💬 Effective Solar Supercap Boost Charger for Small Solar Panel

[ncollins]

@neverdie according to a few datasheets 2v at 200lux. The solar cells delivered were HENGYANG SC-3514.

http://www.vimun.cn/en/ProductInfo.asp?pid=18
http://www.solars-china.com/solars/indoor-solar-cells.pdf

On my windowsill, 2.5v typical in-direct sunlight, 2.7v is the highest I've recorded in direct sunlight.

So far, most of my Aliexpress buys have been fine for my needs. Haven't really gotten close to making a "production ready" module, so for prototyping it's been fine.

One exception: I thought I had a genius idea to repurpose these $1 solar powered keychain flashlights. They had an amorphous solar panel, rechargeable battery, button, leds and a housing...for $1. I bought 20 of them. First one I opened up, I realized it's a lie!

The solar panel is just glued to the circuit. The battery is just a normal non-rechargeable CR2016.

[alexsh1]

@neverdie said in Effective Solar Supercap Boost Charger for Small Solar Panel:

@ncollins said in Effective Solar Supercap Boost Charger for Small Solar Panel:

@neverdie they work super well. I've been testing them for a few months now, very reliable.

Great! Nice to have your feedback.

I had high expectations for Ceech's LTC3108 design:
https://www.ebay.com/i/331654685113?chn=ps&norover=1&mkevt=1&mkrid=711-117182-37290-0&mkcid=2&itemid=331654685113&targetid=809743845025&device=c&mktype=pla&googleloc=9028292&poi=&campaignid=6470262913&mkgroupid=80364172271&rlsatarget=pla-809743845025&abcId=1141166&merchantid=115344895&gclid=CjwKCAjw29vsBRAuEiwA9s-0B7exYA8yU9w--t63jcXP7QWO_f05DzJKNxw2XS6pwlLkPkfK1UYqkxoC084QAvD_BwE
but when I tested it, it turns out it needs about 5ma of current to run, which is at least 10x+ more than most small solar panels will deliver under even quite bright indoor lighting. Basically, if a panel can produce 5ma of current, the voltage will also be fairly high, thereby completely defeating the purpose of a chip like the LTC3108, which can be powered from 20mv. In my testing it took a minimum of 50mv and 5ma of current to get it to actually harvest any energy. 50mv would still be impressive, but the 5ma minimum is what kills its usefulness.

One of the best modules @ceech made was this one

https://www.ebay.co.uk/itm/BQ25570-thermal-solar-energy-harvester/332071662285?_trkparms=aid%3D1110001%26algo%3DSPLICE.SIM%26ao%3D1%26asc%3D20131231084308%26meid%3D759b68bcf5834d74bfa46f9df6b8ae4d%26pid%3D100010%26rk%3D1%26rkt%3D6%26sd%3D332393146893%26itm%3D332071662285%26pmt%3D0%26noa%3D1%26pg%3D2047675&_trksid=p2047675.c100010.m2109

It has been running non-stop for many months now

[NeverDie]

@alexsh1 Does it work well even under extremely low light indoor conditions?

[NeverDie]

What the BQ25570 has working against it is a cold-start voltage of >600mv. The LTC3108 doesn't really have a cold start voltage. It can in theory start with as little as 20mv (though in the case of Ceech's module, 50mv).

On the other hand, the BQ25570, once it reaches its cold start voltage, can operate with far less current than the LTC3108. And once the BQ25570 gets going, it can continue operating down to 100mv. That part is very attractive indeed. I think I'll give it a closer look.

[NeverDie]

I found the smoking gun:
"Q. What about solar harvesting applications with the LTC3108?

A. In general, you should look at the LTC3105 first for these applications. The LTC3108 requires a minimum input current of a few milliamps (at the converter input) just to startup, and may not be a good load match for a PV cell. Therefore, small solar cells that have a short circuit current of less than a few milliamps will not work with the LTC3108 (or LTC3109)."
https://www.analog.com/en/technical-articles/frequently-asked-questions-thermoelectric-energy-harvesting-with-the-ltc3108-ltc3109.html

[alexsh1]

@neverdie It does work under very extreme conditions (dark or very cloudy days for weeks), but I have not tested it extensively with internal lightning. The solar panel is very close to a window. I may be mistaken by I do not think LTC3108 is designed for low current (< 1 mA) application. BQ25570 is absolutely brilliant.

[NeverDie]

Because of the passage I quoted above, I re-visited my earlier LTC3105 project:
https://www.openhardware.io/view/281/Solar-Energy-Harvester
and gave it a closer look. As a result, I now have it starting the up-conversion at 270 millivolts (a slight derating of the advertised 250mv). It can fairly quickly upconvert from 270mv input to 2.63v output (which is what I picked as a "safe" voltage for charging a supercap). It does consume much less input current than the LTC3108 does, but if you put it in a dark enough room it will stall before starting up, which is a bit disappointing. That said, though, I can fix it using a similar technique that I used for the topic of this thread.

Meanwhile, from amazon.com, I ordered the solar boost charger module that recently won the hackaday prize: https://www.amazon.com/gp/product/B07JFNSPQ3/ref=ppx_yo_dt_b_asin_title_o01_s00?ie=UTF8&psc=1 I'll be interested to see how well it handles a dimly lit room!

[NeverDie]

@ncollins

I'd like to do the same trick with an LTC3108 (namely, pulse it on when it hits a voltage threshhold at, say, 200mv or less), but the trouble is that the voltage detection chips need a supply voltage higher than that to work. So, it's a bootstrap problem that's difficult to get around. Including a small button cell battery as a workaround to that problem feels like cheating, but the alternative might be to build some kind of ultra low voltage trigger, and I'm not sure how to do that. An SCR trigger would maybe work, but it might have a 0.6v threshhold. The EPAD Mosfet's might turn on a 0.2v. Just not sure if it would be a clean switch-on or not.

I suppose using a rechargeable watch battery for the same purpose would feel a bit less like a cheat. Do you happen to know long do they typically last, and what's their rate of self discharge?

[ncollins]

@neverdie this is the data sheet for the authentic version of the watch batteries I'm using:
https://www.sii.co.jp/en/me/datasheets/ms-rechargeable/ms412fe-5/

I tried to order a few of the AEM10941 ICs but they quoted me ~$100. I guess they still don't have a US distributor?

Personally, I'm struggling to justify the cost of the harvesting modules. I think ultra high efficiency indoor solar panels that put out usable voltage at very low light levels + supervisor IC with adjustable hysteresis should provide comparable performance with significantly less complexity/cost.

[NeverDie]

@ncollins Well, you're in good company then. Cypress Semiconductor seems to have come to the same conclusion as you: using higher quality solar cells in series to jack up the voltage. Cypress seems to like the ones made by panasonic.

On the other hand, why not do both? This guy has a design for what looks like an inexpensive boost converter that he says requires as little as 10ua standby and has 90% efficiency:
https://www.davidpilling.com/wiki/index.php/DCDC
Not sure what the start-up voltage is though.

[NeverDie]

@ncollins Maybe TPS61200 would be a good compromise. It's cheaper than the other harvesters, and it has a UVLO centered on 0.3v, so maybe it wouldn't need a separate circuit to keep it from crashing potentially weak output of a solar cell.

My main reservation about it is that it won't be as easy to solder as an LTC3108 or LTC3105.

[ncollins]

@neverdie said in Effective Solar Supercap Boost Charger for Small Solar Panel:

TPS61200

Looking at the example Single Solar Cell usecase http://www.ti.com/lit/an/slva345b/slva345b.pdf, it seems the expectation is minimum input 3ma at 0.5v? Maybe for startup, but lower current requirements after?

As for the small solar cells in series, I have thought about that too, but occasionally I have sensors that get hit with direct sunlight. I've fried a couple of NRF24 modules that way. I like the idea of using a 5v amorphous solar panel charging a 5.5v supercap + an LDO (with a passthrough?).

I want a harvesting module that boosts and dumps voltage into a very small supercap 2.7v with enough capacity for one boot/wakeup + sensor read + transmission. Once the small supercap is charged, the input voltage boosts up to 5v and is redirected into a larger storage supercap. , the the small super cap should be charged by the storage supercap when charge input is too low to be boosted. If necessary, the storage supercap could be boosted to charge the small supercap.

Basically, do whatever it takes to get ready for the next transmission, then store in a reserve. If under voltage, boost it, if over voltage route through LDO, else passthrough.

[NeverDie]

@ncollins I think the booster designs are good fun for seeing just how far one can push the limits. So, with that in mind, I decided to do an exploratory breakout board for the most extreme booster out there , namely the LTC3108:
https://www.openhardware.io/view/732/Extreme-Energy-Harvester
For solar cell applications in dim light it may need to be managed though, because otherwise it may (?) collapse the input voltage and simply stall. i.e. it needs to be pulsed. If I'm lucky, maybe with a large enough input capacitor it will pulse itself. If not, it will need some kind of external control to manage it. Unfortunately, datasheets rarely provide that kind of insight, so in most cases one simply has to build and test in order to know for sure.

Maybe a trigger circuit simiar to this would work, but it woud need to work at a very low voltage level, and therein lay the rub:

Maybe one of these jfet oscillator circuits would work at an appropriately low voltage:
http://www.dicks-website.eu/fetosc/enindex.htm

The trick would be getting it to work not only at a very low supply voltage, but a very low supply current as well.

If anyone has ideas or suggestions, please post.

[NeverDie]

@ncollins I've noticed that a number of the dollar store "solar calculators" are fake as well. They give the appearance of being solar powered, but in reality they run on a small battery instead.

[ncollins]

@neverdie I was surprised to find the deceptive $1 solar keychain actually used a decent indoor solar cell. I can't find similar spec'd solar cell for less than $3 (at 10-20pcs). Despite the blatant false advertising, they're actually worth salvaging, especially with a solderable CR2032.

[NeverDie]

@ncollins Is there a better forum than this one for discussing solar energy electronics? I've posted what I can, but it's difficult to find help/collaboration here.

[ncollins]

@neverdie Not that I've found. A thread here, a thread there, but nothing focused. Might make sense to create a new mysensors forum category, Energy Harvesting?

Forum is a good place for conversation, but I think a wiki would really help document the results of those conversations. Also a place to provide some intro material: What is energy harvesting? Why pursue it? The easy route, the advanced route...

I'd gladly help with the content.

[NeverDie]

I recently went looking on some of the Joule Thief forums in the hope of perhaps finding some useful circuits there. However, the problem that I repeatedly ran into was that because the inductor used by Joule Thief devotees is typically hand wound from whatever junk is laying around, it looks as though it may be difficult t to repeat their successes. It might have been easier to leverage their work if they had used standardized parts.

[NeverDie]

@ncollins said in Effective Solar Supercap Boost Charger for Small Solar Panel:

@neverdie I was surprised to find the deceptive $1 solar keychain actually used a decent indoor solar cell. I can't find similar spec'd solar cell for less than $3 (at 10-20pcs). Despite the blatant false advertising, they're actually worth salvaging, especially with a solderable CR2032.

How much voltage/current/power can they produce?

I've noticed that some solar cells, especially cheap ones, can leak away their current almost as fast as they can produce it.

The "fake" solar cells in your photos loosely resemble these amorphous solar cells from Panasonic: https://www.digikey.com/product-detail/en/panasonic-bsg/AM-5907CAR/869-1013-ND/2165198

[ncollins]

@neverdie In direct sunlight, I've seen 5.5v at around 4ma.

These seem to be the cells SC-3222-9 SC-3722-9 (edited).
http://www.vimun.cn/en/ProductInfo.asp?pid=17

[NeverDie]

@ncollins said in Effective Solar Supercap Boost Charger for Small Solar Panel:

@neverdie In direct sunlight, I've seen 5.5v at around 4ma.

These seem to be the cells SC-3222-9.
http://www.vimun.cn/en/ProductInfo.asp?pid=17

Come again? I don't see a cell on that list with the name of SC-3222-9.

[ncollins]

@neverdie SC-3722-9*

[NeverDie]

Which do you prefer? Those or the SC14351? For powering up the cold start electronics quickly, what I need is high voltage more than I need high current. Your solar cells are inexpensive enough that they might be a good fit for that.

[NeverDie]

@ncollins For instance, for the price it might be a good fit for this:
https://www.openhardware.io/view/733/Buck-Energy-Harvester

[ncollins]

@neverdie I've primarily been prototyping with the SC14351 because I don't have to worry about voltage regulation when coupled with supercaps.

[NeverDie]

@ncollins Are you using the Tantalaum's primarily because they're cheap, or for some other reason?

You may be interested in this as a tiny, yet cost effective alternative: https://www.digikey.com/products/en?keywords=728-1067-1-ND
The ESR is rather high on it, but it might be good for running very low current circuits, such as logic.

[ncollins]

@neverdie I used tantalum because they were laying around and I was too impatient to wait for the 1206 100uf ceramic caps I had ordered.

Those are definitely interesting. The high ESR on the lower capacitance coin-cell "H-type" 0.47F super caps caused a voltage drop significant enough to trigger brownout on my first few nodes.

I was hoping to try these out, but they seem to have been discontinued https://www.mouser.com/new/Murata/murata-dmh-supercaps/.

[NeverDie]

@ncollins I like your idea, so I ordered some similar solar keychains from Amazon that got good reviews: https://www.amazon.com/HDE-Emergency-Flashlight-Energy-Keychain/dp/B00NFZUTR6/ref=sr_1_4?keywords=solar+keychain&qid=1572300649&s=hardware&sr=1-4

In the worst case they'll probably be no better than solar garden lights, and with the same DC boost converter ASIC, but if I'm lucky their small solar panels will put out better voltage than the 2v that's typical for garden light solar cells. Also, the packaging might be nice for a solar sensor node.

[ncollins]

@neverdie Awesome. Interested to see how they work out for you.

For other options, this is the only retail outlet I've been able to find that carries different size/volt versions of the chinese amorphous panels.
https://cnmarsrock.aliexpress.com/store/group/Amorphouse-solar-panel/400691_511437499.html

[NeverDie]

@ncollins Strange! In that case it appears that it's cheaper to buy them as part of solar keychains....

Do you happen to know: what difference, if any, is there between the solar cells that they label as "dim light indoor use" as compared to the outdoor cells?

[ncollins]

@neverdie In theory, indoor is optimized for visible light spectrum. Outdoor, a combination of visible and IR. https://www.powerfilmsolar.com/about-us/the-horizon-blog/2018/08/10/outdoor-vs-indoor-solar-the-key-differences

But, who knows with these chinese panels. I haven't tested the indoor vs outdoor side by side. I do know these amorphous "outdoor" panels work significantly better indoors than the mono/polycrystalline panels I have.

[ncollins]

@neverdie More information here https://panasonic.co.jp/ls/psam/en/products/pdf/Catalog_Amorton_ENG.pdf

[NeverDie]

@ncollins said in Effective Solar Supercap Boost Charger for Small Solar Panel:

@neverdie Awesome. Interested to see how they work out for you.

I opened one up:

It appears to be the simplest possible solar charger: a solar cell with a diode and some (presumably) rechargeable batteries.

The tiny circuit board was affixed to the solar cell with sticky tape but can be pried off. Doing so reveals that the model number for the solar cell: 3722-9L,

Maybe it's this one?
http://semtronics.net/shop/index.php?rt=product/product&product_id=1123

Or maybe this one?
https://www.aliexpress.com/item/32774648368.html

Their specs seem to mostly agree. I'm guessing that perhaps the "L" might be for "low light"?

GND on the circuit board was attached to the negative terminal on the solar cell by the faintest possible wisp of threaded wire. Removing the circuit board severed the wisp.

I found that someone else did a teardown with a similar numbered solar cell:
http://tarkus-notes.com/en/mini-chinese-solar-power-flashlight/
although his was missing the diode and it looks as though his solar panel wasn't connected. According to him, the panel would have 10.0uA at 200 lux, where 200 lux is the usual reference for indoor lighting.

My own quick first test shows that I'm getting 4v open circuit under ordinary indoor LED lighting and 9uA of current when I short the leads. I would guess that it could probably power the "Effective Solar Supercap Boost Charger for Small Solar Panel" under even dimmer light, provided that the combined quiescent and leakage currents are small enough. With only a few net microamps flowing in, the margins are pretty thin.

[NeverDie]

@ncollins Aha! So, it appears you and I may have more or less the same solar cell now.

[NeverDie]

I went ahead and hooked up the 3722-9L to 600uF of ceramic capacitors and fed that into the "Effective Solar Supercap Boost Charger for Small Solar Panel" under indoor LED lighting, and with a red LED on the output. Sure enough, it sends out a stream of LED pulses (not very bright, but at least minimally visible if you cup your hand around the LED to shade it). I could make the pulses brighter by adding more input capacitance.

For comparison, I just now wired a brand new IXYS SM340K10l solar cell (http://ixapps.ixys.com/DataSheet/SM340K10L.pdf), which I thought would trounce the the scavanged 3722-9L under the same indoor LED lighting conditions. Granted, it is a bit smaller, but it's open circuit voltage was just 1.35v, and it's shorted current was 6uA. And it cost me $3.86 on Digikey. So, quite a surprise there. Go figure.

[ncollins]

@neverdie at ~$1.90 per keychain, seems like a great way to get into harvesting.

I bet it would be easy enough to design and 3d print a replacement to the button half of the housing with a little more space for a radio and sensors.

I mean, you could probably just add an LDO, HT7333 or XC6206, and use it as is.

[NeverDie]

@ncollins I'm curious as to why you aren't using your 3722-9L's on your solar motes?

Is it because you're more geared to collecting sunshine instead of nighttime indoor lighting?

Curiously enough, during the daytime (but still indoors) I found the IXYS is pumping twice the microamps as the 3722-9L So, clearly, the two boards respond to different spectra differently. The IXYS gives its "external quantum efficiency" for different wavelengths. In contrast, I haven't seen any reference to wavelength efficiency by the 3722-9L, but apparently I got lucky and it's optimized for the kind of light the LED's in my office are emitting. Regardless, the 3722-9L always has more open circuit voltage than the IXYS cell I referenced earlier.

By the way, as an aside, I put a 10,000 uF capacitor on the "Effective Solar Supercap Boost Charger for Small Solar Panel," and now it blinks the red LED plenty bright enough to see without having shade it.

[ncollins]

@neverdie It's mostly because I haven't figured out how to appropriately, efficiently handle the wide voltage range of the 3722s.

In ambient light, they'd be perfect, but most locations in my house, there's a chance they'd get hit with direct sunlight. In a couple of my earlier prototypes, I fried the NRF module after the supercap exceeded 3.6v (damaged something that causes them to draw ~1-2ma).

In my case, I don't have a lot of artificial lighting, but I do decent direct/indirect sun from windows for at least part of the day.

Thoughts:

• add a really efficient LDO between the 3722 and the super cap
• Sacrifice size compact size and jump up to 5.5v supercaps. Add LDO between the supercap and mote
• rig some kind of a pass through when vout < 3.6, else go through LDO

I'm mostly avoiding the issue by using the SC-3517s. A single cell wasn't really cutting it unless in direct sunlight, so I double up in parallel.

Bonus: other prototypes

Uses your boost charger

Testing out 50mF low ESR super cap

[NeverDie]

@ncollins That AVX-BestCap looks like a great find. I love the compact form factor of it. Looking at the datasheet though, I worry that the self-discharge of 5-10ua might be a bit much: http://catalogs.avx.com/BestCap.pdf
That's high enough that it might consume either all or at least much of the microamps produced by the 3722-9L (assuming just a single cell is used) when there's light available, and the leakage would continue overnight in the darkness to eat into whatever had been accumulated.

On the other hand, the all aluminum caps at that same 50mF capacity are huge and for that reason not such a good fit either for a sensor that aspires to be compact.....

[ncollins]

@neverdie With all of your energy harvesting approaches, how are you handling unforeseen restarts? Almost all of my harvesting sensors have the same issue: if they drop below 1.8v (depending on BOD) they can't recover without manual intervention. Only one of my nodes has ever recovered and I think it's because the solar panel produced enough current to push through the MCU and radio startup.

My thought was to use the MIC2778 supervisor ic (datasheet), which has programable hysteresis, and a mosfet to wait for supercap/battery to reach a voltage high enough to handle a cold start.

[NeverDie]

@ncollins Yes, that should work. Since I disabled BOD in order to sleep at very low currents (100na for the atmega328p), I built my own in hardware using a voltage detector. So, if voltage drops below 1.8v, it cuts off the power. When it goes above the hysteresis point, it restarts.

If you leave BOD enabled, I think it may do this work for you, and then you wouldn't need more hardware. But then your sleep current would be higher. So, it's a tradeoff as to how you want to do it.

[ncollins]

@neverdie yeah, the fixed hysteresis of the internal BOD (328p, NRF51) isn't wide enough to handle startup for my nodes. I'm guessing it's due to the high ESR and corresponding voltage drop of the supercaps I've been favoring.

I also didn't realize how chatty mysensors can be during start up. My latest nodes have MY_PASSIVE_NODE enabled, skip presentation, and use dumb sleep.

[NeverDie]

@ncollins I forgot to mention that I use a load switch, controlled by the voltage detector, to turn on and off the atmega328p. I don't drive it directly from the load switch. You could use two different voltage detectors to get a much wider hysteresis, or you could use a hysteresis chip for a more elegant solution. I know for sure that the voltage detectors draw a minuscule amount of current. I haven't looked into the hysteresis chips enough to know how they would compare along that dimension, but I'm guessing they would be similar.

[NeverDie]

@ncollins Here's the load switch that I've been using: http://www.ti.com/product/TPS22860 I like it a lot. I think it, or something similar, should be in everyone's tool box.

[ncollins]

@neverdie said in Effective Solar Supercap Boost Charger for Small Solar Panel:

@ncollins Here's the load switch that I've been using: http://www.ti.com/product/TPS22860 I like it a lot. I think it, or something similar, should be in everyone's tool box.

Awesome, I'll order a few. Appreciate the info, thanks.

[NeverDie]

@ncollins said in Effective Solar Supercap Boost Charger for Small Solar Panel:

@neverdie yeah, the fixed hysteresis of the internal BOD (328p, NRF51) isn't wide enough to handle startup for my nodes. I'm guessing it's due to the high ESR and corresponding voltage drop of the supercaps I've been favoring.

I also didn't realize how chatty mysensors can be during start up. My latest nodes have MY_PASSIVE_NODE enabled, skip presentation, and use dumb sleep.

You could either add a capacitor to get it over this hump, or you could just have the atmega328p check it's voltage level at startup. If it isn't high enough, it goes to sleep and keeps checking upon subsequent wake-up's until it's high enough to support radio transceiving without crashing below 1.8v.

[NeverDie]

@ncollins What I tried in the past, and which I rather liked, was to first charge up, say, a 100uF capacitor and use that to power the 328p when voltage threshhold is reached and to then also redirect the incoming solar charge current into a supercap for bulk collection That way, in a cold start scenario, the MCU powers up very quickly--nearly instantly with larger solar cells--rather than having to wait hours for a large supercap to fill up. 100uF is enough to power numerous transmissions.

All this comes at the cost of more circuitry though, so I can understand if someone would prefer to omit this feature to keep things simple.

Maybe we should start a separate thread for this kind of discussion?

The only downside I see to the hysteresis chips is that they consume more current, especially when you consider the amount of current consumed by the voltage divider circuit. In the case of the MIC2778, it recommends the total resistance of the voltage divider circuit be no more than 3megaOhm. At low voltages, though, I suppose that's not much of a problem. Still, for comparison, some voltage detectors might consume a mere 350na in total.

[ncollins]

@neverdie Yeah, I think starting a dedicated thread is the right move.

That sounds a lot like what I'm trying to do. Prioritize load (boot/transmission) power, then dump into a reserve. This make sense when light is strong, but when solar isn't producing, how do you get power from the reserve to the load capacitors. Would you boost it?

[NeverDie]

Continuing the discussion on a new thread here: https://forum.mysensors.org/topic/10748/solar-energy-harvesting-for-wireless-motes

[MarcosBe]

Very interesting information, thank you very much!