low power microcontrollers, RTCs, ePaper and solar cells

[FrankBuss]

About those IXOLARTM solar cells, they are 0,089*0,055= 0,004895 m2 @ 22% efficiency @ 1 sun wouldn't that be 0,22*1000*0,004895= 1,0769 Watts ? Why do they say 22% and 714mW?

Good catch. That's what the datasheet says. I guess it clips above certain lux.

[GeorgeOfTheJungle]

My kindles draw ~0,71 mA in sleep mode (memory on, data not lost), that's about 0,00071*3,7= 0,002627 W= 2.63 mW. That's a problem because I want them to run on batteries for a year or more, and @ 0,71mAh day and night during a year that's 0,71*24*365= 4515 mAh. Do you think that those tiny PVs could supply 2.63 mW on average *indoors*, perhaps? So that sleep time will cost me nothing in terms of battery life?

[FrankBuss]

My kindles draw ~0,71 mA in sleep mode (memory on data not lost), that's about 0,00071*3,7= 0,002627 W= 2.63 mW. That's a problem because I want them to run on batteries for a year or more, and @ 0,71mAh day and night during a year that's 0,71*24*365= 4515 mAh. Do you think that those tiny PVs could supply 2.63 mW on average *indoors*, perhaps? So that sleep time will cost me nothing in terms of battery life?

Yes, that's possible, see my log article:

https://hackaday.io/project/20466-wifi-epaper/log/58513-solar-cell-tests-and-bigger-display

I measured about 5 mW max with 200 lux, which really isn't very bright, and if you have some daylight during the day, it would be no problem at all. I guess you could stick the solar cells on the back. You could even use less efficient and cheaper cells, because there is more room on the back of the Kindle than I planned for my device.

But you would need to modify your power supply. With the LTC3106 I plan to use for my project, everything is done for you, like switching between a backup battery and the solar cells, and even charging the battery when enough solar power is available. I guess if you place it somewhere with lots of sunlight (maybe not in direct sunlight to avoid damage) you even would not need to charge it at all from mains anymore. Would be a nice hack for hackaday.io.

[GeorgeOfTheJungle]

Ok, thanks. I'm going to order a couple of those to check it out. I've added a picture of ghosting in a message above so you can see what my kindles do after a couple of hours without full screen updates, as I'm painting the time every minute, that's after about 120 screen updates. In my case a full update every hour (1 every 59) seems to work fine and fix the issue. But I suppose YMMV because your epaper surely won't behave exactly like these.

[FrankBuss]

I didn't check partial updates so far as for your clock, might be the same problem with my display. Maybe do a full black/white cycle for the rectangle area where the digits are painted.

I also noticed in the current firmware for my Pervasive Displays that their firmware changed the update process: they don't use the inverse version of the old image anymore, but the process is like this: inverse version of the new image, full black, full white (I can't test it with my old display, could be full white/full black as well), and finally the normal version of the new image. Maybe this helps, too.

[GeorgeOfTheJungle]

they don't use the inverse version of the old image anymore

What I do is paint the inverse of the *new* image then normal. Also it's very easy to do that directly on the pixelmap of the framebuffer. Works fine for me.

[FrankBuss]

Ok, maybe try the full black / full white for the rectangle area where the clock is before you paint the normal image, then the full display update should not be necessary every hour.

[splin]

About those IXOLARTM solar cells, they are 0,089*0,055= 0,004895 m2 @ 22% efficiency @ 1 sun wouldn't that be 0,22*1000*0,004895= 1,0769 Watts ? Why do they say 22% and 714mW?

Good catch. That's what the datasheet says. I guess it clips above certain lux.

Solar cells don't clip. Looks like another case of marketing bull*hit. According to the diagram the actual net, active area = 16 x 12mm x 20mm = 3840mm after you account for the strip round the edge and the tabbing. That makes the cell efficiency more like 18.5% at best. Not that that is important - the relevant number is 714mW @ 1000W/m^2

Unfortunately the spec has nothing to say about low light performance despite devoting a fair portion of the spec talking about indoor light levels. Curious that they bother to provide Voc plots against irradiance but not the more important Isc. It's hardly because they haven't done the tests after all.

I'm very surprised you managed to get 5mW indoors; I just measured some 45mm x 65mm polycrystalline panels indoors in artificial light - a single 13W, 1500 lumen LED bulb room light, with some light still coming through the window. Unfortunately I don't have a light meter. Placed on my desk with no shading, Voc = .37V, Isc = 70uA so power probably around .7 * .37 * 70 = 18uW. Scale that up to your panel (based on net area) would give 35uW!

Moving the panel 500mm horizontally and raising it 300mm so that it is approx 1m directly below the room light the output goes up to .62V, 190uA or approx 83uW. The efficiency of my panels is probably pretty poor but the disparity is still enormous. I don't consider this room to be badly lit but the illuminance is likely far lower than in a typical office.

I just tried another panel which I know performs fairly well; it believe it is monocrystalline but I'm not certain. That produces 81uW and 280uW (scaled) in the two positions. My point is that you may well get much less output in many home situations - I have a Casio solar calculator which is fine for office use but is hopeless at home as you have to hold it right up to a bulb for several seconds to get it to start up!

Another data point - the panel spec shows typical office illumination levels starting at 9W/m^2. That panel would produce around 6.4mW at 9W/m^2 assuming its output doesn't drop off at low levels (unlikely). That suggests your test was conducted at light levels more typical of an office. That may be pefect for your application but be aware that it probably wouldn't be useful in many home situations.

[EDIT] And be aware that there can be considerable differences in low light performance between different samples - eg. I tried another panel of the 45x65mm panels which produced less than half of the output of the first (but are similar in full sun). This may be less of an issue with high quality panels but is of course unspecified.

[GeorgeOfTheJungle]

Ok, maybe try the full black / full white for the rectangle area where the clock is before you paint the normal image, then the full display update should not be necessary every hour.

That costs "a lot" of energy... I have to account for a full year of updates!

[GeorgeOfTheJungle]

@splin I have hope that what @FrankBuss has measured will be about right. And as thousands of mAh are at stake... I want to believe!

[FrankBuss]

I have measured the 200 lux with this instrument, cheap but very useful for such measurements: http://cgi.ebay.de/182356676136
The light source are 3 halogen bulbs, 50 watt, 1.5 m distance and no light from outside (was evening). But the lux value looks like it is standard for normal home lighting:

http://worldofleds.co.uk/lux-levels.htm

Problem is, I measured it lying on my workbench, will be less when sticking on the fridge, as one of my application idea is. But even with 1 mW it should be possible to do an update once per hour, which is my goal.

I think there are big differences for the solar cells, not only the efficiency, but also what part of the spectrum can be used. My halogen bulbs are very yellowish, such warm light is typical for many home lights. Some solar cells are more optimized for the colder sun light spectrum. And look at the price for my solar cells, there has to be a reason that they are multiple times as expensive than some cheap polycrystalline cells from eBay

[splin]

I have measured the 200 lux with this instrument, cheap but very useful for such measurements: http://cgi.ebay.de/182356676136
The light source are 3 halogen bulbs, 50 watt, 1.5 m distance and no light from outside (was evening). But the lux value looks like it is standard for normal home lighting:

http://worldofleds.co.uk/lux-levels.htm

Agreed, but in many homes living areas may be more like 100 lux or less - depends on personal preferences.

Problem is, I measured it lying on my workbench, will be less when sticking on the fridge, as one of my application idea is. But even with 1 mW it should be possible to do an update once per hour, which is my goal.

I think there are big differences for the solar cells, not only the efficiency, but also what part of the spectrum can be used. My halogen bulbs are very yellowish, such warm light is typical for many home lights.

Ah-ha! You didn't provide that info in your blog. So....

Some solar cells are more optimized for the colder sun light spectrum.

I swapped the LED bulb for a 150W incandescent. Perhaps 2200 lumens so approx 45% more than the LED. The output of the first panel, on the desk, was now 541uW - approx 30x as much. Moving it to 1m below the bulb increased that to 1410uW or approx 17x LED bulb.

Scaled up to your panel that would be 1.05mW and 2.73mW. In the second case at 1m from the bulb lux = lumens/area = 2200 lumens / (4 x pi x r^2) = 175lux. 2.73mW/175*200 = 3.12mW. If your panel is 22% efficient that makes mine 13.7% which is about what I'd expect given their age and being cheap polycrystalline cells from eBay

So from this I think the conclusion is that you should expect *much* lower outputs with LED lighting at similar Lux levels - which you'd expect as they put out much less power in total. The colour temperature is almost irrelevant - a 100W incandescent bulb puts out 100W of energy, of which the panel is responsive to perhaps 35% of the spectrum - ie. visible light plus some of the near infra-red. That means a solar panel completely surrounding the bulb could receive up to 35W of light plus near infra-red.

An equivalent 13W LED bulb may lose 3W in the driver; of the remaining 10W the LEDs may be around 14% efficient so perhaps 1.4W would be available for the solar panel with rest being dissipated as low temperature heat which the panel can't utilise. 1.4W is 1/25 of the power aailable from the incandescent - which isn't far of what I observed notwithstanding the very crude experimental conditions and broad assumptions.

This is made worse by the severe drop off in efficiency of a solar cell, including monocrystalline, at low light levels. At 200 lux when provided by the sun or an incandescent bulb, the drop off may be minimal but 200 lux from an LED is a very much lower power level and the efficiency may well be much worse. I say *may* as the spectra are very different so more research is needed to be sure. Its easy to measure though using your panel of choice and a suitably LED lit room.


Some/many people may still use halogen bulbs in downlights, but lots have switched to LEDs and it's probably only a matter of time before they are banned altogether. Unfortunately I think you are going to have to rethink this project.

[FrankBuss]

Right, I noticed the same problem when trying to increase the power with my LED flashlight. It looked brighter and the lux meter showed a higher value, but the solar cell output power was actually lower. So it might only work if it is installed somewhere with enough daylight. And for better exposure maybe a construction like this would be a good idea:



Would make it more stable, too, instead of just relying on magnets.

[GeorgeOfTheJungle]

Thanks @splin, el aguafiestas