Expected solar cell power production in shaded areas?

[ChromaticMultiplexer]

I built a few very low SWaP (size, weight, and power) LoRaWAN sensors to be installed in a forest like area and now exploring solar solutions. I am hoping to hear experiences and solutions used for similar projects.

My sensor's worse case power consumption is 20mW continuous and, according to a generic rule of thumb, a panel wattage rating needs to be ~20x the continuous load. 20mW * 20 = 400mW. However, this is for direct sunlight, facing south, and at an angle (~30 degrees), but forest conditions will have minimal if any direct sunlight.

Using two SM141K05L solar cells (154mW rating x 2 = 308mW total rating) I measure 261mW draw from the solar cells in direct sunlight from an E-Peas AEM10941 energy harvester, which is about 85% of the solar cell's rating, not bad right? The AEM10941 looks to have a max charge current of 110mA and using an INA228 power sensor the charging current draw was about 85mA, so I am assuming the 261mW power provided is the best that can occur due to my location, daylight conditions, and solar cell.

In shade the same setup draws 5mW of charging current which is 1.72% of the solar cell's power rating, which is disappointingly low. Is 10% power production the rule of thumb for overcast days? If 1.72% production of rated solar cells is the best case, how much solar cell is needed? Topping off the battery requires ~60mWh for eight hours (a guess) which is: 60mW/0.0172 = 3488mW! (too large for my small sensors)

Similar results were also found using SPV1050 an BQ25570 solar charge controllers and similar panels.

My questions:
Is the low 1.72% power produced of the solar cell rating normal for shaded areas?
Is there a clear choice between energy harvesters: AEM10941, SPV1050, or BQ25570? In terms of charging capability regardless of other features like output DC-DC.
Are there other chip sets I should consider for low-power setups like this one?
Any other thoughts?

Thanks!
 

[Peabody]

I recently did a test of a larger 5V panel rated at 2.5W, 500mA, and it tests pretty close to that.  I took the panel outside and pointed it straight up toward a clear blue sky, but in the shade.  I measured 9mA going into my charger, which is a TP4056 linear charger.  So if my math is right, that's 1.8%.  So that would be in the same ballpark as your panels.  My general experience with solar panels is that they do very poorly on cloudy days.

I would be surprised if a specific energy harvester or charger made that much difference in battery life.

Have you done all you can to reduce current consumption - increasing sleep or power-down periods and such?  Could you give us an idea of what you're sensing, and how often you sense and report?

[uer166]

6mA consumption at 3.3V? That's like 60% Rx duty cycle, or 6% Tx duty cycle at very high power on a SX1276. Are you sure you can't reduce those duty cycles as above poster mentioned?

[MarkT]

The eye's response to light is logarithmic, and we have auto-gain control (irises with variable aperture) so we don't tend to realize just how big a contrast in brightness there is - overcast can be 2% to 10% of full sun easily, inside a forest can be sub-1% in the height of summer with all the folliage.  Cloudy day in a forest - think <0.1% of full sun and you'd not be far wrong.

As an anecdotal data point I've driven into a forested section of road on a sunny day in summer (July) and was forced to turn my headlights on...  That was a drought summer and two weeks later enough leaves had died on the trees that it was easy to see on the same road...

[ConKbot]

The eye's response to light is logarithmic, and we have auto-gain control (irises with variable aperture) so we don't tend to realize just how big a contrast in brightness there is - overcast can be 2% to 10% of full sun easily, inside a forest can be sub-1% in the height of summer with all the folliage.  Cloudy day in a forest - think <0.1% of full sun and you'd not be far wrong.

Furthering this, an example correlating lux to everyday environments.


Lux is compensated for the eyes response, but it's still an apples to apples power/area comparison.  Add non idea factors (i.e. photodiode leakage not varying with current output, but voltage, so potentially remaining about constant as power varies by a few orders of magnitude) and low light output can fall off a lot.