Hi,
I am currently doing research in compressive sensing. I have successfully built a single pixel camera using the DLP Lightcrafter Evaluation Module.
I currently run binary frames at 30 fps, HDMI external streaming patterns. I have tried to run 60 fps, but there are missing patterns (roughly 10-20 missing patterns in a 60,000 pattern sequence). Further, if I miss one measurement for one pattern, the measurement is useless. Everything needs to be there. I would like to speed it up to 60 fps. It is not a limitation due to the detector (a photomultiplier tube).
Part of the problem is that each pattern projected by the DMD is not on for the same length of time. I have noticed a very systematic variation. There are three lengths of time that the patterns are on for. A short, middle and long time. The vast majority are the middle length. The problem is that some patterns are on for 22% shorter and 22% longer. It is quite specific. There is some variation (1-2%) for each of short, middle, and long, but this is understandable. I am concerned with the very systematic and reproducible length of time that the patterns are on for. There is no other length besides the short, middle and long.
My research group is about to purchase a new computer and would like to know if this problem could be solved with a different computer. Right now we are running it with my 6 month old Lenovo X3 Carbon Thinkpad. It has i7-5600U processor (4MB cache up to 3.2GHz) 16 GB RAAM and 512 GB SSD PCIe. It is a nice computer, but there is still significant room for improvement.
Can you tell me if the pattern variation length is likely due to the computer or just the inherent limitations of the lightcrafter.
If it is the computer, would: HDMI 2.0, a faster processor, more RAAM help? Further I use windows 10 right now but had the same problem on windows 7. Could a different operating system solve the problem.
If it is the lightcrafter, do the more expensive lightcrafters perform better at higher FPS?
We are analytical chemists so this is not necessarily trivial for us.
Thanks,
John