If that question is for me Terry then you either want a 4” inch diameter “P” trap which is what mine is or what a lot of people use 3 x 4” diameter swept 90” degree bends.
You can use push fit or solvent weld (glue) & if you get ones with male & female ends you won”t need any 4” diameter pipe to join them together 
Then you can get either a plastic or rubber 4” diameter blanking cap to mount the light sensor head in one end
This is what I ended up getting at lowes. Ot should work once I get the meter attached and pit some foam around the top hole...
I painted the inside of the box, still needs drying time...
Lots of info in this thread on mystery pipes.
I think there a slights issue with the my metering app. It was working great until I painted the inside. Now everything us reading about 90 percent low.
My meter will be hee in a little while and I will check it again
When you have the time do me a favor please.
Let me know what the Ft candle reading was from that review. I'm just curious.
Thanks very much...
since Im only testing AAA lights, this works for me
the vitamin jar cap has a couple of pieces of paper to “calibrate” it so I can read lumens without a calculation.
“calibration” is based on a ReyLight CuTool that has a claimed 1 lumen low.
here you can see the calibration is reasonably accurate, the pictured Worm has an 80 lumen spec on high:
Okay I had to do my box all over again, and finding a second one was not as easy as one might think.
I decided to calibrate it totally from scratch. I have a ton of Lux Pro, Ozark Trail and Rayovac almost indestructible lights. One of the good things about these is that first they are inexpensive, and most of the time the lumen output is what the manufacturer claims.
I have gone through about 8 getting a good calibration, here are some of the results of some higher lumen lights. Most of them I did with 2 different 18650 cells.
Thorfire G-10
Highest setting using a Panasonic NCR18650PF Protected 2900mAh
Current was 2.31A
789 Lumen
Same light same setting but with an unprotected 30Q
Current 2.69A
879 Lumen
Next was the last review I did the Rofis TR-20
Using the Rofis 18650 cell provided with the light current was 2.68A
Again Highest setting which is turbo on this one 831 Lumen
Same light same setting this time with the 30Q
Current 3.07 and 930 Luman
And just for kicks I did my Jetfire C8 using a Panasonic 18650 3400 mAh protected cell. Did not have time to check the current but I will post it after while.
862 Luman
Same light again highest setting with the 30Q
918 Lumen
I will be testing different things most of the night and almost all of tomorrow.
I still have to get the pipe calibrated, hoping it goes as well.
The problem with a 3 piece 90 degree pipe is you wont get proper light integration.
This means that if you put a 1000 lumen 300kcd thrower and then compare a 1000 lumen 10 kcd light they wont read equally because the high intensity of the 300kcd is not tumbled or mixed enough and will read more than 1000 lumens.
How i overcame this is to use 4 sections of 90 degree pipe and a straight dropper pipe 200mm length to the lux metre. This allows all the light to integrate properly and now means a thrower or a flooder of equal lumens will measure the same. For example i can put my tn32 1550 lumens 200 000 candela light and measure accurately between this and a little aaa olight 13s [ 400 candela 80 lumen] for accurate lumen measurement.
For the inlet pipe i used a cut down clear heat resistant welding helmet protective external lense which i place my lights on. I then cut white cardboard with the right od to fit the light tightly and to prevent any light escaping. I also sprayed the whole outside of the light pipe with two coats of grey etch primer to stop light escaping through the pipe.
I made mine a couple of years ago taking about a year to get proper calibration and integration with lights like the tk75, tn32 zebralights sc series, i3s and many others with known genuine outputs and varying candela.
Now 1 lux on my pipe and meter equals 5.35 lumens . So for example my zebra sc600 mk11 cw at 30 seconds measures 198 lux [198 x 5.35 = 1059 lumens.]
It is very useful, you can plot full runtime outputs, indentify step downs, check your mods output success or failures , check outputs of the same light with different cells and of course validate your newly purchased lights are indeed as powerful as claimed.
My biggest reason for even attempting the pipe is that it is just easier to keep around and move. The boxes are a little bulky and get in the way. And now that I finally have it very close there is the chance that some trips over it and all that will be left is white stuff all in the floor.
I am sure it will take me much longer to get everything just right. I have 3 90 degree, and like you said may still have to add another 90.
I like the challenge, when I got the box calibrated well enough to trust, just another hurdle, now on to the next one.
Has anyone played around with the placement of the sensor, like on the left of the light at what height, or right, or above, just curious?