are my intervalometer settings wrong?

[calypsob]

Using my 550d with ML I am shooting timelapse for astrophotography, lets say 300s images "5 minute".  When I get home I always have data which is 1 second short, some are 299s and the rest are 300s.  Does anyone know why this happens?  It messes me up because I need to later add dark frames and end up having to take 299s and 300s subs.   

[a1ex]

Just use the same dark frame for both.

Imagine the actual exposure would be between 299.99 and 300.01 seconds. Guess what will happen when truncating it to an integer.

If you have time to test, adjust BULB_EXPOSURE_CORRECTION in consts.h and report if you can find a better value.

[calypsob]

I tried stacking the data today and got some strange results with the 50s and 49s darks combined together.  Fortunately I also had a set of staggered 49 and 50s dark frames so I had almost and equal amount of darks to calibrate together and independently.

on my first try I calibrated everything together and scaled the darks because I used 50 and 49s.

light_BINNING_1_integration
Calculating noise standard deviation...
* Channel #0
σR = 9.257e-005, N = 1909520 (10.59%), J = 3
* Channel #1
σG = 8.183e-005, N = 321791 (1.79%), J = 4
* Channel #2
σB = 9.191e-005, N = 338895 (1.88%), J = 4

on the second try I applied calibration frames independently to 49s and 50s.  After applying indepent calibration frames I stacked the 49 and 50s frames together and got much lower read noise, but it was a little higher in green? 

integration
Calculating noise standard deviation...
* Channel #0
σR = 1.072e-004, N = 212831 (1.18%), J = 4
* Channel #1
σG = 8.856e-005, N = 365346 (2.03%), J = 4
* Channel #2
σB = 1.001e-004, N = 395618 (2.19%), J = 4


A1ex, I will try and figure out how to adjust the script to give consistent subs, thanks for the tip. 

[a1ex]

I'm not sure what to look for - your second data set has higher stdev values on all 3 channels...

Also, why did you scale the darks in the first step? The difference between the two sets of frames is likely much smaller than 1 second (see my previous post).