There are now some Chinese designed and manufactured lux meters being sold on Ebay and Amazon websites. These also use silicon photodiodes as a sensor, and they have a filter on the photodiode that changes the above spectral response to:
Because this meter is measuring light intensity in lux and not PAR units, it is measuring primarily green light, and the filter used to adjust the silicon photodiode spectral range largely eliminates all light below about 500 and above about 640 nm. This makes the photodiode plus filter unsuitable for use in a PAR meter, even with added filters. Infrared blocking filters are not readily available, are normally made of glass, and are expensive. A lux meter can be modified to read in PAR units, but to do it economically a new photodiode with a more appropriate spectral response is needed.
Fortunately, there is at least one silicon photodiode that does have a more appropriate spectral response, the Excelitas Tech VTB8441BH, costing only about $5-$6 depending on shipping costs and how many you buy. The spectral response of this photodiode looks like:
You can see that the Excelitas photodiode spectral response covers the entire 400-700nm range of the PAR spectrum. With proper filters it should be possible to convert that spectral response into something at least as good as the Apogee Quantum PAR meter spectral response.
Rosco is a company that manufactures gel filters, primarily for theatrical use, adjusting the colors of lighting used in stage productions, and for photography, to adjust he colors of studio lighting. The Rosco website, http://www.rosco.com/filters/roscolux.cfm
is a great source of information about the filters they make, giving the spectral energy distribution curve (spectrum) for light transmission through each of the hundreds of filters they manufacture. You can use the data from those spectral energy distribution curves to calculate the spectral distribution of light transmitted through any combination of those filters and the spectral response of a photodiode which has those filters between the light and the diode. This is easy to do, using a spreadsheet and the tabulation of transmissivities for each increment of light wave length, which is also on the website. The data for a typical filter looks like:
A typical spreadsheet, using this data is:
After about a year of off and on calculating, with various filters and combinations of filters, this is the final result, showing the spectral response of the Excelitas photodiode plus 3 filters, #3313 (Tough 1/2 Minus green), #373 (Theatre booster 3) and #1995 (Thermashield):
Note that this duplicates the Apogee Quantum PAR meter response at the low end, and expands it just past 700 nm at the high end. A PAR meter based on this sensor should be at least as accurate as the Apogee meter.
Actually making a PAR meter, using this information, is a bit complicated, but I will post details about that in the DIY forum.