ap--
Implementation dark current and nonlinear correction
Here's some code for the implementation of the dark current and nonlinearity correction. When using the the dark current correction it is possible that the datasheets are wrong about which pixels can be used, at least that was the case for the QE65000.
When using the the dark current correction it is possible that the datasheets are wrong about which pixels can be used, at least that was the case for the QE65000.
The data sheet suggest 0:10, 1034:1044 via USB. The seabreeze library uses 1025,1026,1027 You are using 1,2,3,1038,1039,1040,1041
How did you get your indices?
Looking at all other spectrometers: Your values differ with all of them compared to the seabreeze library. We should double check with the datasheets and see if there maybe is some extra documentation on the topic provided by oceanoptics.
Update: I just posted the question on the seabreeze source forge repository. They should be able to help.
First I used the indices from the datasheet, and noticed that the average was above 2600, when measured by switching of the illumination it is usually ~2400. So I just printed all the dark pixels and just used the ones which had a value of about ~2400, the "bad" ones had ~3000. Interestingly SpectraSuite seems to make the same mistake, when activating the dark current correction the baseline drops to -200 instead of 0. For the other spectrometes I just used the values from the data-sheets, I didn't look at the seabreeze library but that would be a good idea.
Hi Simon, @mzieg is working on a fix for this. The dark pixel indices have to be double checked for each individual detector model. (Exactly like you did.)
Until then Issue #16 stays open.