There were several issue related to data averging. The VI became too complicated due to unwise choices. modifications involved:
1. Previoulsy the averging and filling buffer was done in same labview. I separated both of them. Also made a subvi called delete buffer which resets the buffer before lc voltage is changed for new data.
2. removed PCI like complicated averging and now m using directly VI from NI for this. This made the labview very compact.
3. The only issue remains when the lc changes the azimuth from -45 to 45 kinda when the voltage is around 1V. The std is high at this point.
So, we have now a rather clean procedure
LC voltage change --->>>> create bufffer VI (initialized array of size equal to average order) --->>>> Fill Buffer VI by loading camera VI(number of times it loads is equal to averge order) --->>>> average VI (takes average of entire buffer) --->>>> Send data for saving --->>> delete buffer
On a good note: The VI datasaving speed is better now. With averging of 10th order the data saving speed is 40Hz.
The STD has reduced to around 0.08 deg in azi and 0.03 in ell at around lc voltage of 0.9V (where the flip happens in sign and ell in wrapped). At all the other voltages of LC the std is 2e-3 deg ish in ell and azi both.
The uncertainity was reduced by adding a wait time(1s) before the polarization camera is read and after lc voltage changes. From my understanding at this voltage of 0.9 ish V the lc molecules take more time to stabilize as compared to other voltages, which was giving rise to high std at just this voltage. Also, the ellipticity getting wrapped was giving rise to high std.
Now, the setup is ready to take measurement with better averaging and more averaging. I have left the setup to warm up. Will start measurement in an hour.