I will explain a couple effects, and how they can happen. This is just for future reference.
1. If we have only have two triggers, T1 and T2, T1 rates is about 1kHz, T2 rates is only about 10Hz. We can still have the situation T1's dead time is small, but T2's dead time is large. The reason is that if T1 is behave like a pulser, it will almost never see itself, but T2 can see a lot of T1, Thus, T2's dead time can be larger.
2. prescale factor play an role. The larger prescale factor, the more your trigger looks like a pulser, for a real pulser, the dead time should be quite small, (low rate), since it can not see itself.
3. In principle, for a pulser, it won't see itself, due to argument 1 and 2, however, it's rate could also have effect on the live time. The reason is that the trigger before the key trigger can cut into the effecti dead time window for this key trigger, in this case, the effective rate is R*(t-delta_t) instead of R*t, then the dead time could be smaller.
With these effect in minds, we can try to understand the following data.
1. Cosmic data Trig perscale CODA scaler live time
1 1 17516 18001 0.968
2 16777215 0 1764 0.000
3 1 2787 2876 0.964
4 16777215 0 1133 0.000
5 1 1710 1755 0.970
6 65535 0 18334 0.000
7 65535 0 1755 0.000
8 100 1408 143753 0.975
The dead time is small, everybody is pretty far apart, thus the dead time is quite close.
2. beam data
1 70 100563 7057815 0.997
2 16777215 0 43 0.000
3 16777215 0 94915 0.000
4 16777215 0 3392 0.000
5 65535 0 3350 0.000
6 65535 91 7223528 0.826
7 65535 0 182 0.000
8 100 1364 158238 0.862
Trig 1 can not see itself and it almost can not see 6 and 8, since the rate are too small. Trig 8 can see both Trig 1 and Trig 6, that's why it's live time is smaller than trig 1. Same argument apply to Trig 6, the reason for Trig 6 live time smaller than Trig 8, can be two. a) beam trip, b) statstical effects.
3. a more complicate situation.
1 250 86307 26073644 0.824
2 16777215 0 291986 0.000
3 20 13407 347965 0.767
4 16777215 0 35752 0.000
5 1 10078 12722 0.789
6 60 79266 5782337 0.819
7 65535 7740 656617422 0.769
8 100 4198 506742 0.825
In this case, every T6 should be T1, every T5 should be a T1 and T3, I do not know much about T7.
First, we can easily understand why T3's live time is smaller than T1 and T6. Simply because T1 and T6 see little of themselves. due to p1. But T3 will see them all.
Second, T6's live time is smaller than T1, but not much smaller, is probabily due to the correlation. T6 is blocked more by T1, but T1 is not. due to p1 and p3.
The reason of T3's live time is smaller than T8, is probabily due to beam trip.
T1's live time is bigger than T5, since it can almost not see T5, but T5 can see T1 a lot.
T3's live time is similar to T5, because their rates are similar. They are see a lot T1 and T6, not like T1 and T6. However, T3's live should be smaller than T5, since the T5 trigger correlated to the T1 trigger, which means T5 should see less T1 than T3.
A real understanding of the dead time require a full simulation. Hopefully, we can get a feeling of it with these very simple and naive arguments. Please correct me if you think there is anything missing.
A copy of this log entry has been emailed to: rom@jlab.org,kalyan@jlab.org