User name R. Michaels
Log entry time 13:11:49 on January23,2004
Entry number 116524
Followups:
keyword=trigger 7 efficiencyI checked run 2044 (waterfall target) for the T7 efficiency.
Firstly, fig 1 shows how the discrimination is working.
Recall that you get a hit in a TDC channel for the summed A1 (A2)
signal when these are above a 30 mV threshold. Normally the
TDC comes at about channel 1500. The four plots are L.a1.asum_c
and L.a2.asum_c requiring a hit or no hit in the TDC. So the
threshold for A1 (A2) are roughly 500 and 150 channels respectively.
The A1 threholds is a bit "high", and A2 "low" -- both conservative.
Fig 2 shows the L.a2.asum_c spectra for T5 and T7 (using the
trigger bits to see if the trigger is present) with the cut
that A1-sum TDC is absent (hence below threshold) and A2 is
above 200 ADC chan. The main point is the spectrum look the same.
One can also get a very rough efficiency for T7 by using the
prescale factor (20) and the number of entries in the plot.
#T7 = 52815. #T5 = 20 * 3426 --> efficiency of T7 approx 77%.
(Is this naive ?).
A second way to estimate the efficiency is to plot
T->Draw("D.evtypebits","(D.evtypebits&0x20)==0x20&&D.a1sumt<5&&L.a2.asum_c>200")
(i.e. plot the bit pattern requiring a T5 and that the A1-sum TDC
is absent, and that A2 sum is > 200). The logic is that all T7's
are T5's but an event with A1 and A2 consistent with being T7 should
never be only a T5. Also I demanded a T5 bit because of its prescale
factor. Hope that's right.... So the result is there are 596 events
which have only T5 bit set in this case, while 2830 events have
both T5 and T7 and the T7 efficency = (2830 / (596+2830)) = 83%
consistent with the 1st estimate.
I think you can run with T7 and keep as many T5 as possible
(200 Hz ?) so that you can later work out the efficiency of T7.
FIGURE 1
FIGURE 2
