After almost 2 years of working on this, trying everything I could think of, I finally shifted my brain into "park", and guess what? This lux meter has a closely fitting cap over the sensor, to block out the light to make the photodiode last longer when it isn't in use. What do you suppose would happen if you left that cap in place, but drilled a tiny hole in the center of the part over the diffuser dome? Duh! It acts as a very effective filter, reducing the lux meter reading as much as you might want to. So, I drilled a 1/16" diameter hole in it, and it reads about 25% low compared to the "PAR" that it should read. No filters needed, just a small increase in the hole size, probably to a #49 drill size (about .073 inches diameter). I will buy a drill bit that size and see if it works. If it does that means no parts, no filters would be required to make the lux meter read "PAR", and you could leave the drilled cover on to read PAR and take it off to read lux. At worst, you would need to add a tiny piece of diffuser filter from the Rosco sample book, over the back of the hole to adjust the reading down a bit.
I started thinking about this when I noticed that even the Rosco neutral gray filters badly distort the spectral response, and will probably make it mis-indicate some types of lighting, where a simple green light measurement would be more likely to be accurate. I should put my brain in "park" more often!
Here is what the modification looks like as a sketch:
Today I found that the best diameter for the tiny hole in the protective cover is .067 inch, which is a #51 drill size. I did this by trying 4 different drill sizes, plotting the resulting meter readings and using that to interpolate to the best diameter, giving the same reading on the lux meter as on a PAR meter. I still need to buy a #51 drill and try it to be absolutely sure of the size needed.
This is an absurdly simple way to make a usable PAR meter. Drill a tiny hole in the protective cover. Use silicone sealant to make the sensor body waterproof. (Silicone worked best for this, by testing various sealants with partially assembled sensors.) The lux meter is then unharmed and, by removing the protective cover, reads lux as it was designed to. This enables you to read from zero to about 50,000 PAR! Take it on a trip to the sun and measure the PAR there!