Had a nearly impossible time getting the photodiode based solar sensor to yield a sensible shaped sunlight vs time curve.
Tried columnators, different aperatures, different orientations etc.
Finally gave up and modified the unit to use three solar cells. These were scrounged from a number of old solar calculators I had laying around from a long time ago. All were 3 volt units.
They were glued together at right angles to each other to form a building (long side up) with solar cells on three “sides” of the building. The cells were wired in parallel. It was then mounted on the lid of a baby food jar. Lots of holes were drilled in the bottom of the jar lid to provide some circulation. The glass jar was screwed on and the unit mounted outside with the blank side facing north and lid/holes facing down.
The modification required removing the photodiode sensor and an additional 15 K ohm resistor. Details can be supplied.
The result is a sensor which seems to more closely follow perceived daylight brightness. It is a bit non-linear – compressing the high brightness levels. I view this compression as a positive since eyes respond non-linearly also. Dips due to clouds are visible but not extreme. The geometry problems seem to have vanished.
I take no credit for the idea, it is described in an application note
If what you are looking for is a “sine” curve like the one below then your best bet is to have all your cells horizontal with a diffuser above.
The compression effect you are seeing is because that circuit is measuring the unbiased open circuit (well, lightly loaded) voltage generated by the cell which is not linear. For a linear measurement you would have to bias (apply voltage to) the cell to get it into a linear portion of its response, but the real solution is to use an amplifier and measure the short circuit current which is very linear over a wide range (this is how the Davis solar sensor works).
In any case well done! Using a small solar cell is a far better solution than the lensed photodiode that I have seen supplied with that kit. The narrow angle lens means that it’s just looking straight up viewing a small cone of the sky, and no amount of diffuser can really fix it.
First, I am not attempting to measure any physical quantity. The goal is to qualitatively correlate sunlight information with temperature variations.
There was a reason for my madness, i.e. mounting the sensors vertically.
I was interested in augmenting the early morning signals relative to later in the day signals.
This vertical mounting also serves to compress the output for several hours around noon time. What results is not a sine wave but something which is squashed with a relatively flat top. The periods of cloudiness are still seen but reduced in maginitude. This is closer to the total ambient light as I perceive it.
Will try and figure how to post a typical shape figure here…
As far as getting the photocell into the linear region, I just took an easy way out. At lower light levels the response is almost proportional. To reduce the the overall light level to this region, I just put a green pill bottle over the photocell “building”. The “gain” of the hobby boards circuit was upped at this lower level by inserting a 500 ohm potentiometer in series with the 390 ohm resistor. Adjusting the pot results is a curve which peaks at close to 100 millivolts. 100 mv is 100% solar when assigning the sensor to the “general purpose ADC” slot on setup.
At this point, I quit fooling around. It was good enough.
:o :idea: