Yep. Just get some cds and replace the temp sensor with them. I did add a variable resistor to my circuit so I could fine tune to keep the readings in the range of the sensor. I adjusted mine during solar noon to make sure I was getting max readings.
No, the modification will replace the temp sensor in the THGR800/810 with resistors that are sensitive to light, hence the solar readings. The temp sensor in a jar a completely different animal that uses a thermal sensor to measure the actual temp inside a jar to figure out the solar condition. More heat, less clouds.
Hi NorCalDan,
I've completed the modification suggested addding a photodiode and two resistors and knowing little about electricity, I'm not happy with the result. Some of the weak parts of this solution you have already mentioned in another thread. The main problems I see today with this solution is that the output seem not to be linear but logaritmic - the unit is very sensitive to low light conditions and goes up steeply, then it does change very little during the full day different light conditions. This is why it is dificult to setup cloud condition thresholds and they are very close to each other and more importantly it is totally biasing the WD interpretted irradiance figures.
I've used Siemens BPW21 photodiode. This is very precise silicon based photodiode with a built/in filter that corrects the silicon natural responsivity in order to achieve the sensitivity similar to the human eye. This is actually the part recomended at the german site.
There are some fundamental doubts about using this part at all
- first, we want to measure
iirradiation not
illumination. This part would be just pefect for measuring light intensity (meaning the light is what we see by eye), but it would ignore important part of the spectrum in the IR zone. The conversion of illumination to irradiation can not be correctly made unless the spectrum composition is known and the spectrum is changing under different conditions (clouds for example). Not a simple formula conversion.
- Second the BPW21 seem to be limited somewhere around 30klux according to its datasheet, the question remains if the lienarity is there for higher illuminations. I will check this when calibrated luxmeter arrives, but my guess is it will not work correctly. For this range, there is another type made by Vishay BPW21R that should be performing linearly up to 200klux, where full summer day noon should be around 100-120 klux.
- reading the other posts where people are adding various photodiodes, often several (connecting them linearly, paralelly?) the result is a mix of completely different opto characteristics and thus very biased results that in my opinion are far from any proper measurement. We are mixing apples and bananas in a hunt to have solar sensor, the question really is what do we have in the end.
So I decided to start a project that would arrive with standard solution for solar radiation measurement using modified THGR800/810. The output should be reasonably precise so that the measured values can be correctly converted into W/m2
- Choosing the right photodiode
I've looked for several photodiodes and the most apropriate one seems Siemens BPW34.
http://www.datasheetcatalog.com/datasheets_pdf/B/P/W/3/BPW34.shtmlThis photodiode registers even the infrared part and its current characteristics is calibrated towards the W/m2. The range goes above 2mW/cm2 which should be 2000 W/m2, sufficient sensitvity in full operation range. It is also about 10x cheaper than the original BPW21, good for budget!
- Checking for built/in original termistor linearity corrections in THGR sensor
The original termistor in THGR810 looks very strange to me, it appears to be actually two extremely small parts connected in parallel. The tip deffinitely is the termistor, fast reacting to the ambient temperature. The middle part I'm still unceratin what it does.. Will post a picture if someone has an idea.
Neverteless the original part is termistor, ie the part that changes it's resistance with changing temperature. Termistors tend to be non-linear and there might be a built in software compensation in the THGR to actually deliver temperature values that are in line with the used termistor charcteristics. This built in compensation could interfere with linearly performing optocell / photodiode.
Logical choice would be to use photoresistor, BUT - photorestistors are non linear to the light conditions too (similar to thermistors) and they are very dependent on the ambient temperature. The same light/irradiation measured in the summer morning and winter noon can be interpretted as two resistance values far from each other. They are also slowly reacting nad have a "memory" effect... not good for us.
- designing the circuit that would transform measured values into output that THGR could interpret
Well this will not be an easy one. Already the basic modification is suggesting the photodiode is connected opposite to its photoconductive mode, i.e. positive voltage on the cathode. This is standrad setup in most of photodiode aplications - photodiode is installed reversally and it is the reverse bias current measured (and amplified). It is fast, current linear to the light conditions etc...
BUT
The HGR seem not to have positive and negative volatge on the two conections. Did some measuring using osciloscope and it is clear that shortly before diode blinks, there is a frequency applied (could be electronics switching on and preparing for taking the sample) and then there are number of impulses going each with different polarity.
I'll try to enclose the signal I got if someone can verify.
If you have modified your THGR, you can try this yourself - any direction you connect your photdiode, you get almost identical reading. photodiode can be used in both directions but to keep the linearity, reverse biased configuration is the best. zero-bias operation actually makes it work in photovoltaic mode (light striking the PN junction creates bias)
There is probably only one direction of the THGR signal used for measurement, the question is which one. Could be that tiny little part at the termistor is there because of the signal used??
- calibrate the electronics
Using a proper solar radiation meter should be easy but hopefully arrive to a simple hwo to that can be then used by any weather amateur to reasinably calibrate his/her device, so that the data gathered are comparable.
- Check for data acquired by WD
I did get some data and I believe the current configuration solar in jar settings in WD does not allow for correct conversion to radiance values. The range defined should be for full summer soltice, ie.e the max the unit can indicate. It is normal that winter days we will be 50-60% of the summer values, yet on clear winter day it shoul be interpretted as nearly 100% of the day max. I thought that this is taken care of by disabling "update the solar value from its max reading for his lat/lon" but it had no effect over the graph and interpretted irradance. In this way, having perfectly calibbrated sensor with linear response would not help as WD would recalculate the value to its percentage using th eactual value / annual max. value in stead of daily max value...
More about this here...
http://www.weather-watch.com/smf/index.php/topic,48510.msg392434.html#msg392434Any of your experience gathered during your experiement are wellcome.
I would just love to have a decently precise solar radiation measurement using a wireless units, I do have very limited possibilities for cabled solutions using 1-wire or labjack.
