Thread Rating:
  • 6 Vote(s) - 4.33 Average
  • 1
  • 2
  • 3
  • 4
  • 5

Measuring Solar Radiation based on Photovoltaic Cells

Hello,

I have now build the Solar radiation sensor and fine tuned its calibration.

At first I had a gain of about 26 (assuming the same values taken in the example published in a different post on solar radiation measurement) and at mid-day the values read were capped at 1400W/m2 in Cumulus. So I've adjusted R1 and R2 to get the exact values giving me a gain of 6.875. Then I readjusted by measuring the PV cell output at a constant irradiance and reading the board output to get the same gain.

Now the values are pretty similar to the national irradiance map published here in Singapore and refreshed every minute! It's a color mapped, so not extremely precise, but when the zone of my weatherduino is of the same color as the neighbourhood, we can deduce the value of the irradiance. I still need more fine tuning though.

My question is on Cumulus: how to adjust the original green gaussian center? Currently it's entered at 3am.. I need to center it around 12pm, right? I tried to find some configuration on the trendT.html or the js files but no success.

Thanks!
Reply
Tongue 

(17-08-2016, 16:36)zitoune Wrote:  Hello,

I have now build the Solar radiation sensor and fine tuned its calibration.

At first I had a gain of about 26 (assuming the same values taken in the example published in a different post on solar radiation measurement) and at mid-day the values read were capped at 1400W/m2 in Cumulus. So I've adjusted R1 and R2 to get the exact values giving me a gain of 6.875. Then I readjusted by measuring the PV cell output at a constant irradiance and reading the board output to get the same gain.

Now the values are pretty similar to the national irradiance map published here in Singapore and refreshed every minute! It's a color mapped, so not extremely precise, but when the zone of my weatherduino is of the same color as the neighbourhood, we can deduce the value of the irradiance. I still need more fine tuning though.

My question is on Cumulus: how to adjust the original green gaussian center? Currently it's entered at 3am.. I need to center it around 12pm, right? I tried to find some configuration on the trendT.html or the js files but no success.

Thanks!

Interestingly, it seems the green gaussian has shifted automatically to get entered properly. Good news Smile
I assume it needs a couple of points (2 days?) to be able to synchronise.


Looks like my calibration is not too bad after all. As you can see the sensor is placed now in a space where the direct sun light is only after noon, for about 2 hours, due to high buildings around. But never-mind it works!

Regards.


Attached Files Image(s)
   
Reply

Hi,
(17-08-2016, 16:36)zitoune Wrote:  in Singapore ....... I need to center it around 12pm, right? !

In Singapore, I believe it should be centred around 1pm. Singapore time is about one hour ahead of sun time, GMT + 8 hours (I believe because they use the same time zone as Malaysia, but Malaysia coordinates its time with the Eastern island).

The "maximum solar radiation" line should be correct in Cumulus provided you enter the correct site Longitude (it was about 104 degrees E, when I was there Wink ) and of course the local time is set correctly.

I can't see your solar.tiff, but certainly a gain of 26 seems far too high.

Cheers, Alan.
Reply

(18-08-2016, 10:52)AllyCat Wrote:  Hi,
(17-08-2016, 16:36)zitoune Wrote:  in Singapore ....... I need to center it around 12pm, right? !

In Singapore, I believe it should be centred around 1pm. Singapore time is about one hour ahead of sun time, GMT + 8 hours (I believe because they use the same time zone as Malaysia, but Malaysia coordinates its time with the Eastern island).

The "maximum solar radiation" line should be correct in Cumulus provided you enter the correct site Longitude (it was about 104 degrees E, when I was there Wink ) and of course the local time is set correctly.

I can't see your solar.tiff, but certainly a gain of 26 seems far too high.

Cheers, Alan.

Hi Alan,
you are totally right - I just checked and 13:08pm seems the exact time nowadays. When I move the cursor on the curve, I find the maximum at about 13:10pm.
I have re-attahed the picture in jpg - you see it ?

Thanks.
Reply

Hi,

Yes, your graph looks fine now. The sky seems to be clearer than when I was there in March. Wink

Beware that the sun is not a good timekeeper, solar "noon" can vary by as much as +/- 20 minutes during the year! If you want to know why, then google "The Equation Of Time".

An easy way to determine the exact time of noon is that it's half-way between sunrise and sunset, as quoted in Cumulus. It's very easy for me because I'm almost exactly on the Greenwich meridian and currently noon is showing as +4 minutes, i.e. 1.04 pm because we're on "British Summer Time" (or Daylight Saving Time). So your 1.08 pm seems exactly correct if you're 16 degrees * 4 minutes = 1 hour 4 minutes ahead of the sun.

Cheers, Alan.
Reply

I found this picture some time ago at an ebay listing.

   

It seems that some company was already produced a commercial pyranometer, based in the principle of a solar cell.
Maybe the picture gives some ideas for our own constructions.

Reply

Happy to have found this thread today as I was looking to have an inexpensive way to measure sun irradiance.  I looked for cheap pyranometer but the cheapest one is way too expensive for a weather station built just for fun.  

Looking at this project, the most difficult part is certainly the low voltage output and the need to have an opamp.  Somewhere in the thread, someone (alan I think) came with the idea to use a panel with more cells.  I think the real benefits were not seen.  The graph shown on the thread's first page is for one cell giving 0.5 to 0.6 volts.  Uncle_bob calculations are based on that : one cell solar panel.  If you use more cells you multiply the voltage by the number of cells, current being the same, which allows to get rid of the opamp. 

To give an idea of the values we could get with 10 cells solar panel, target voltage is then 1,6 volts.  So for 1600 W /m2 we now have 1600 mv or 1 mv /w / m2 (10 times lower).  With arduino internal ADC, this gives 4.88 W / m2 as the smaller step (again 10 time better than original design), which is quite good and really easy to do. Using 2 resistances for shunt to make a voltage divider (total becomes 17,33 ohms) we could use the 1,1 v arduino ref voltage and get 1,07 W /m2 as the smaller step and again no opamp needed ... Too good in fact for other imprecision factors like no real good reference irradiance.  
 
Some time ago, to make tests for building a sun irradiance meter, I ordered a 10 cells panel which is a 5 volts solar panel. In fact this is a 2 times 10 cells to have higher power.  I can expect higher short circuit current. As expected, open voltage is 5 to 6 volts.  I will make calculations and test this but here's what I expect with this panel : 

Short curcuit current : 0,27 amp for 810 W /m2 (bad conditions for a test today so I took a nice irradiance number to facilitate calculations)
Shunt resistance : 3 ohms
Maximum irradiance : 1500 w / m2 (1000 would probably be better for Quebec)
Max voltage measured: 1500 mv
1500 W / m2 -> 1500 mv or 1 mv / W / m2 
arduino gives 4,88 mv / step -> 4,88 w / m2

Also, with arduino, there's an easy way to get better precision : 16 bits ADC which has 4 inputs to do other things (like wind direction which is analog output for me).  

J Guy
Reply

Hi, welcome

The voltage of the solar cells doesn't matter, as what we need to measure is the voltage drop at the shunt resistor (short circuit current), which should be as low as possible to keep the cell at the short circuit zone.

Increasing the number of cells will only make that a lot of power should be dissipated on the shunt resistor.

Maybe you would like to read the original study from Charles G. Wright, on which our solar cell pyranometer is based. Here is the link:  

http://chuck-wright.com/projects/pv-measure.html

Reply

I agree with you we are measuring voltage drop across a shunt but ohm's law says that when in series voltages sources (like batteries) add. Do we agree that cells are "like" batteries ? When in series cells give higher voltage, when in parrallel, cells give higher current. So here, we have 10 series cells giving an open circuit of 5 volts. To keep each cell in the current source mode (as explained in the link), we HAVE to keep each cell to 0,1v and this is done with the shunt. But as total voltage is 10 times higher (parrallel batteries) we have to have 1 volt voltage drop in the shunt. This way each cell will have a voltage drop of 0,1 v exactly like shown in the link.

In fact, another way to see it is to imagine having 10 reference circuits (like the one in the link) and put them in series. Total current will not change, but the total voltage from end to end of the 10 circuits will be increased by a 10 factor (like we add batteries voltage when batteries are in series). Some people working with BWP34 add some in series to higner the generated voltage.

if you think cells are batteries, and we should easily agree on that, it's clear that total voltage from 10 series cells increases by ten. For having each cell in the current source mode, which is the basis of the theory, we have to have each cell at 0,1 volt. With 10 series cells, total solar panel voltage MUST be 10 X 0,1 volt which gives a total voltage of 1 volt for 10 cells and easier treatment.

J guy
Reply

(27-08-2017, 00:03)jgveill Wrote:  Uncle_bob calculations are based on that : one cell solar panel. 

Just to be clear, and not take any credit, I only ran Werk_AG's post through Google translate and reposted it.
Talk maths to me and my eyes cloud over Smile
Reply




Users browsing this thread: 5 Guest(s)