I know that I had similar difficulty when testing my Speed. My light source was too big and my sensor was way too far from the FP shutter. Switched to a very focused beam (sounds like you did a similar thing with the soda straw) and moving the sensor closer to the FP to mimic the film plane made a huge difference in my measured speeds. My 1/1000 tests out at 1/500 and 250 and below is close enough for my work. Try using a laser pointer--very little scatter. I also thought about using the flash sync conductors on the FP to time the speed of the curtain and do the math to figure out the accurancy but then I decided to just get out and shoot--way more fun.