|
Post by r3dunlop on Apr 15, 2015 12:37:34 GMT -8
I am experiencing non-linear gains in my data. This wouldn't normally be an issue, but some of the non-linearities are drastically different. In some crystals I am seeing an effect on the order of over 1%. This means that if I gain match on 240 and 776 keV transitions, I see a change in some centroids of peaks of about 3-4 channels at 337 keV! I'm considering some solutions including 3-4 point gain matching.
|
|
|
Post by r3dunlop on Apr 22, 2015 11:41:18 GMT -8
|
|
|
Post by jsmith on Apr 24, 2015 8:28:31 GMT -8
Jenn and I have both seen a bit of this as well. We've noticed that we could get a linear gain-match to work for E>400, when we only used peaks above 400 keV. From this, we think that most of the non-linearity is below 400 keV.
|
|
|
Post by r3dunlop on May 12, 2015 5:09:32 GMT -8
I have performed "energy calibrations" for a three different digitzers. These were produced by fitting K-value corrected charge spectra. I then took the centroid vs energy, and fit a straight line to these data to get the nominal gain + offset (See next message). I then applied this linear gain to the data and found what the calibrated energy for each peak was in each of the three digitizers. (The plot labeled 0x100a should read 0x000a) This is not good. They each show different "functional forms". I am not sure if the 64 channels can be combined into groups where the ADC's within the group share the same behaviour. Attachments:
|
|
|
Post by r3dunlop on May 12, 2015 5:11:07 GMT -8
This is the gain + offset part for one of the digitizers
|
|
|
Post by r3dunlop on May 12, 2015 5:30:48 GMT -8
The transition at 448 has some strange background and looks like it could be influenced by some background line. That would make 0x000a acceptable. I also did 0x000e, and I get the same looking curve.
|
|
|
Post by r3dunlop on May 12, 2015 7:00:53 GMT -8
Having looked at a few more runs, it seems one might get away with doing two separate gain matchings. I agree that many of the runs appear to have different linearities that switch at approximately 400 keV. Therefore, I am going to proceed with my analysis gainmatching two peaks above 400 keV, and then again with two peaks (if possible) below 400 keV for any transitions required down there.
|
|