[danner]

I'm used to being able to figure things out. My being frustrated is not with you or the weather station, but with my inability to sufficiently comprehend the system to manipulate it to make it do something (I am admittedly over my head because I've never undertaken this sort of hands on project with something I knew so little about). I had incorrectly assumed the battery charger kicking on and causing voltage fluctuations may have contributed to it, that's why I swapped out the battery for the power supply. That's my issue and no one else. I will refrain from using colorful descriptors in the future. 

When the rain stops I'll reduce the gain further.

[werk_ag]

   

"clipping" set to 5

http://www.meteocercal.info/recentGraphs.php

[danner]

Looks good.

[danner]

Among the other hobbies I do is amateur astronomy so I'm quite aware that there is light from the sun before the sun comes up and after it goes down. How sensitive the sensor is to the small light levels is something I do not know yet, but hopefully will understand soon. Which lead me to another examination of what is behind the readings I am getting at night.

Something else I was looking at was whether the full moon could possibly trigger the sensor as it is quite bright and high in the sky. There is some correlation in the solar readings (the spec sheet says it's a type of photo-diode, which I was able to get output from by shining my flashlight on it) with the full moon rising and the higher in the sky the moon is the higher the readings above 1 W/m2 become. But I am also aware that correlation does not equal causation, that it could simply be a coincidence. I wont know if there is any correlation until the moon cycles and isn't out at night, or I could climb on my roof and put the black cloth on the sensor.

[werk_ag]

My being frustrated is not with you or the weather station, but with my inability to sufficiently comprehend the system to manipulate it to make it do something (I am admittedly over my head because I've never undertaken this sort of hands on project with something I knew so little about).

I also haven't interpreted the expression, as something directed to me or to the project, perhaps just some impatience to get results, something I can very well understand, as sometimes I suffer from the same problem.

I will try to give some simple figures that may help to understand to what we are dealing:

Supposing that R2 have a value of 4900 Ohm, the gain of the OP is around 3, this mean that by each mV on the input the output will be 3 mV.

So, if the pyranometer (or any interference) presents a voltage of just 0.33mV in the OP input, that is enough to produce 1mV at the OP output, which by turn will produce a reading of more than 1W/m2.

I think this figures will illustrate how easy is having some spurious readings. Cutting the readings below an acceptable threshold will be almost always necessary.

I'm aware that there are better designs than a simple non-inverting circuit for the OP, but they can't also give 100% insurance against spurious readings.

[danner]

I will try to give some simple figures that may help to understand to what we are dealing:

Supposing that R2 have a value of 4900 Ohm, the gain of the OP is around 3, this mean that by each mV on the input the output will be 3 mV.

So, if the pyranometer (or any interference) presents a voltage of just 0.33mV in the OP input, that is enough to produce 1mV at the OP output, which by turn will produce a reading of more than 1W/m2.

Out of curiosity, is that change in gain linear?

If the gain of the OP-amp is "3" with the pot at 4.9k ohms, what was the gain with the pot set for 4.7k ohms or after I changed it to 4.81k ohms?

[werk_ag]

Out of curiosity, is that change in gain linear?

Yes, otherwise things could not be done so easily.

If the gain of the OP-amp is "3" with the pot at 4.9k ohms, what was the gain with the pot set for 4.7k ohms or after I changed it to 4.81k ohms?

I said around 3

R2 = 4900  Gain = ~3.040816
R2 = 4810  Gain = ~3.079002
R2 = 4700  Gain = ~3.12766

[AllyCat]

Hi,

I've been watching this thread, but stayed out because, although I understand (some of) the theory, it's a very complex subject and I'm in danger of writing pages and pages....  Also, I suspect that it's fundamentally a "physical" problem, but don't have much of an idea what the cause might be.  So just a few brief comments for now:

For photography, one unit of light level is the Lux, with full sunlight being approximately 100,000 Lux.  But moonlight is less than 1 Lux (i.e. >100,000 times smaller) so if we apply that to the 300 mV full-scale from the SP-110 we would expect only a few microVolts in moonlight.  Those voltage levels are very difficult to measure because of  "noise" and thermoelectric effects.  That just might be the cause of your problem - junctions between different metals at different temperatures can generate a (small) voltage difference.

The light (or energy) level that we're trying to measure is proportional to the current which is generated (and actually it's not too difficult to measure a current ratio of 1: 1 Million, or more).  However, voltages are generally easier to "understand" and measure, and the conversion is easy - it's what a resistor does (V = I x R) !  Werk puts the resistor directly on the (PV) sensor (as presumably do Apogee) whilst Davis effectively amplify the current first and then convert to a voltage.  

The voltage gain is indeed (almost) proportional to the feedback resistance (ratio).  So for the very best stability (but not as convenient), the "Pot" could be replaced by a Metal Film (fixed) resistor (perhaps re-using the pot to balance the Input Offset) and the gain adjusted purely within the software.  

It's worth considering the specification of the Op Amp:  The "Input Offset Voltage" can be several hundred microVolts, which represents an "error" in the input voltage, that gets multiplied by the amplifier gain to produce an output that might represent several watt/m2.  Also, although it is a "rail to rail" device, it cannot pull the output completely down to zero (milli-) Volts.  The data sheet specifies about 6 mV at 20 microamps (pull-up), but I can't see why the current would be even as large as that.  However, I do wonder if the designers of the SP-110 expected it to be used with a true "instrumentation grade" (i.e. very expensive) amplifier.

Cheers,  Alan.

[danner]

R2 = 4900  Gain = ~3.040816
R2 = 4810  Gain = ~3.079002
R2 = 4700  Gain = ~3.12766

So correct me if this observation is incorrect. When R2 = 4700ohms I was getting a ~12 W/m2 well after sunset and into the night. I increased R2 = 4810ohms and now I am seeing ~7 W/m2 at night. Does that mean if I increase R2 = 4900ohms that should give me ~2 W/m2 at night? Then I can set the clipping to 3 W/m2. Assuming, of course, the previous is true.

[werk_ag]

Probably! Try it.
Setting the "clipping" to 4 or 5 will be safer.
Avoid setting R2 above 4900