[danner]

sounds good, thanks.

[danner]

Grabbed my flashlight and removed the solar/uv board, adjusted it to 4.9kohms in the garage, and removed the ADC1 wire. Unfortunately there is no change to the readings. It was 7-10 W/m2 before the change and it is fluctuating between 7-9 after making the changes. I can set it to clip at 10 W/m2 for now. Might have to set it to 11, will have to see what it puts out.

Any chance there is a problem with the TX board? How about the SP-110 itself?

[AllyCat]

Hi,

You're not measuring the resistance between the "ends" of the pot are you?   It should be between the "wiper" (ground) and the end that is connected to the Op Amp input.  The value should be about 3.33 kohms to give a gain (amplification multiplier) of 4.  Increasing the resistance gives lower gain, which you possibly need because at 1500 w/m2 the SP-110 should give 300 mV output. 

I believe to give "full scale" with the 1.1 volts nominal reference voltage of the Nano needs a gain of 3.667, which I calculate requires 3.75 kohms.  With the wiper at "end of track" (5 kohms) the gain should be close to 3 to give a "full scale" (limiting) ADC reading at around 1833 Watt/m2 (I don't know what scale factor the software uses).

But all this should make no difference to the "zero" (night time) value.  Zero times 3.75 (or any value) should still be just zero!  So in the "dark" you need to discover what isn't zero, which with a good multimeter shouldn't be too difficult:  

Measure the voltages on pins 1, 2 and 3 of the Op Amp relative to a good ground (probably the PCB track, wiper of the pot, or ground connection of the SP-110) to a resolution of better than 1 mV.  Those three voltages (which could be measured on the two ends of R1 and the PV input) should give an indication whether it's the SP-110, the Op-Amp (input offset error), the Nano ADC (zero reference error) or something else that is "wrong".

Cheers,  Alan.

[danner]

Hi,

You're not measuring the resistance between the "ends" of the pot are you?   It should be between the "wiper" (ground) and the end that is connected to the Op Amp input.  The value should be about 3.33 kohms to give a gain (amplification multiplier) of 4.  Increasing the resistance gives lower gain, which you possibly need because at 1500 w/m2 the SP-110 should give 300 mV output. 

I'm measuring between pin 2 and the ground pin on one of the screw terminals with the connector removed from the TX board and with the SP-110 removed from the Solar/UV board.



I was considering swapping out the TX1 board and nano, or moving the Solar/UV to my other transmitter, TX0. Either way that should allow me to see if the behavior follows or goes away.

But all this should make no difference to the "zero" (night time) value.  Zero times 3.75 (or any value) should still be just zero!  So in the "dark" you need to discover what isn't zero, which with a good multimeter shouldn't be too difficult:  

Measure the voltages on pins 1, 2 and 3 of the Op Amp relative to a good ground (probably the PCB track, wiper of the pot, or ground connection of the SP-110) to a resolution of better than 1 mV.  Those three voltages (which could be measured on the two ends of R1 and the PV input) should give an indication whether it's the SP-110, the Op-Amp (input offset error), the Nano ADC (zero reference error) or something else that is "wrong".

Thank you for that information. It is very helpful and much appreciated! I can do that.

[hornychz]

Let me put myself a bit in this debate. Somewhere here (about Werk_AG?) has already been said that measurement
in the mV area is quite challenging and the results in many cases are not relevant. Among other things, this is mainly
due to the electromagnetic field in which the measurement takes place - the proximity of the power supply, the RF transmitter,
the ground loop, etc. The 1 m of the measuring conductor can be induced to voltage up to tens of mV. We help with
a low impedance of the load, but it is necessary to think of this - therefore also the shortest conductors, twisted, shielded etc.
Simple verification - by measuring conductors we check the voltage on a resistor that is not connected anywhere - eg here
in the size of 3K3. Very often, we do not need to measure zero.

[werk_ag]

I believe to give "full scale" with the 1.1 volts nominal reference voltage of the Nano needs a gain of 3.667, which I calculate requires 3.75 kohms.  With the wiper at "end of track" (5 kohms) the gain should be close to 3 to give a "full scale" (limiting) ADC reading at around 1833 Watt/m2 (I don't know what scale factor the software uses).

Hello Alan

Thank you for sharing your thoughts on this matter.
Can you please give me some light about how you did your calculation?
Perhaps I'm wrong, but I believe that to get the full scale from the SP-110 the variable resistor should be 4666 Ohm

http://mustcalculate.com/electronics/non...=&r2=10000
   

On the solar interface R1 is R2 and R2 is R1

Same calc from other source
http://yano.wasteonline.net/electronics/...=&rf=10000

   

On the solar interface Rf is R1 and Rg is R2

[werk_ag]

Dan,

For testing purposes, you can try connecting the SP-110 directly to the ADC0 input on the TX board. Then, on the RX software, comment this line

Code:

       SolarRad = map(RX_Data[2], 0, 1023, 0, 1600);              // Using OP - Formula outputs an integer, between 0 and 1600 W/m2

and uncomment the next line.

Code:

       //SolarRad = round(((RX_data[3] * (1100 / 1024.0)) * 5);   // Not using OP. Sensor connected directly to ADC input

result should be like this:

Code:

     #if (Solar_Sensor == 2)      // Apogee SP-110: Self-Powered Pyranometer (Solar / UV interface R2= 4700 Ohm)
       //SolarRad = map(RX_Data[2], 0, 1023, 0, 1600);              // Using OP - Formula outputs an integer, between 0 and 1750 W/m2
       SolarRad = round(((RX_data[3] * (1100 / 1024.0)) * 5);    // Not using OP. Sensor connected directly to ADC input
     #endif

[danner]

I see I missed some posts. Sorry about that. Good news and bad news.

The good news is I swapped out my TX1 board using the same Nano and it appears the ghost readings are gone, so whatever was causing the readings is likely related to the TX board. But as with everything, time is the ultimate arbiter of truth. And now I have the unenviable job of figuring out what is wrong with the TX board that worked fine with the wind and rain, but not solar.

The bad news is now it appears there isn't enough gain, LOL. The sun is almost set, but the overcast sky is still back lit, and the other stations nearby are still showing solar readings, and mine is at 0.

While I was swapping the TX board out I reset the Solar/UV board to 4.7k ohms and I put the ADC1 wire back and twisted with the ADC0 wire. But at least it is now reading 0! Also set the clipping to 0.

[AllyCat]

Hi,

.. at 1500 w/m2 the SP-110 should give 300 mV output.   ....  I believe to give "full scale" with the 1.1 volts nominal reference voltage of the Nano needs a gain of 3.667, which I calculate requires 3.75 kohms. 

@ Werk:  Sorry, I split apart my first two paragraphs at the last moment; I was considering 1500 W/m2 (i.e. 300 mV) as "full scale" in my "back of envelope" calculation (actually done in my head).  I don't know (didn't check) what scaling factor the software actually uses;  I think we agree that the minimum gain (at end of track ~= 5k) is at, or just over, 3.00   .

@danner:  So it looks as if there is a "zero offset error" on the ADC input of your other Transmitter.  Try connecting the "solar" input pin (preferably through a "safety" resistor of say 1 kohms) to various points on the PCB which "should" be earth (zero volts) and see what the software reports (preferably with little or no clipping applied).

I think we should assume that the "precision" calibration aspect of the SP-110 applies to high level (i.e. peak daytime) solar measurements.  As Zdenek says, measurements around and below 1 mV are always difficult.  Also, there are many other "complications" in making measurements when the sun is at low elevation (close to the horizon).  In particular, the "Cosine response" of the (required) vertical-facing sensor gives a much higher proportion of the light from the "sky" compared to the sun itself.

IMHO if you want to improve the reporting of "sunrise/sunset" levels, the Op-Amp needs both gain and offset ("zero") potentiometers.  Perhaps you could use the UV channel amplifier (if you don't plan to use it for an analogue UV sensor).  The Weatherduino PCB is not an "ideal" starting point but you could try the following:

Swap the 5k and 50k pots (of the UV channel) so that the 5k goes to the output of the Op-Amp (it will still be used to set the gain, but now increasing its resistance will increase the gain).  Put an accurate 1k resistor across the wiper-to-active end of the 50k pot (i.e. Op-Amp input to ground).  Ensure that the wiper of the 50k is always towards the "open" end, then the gain of the Op-Amp will be adjustable (with the 5k) from about 1 to 6.

Then the 50k (R4) might be used for the zero offset adjustment, but I'm not sure how best to do that yet.  It will need the addition of a few high value (Megohm) resistors and a slight modification to the software (subtracting a few W/m2 to accommodate the "sit up" of the output voltage from the Op-Amp).  More details if you decide to follow this route.

Cheers,  Alan.

[MikeM]

As a side note, for increased ease of adjustment, the pot can be replaced by multi-turn (10-turn) presets.