Calibrating the sensor as it was didn't work very well. It was too sensitive to the angle and not as linear as I wanted, plus it was too sensitive - too high readings. So, I coated it with two coats of white fingernail polish. On recalibrating it, it worked fine, but with a different calibration for LED light and 6500K CFL light. That is ok with me, since I will only use it with one of those. I now dub this the "PAR Volksmeter".
You can see that this has sensor a little smaller than the Quantum PAR meter. It is also easier to use than the Quantum meter for low PAR readings because of the 3 position switch on the readout.
The calibration is:
I will let the nail polish dry a few hours, then test it underwater. I see no reason for it to either leak or change calibration underwater.
Success, an easy DIY of a cheap lux meter to get a very cheap "PAR" meter, usable for most of our uses. I trust the calibration with LEDs more than the one with the 6500K CFL, because the latter was much harder to do, due to the small area of highest light under the CFL bulb. The cost: about $15 plus shipping for the lux meter, about $5 plus shipping for the photo diode, about $2 for the acrylic parts, $2 for the electric cable, plus odds and ends.
I recalibrated the PAR Volksmeter today. Yesterday I modified it by attaching a strip of acrylic to the bottom of the sensor, with a post at one end that will fit snugly into a 3/4" diameter acrylic tube, as a handle, so it is much easier to position and hold accurately.
Then, after a lot of thought I realized how much accumulated error is in each calibration point. For example, the PAR meter is only accurate to +/- .5 micromols, which is a large error at low values of PAR. Then just tiny mistakes in distance from the light source makes a big error in PAR readings. Plus minor mistakes in location of the two sensors adds another error. So, I tightened my positioning control, and weighted the data points by how much possible error each data point had. That gave me a big change in calibration constant, so I corrected the chart above. It also made the calibration constant be the same for the LED and 6500K CFL lights. That constant is 80 +/- 5 Lux per PAR unit. The chart above doesn't include a couple of data points that were way outside the area covered by the chart, but those points were included in calculating the constant. (Very high PAR means the sensor is close to the light source, where errors in distance are very significant and very hard to eliminate.)