - even taking into account that we might spend somewhat more on a clock which will never need batteries, maintenance, nor replacement - is more than we might be willing to spend. That the sheer volume of such capacitor battery may be larger than we would like to allocate for a clock, is another (albeit slightly less important) negative factor.
So is there any hope we might ever accomplish our task?
It's time to start to think a bit out of the box.
What is the reason why foil capacitors are bulky and expensive, compared to other types of capacitors? Price is given by the amount of material used times the cost of material, plus the cost of time/labour spent on preparing them. As the first factor is of the same reason as the "bulkyness", let's have a look at it first. Given similar thickness of the needed metallic contact layers, capacitance-to-volume ratio is given by the basic properties of capacitors: permittivity of the dielectric layer, its thickness versus breakdown voltage, and its area. Electrolytics and tantallum are unbeatable in volumetric efficiency because of they employ a "trick" to increase the latter two to its extremes: dielectric area is increased several orders of magnitude above the basic area of electrodes by creating a "spongy" structure using various methods, and the dielectric is "grown" electrochemically to exactly the minimum thickness needed for a given breakdown voltage. That the bulk of capacity of elyts/Ta is given basically by "growing" also decreases manufacturing cost per microfarad. High capacitance ceramic capacitors employ a high permittivity material, sandwiched with metal foils to a multilayer structure in huge sheets, which is then cut up into thousands of individual parts, so there is significant parallelism in making them (and, given the huge demand for them, they are manufactured in highly automated way which drives down the manufacturing costs too). Foil capacitors can do neither of these: the plastics have a relatively low permittivity, they are based on thin plastic foil and their breakdown voltage is given by the thickness of the foil so it has to be perfectly smooth and flat and any "sponginess" is excluded. Then they are metallized, wound up and packaged, and while there may be some amount of parallelism and automation applied, it's way lower than at the other types of capacitors.
So in capacitance the foil types simply can't beat other types. Do they have some other distinguishing feature then, which could be utilized?
The fact is, that foil capacitors are good at insulation voltages. So good, that PP capacitors rarely have rated voltage below 100V. Can't we utilize this fact?
So far, we asumed, that in the RTC "emergency supply" line there is no active electronics, just the capacitor, charged through...
Before delving into the details of chemistry-less hours-lasting storage, first, the clock itself.
Let's make at least this thing simple. I am going to use a STM32L476-DISCO board. This is based on the STM32L476VG microcontroller, which is member of a low-power 'L4 mcu family by ST Microelectronic. The chip contains both a built-in RTC (with an oscillator for the 32.768kHz crystal) and a segment-LCD controller, making it a suitable choice for building a clock. The DISCO board, besides other things, features a 6 x 14-segment LCD, an on-board ST-Link debugger, and also an auxiliary microcontroller with circuitry for measuring the main mcu's current consumption.
There's one more thing making the 'L476 very suitable for our project: the RTC has its own power domain (intended to be powered by a battery, hence the respective pin is named VBAT), and it isolates itself from the main power domain automatically when the main power drops below the power monitor's threshold voltage.
Of course, there are drawbacks, too. The RTC alone, running with the 32768kHz crystal, consumes around 500nA (the exact figure is voltage dependent), which - while in itself impressive and perfectly suitable for all usual practical purposes - is not exactly the lowest current which could be found on the market. As mentioned in the "root" project, there is a commercially available integrated RTC-oscillator-crystal, with order of magnitude less current consumption, at around 40nA.
So, keeping things simple for now, let's just try to run a simple clock on the 'L476, using its RTC; trying to maintain backup for its relatively large power consumption for maybe one or two hours - and if we succeed in this, there's hope that using a less power-hungry RTC may result in the desired night-long endurance.