2007 EEMC Tower Gains

Under:

Run 7 EEMC Tower Gains - Using Slopes


Goals: Use "inclusive slopes" from fast-detector only, min-bias runs to determine relative (eta-dependent) gains for all EEMC towers for the 2007 run.  More specifically, analyze ~30k events from run 8095104 (thanks to Jan [production] and Jason [fitting] !), fit slopes to all ungated tower spectra, in order to:

  1. check if new / replaced PMT's (only 2 for this year) need significant HV adjustment
  2. make sure tubes with new / replaced bases are working properly
  3. search for towers with unusual spectra, anomalous count rates, or slopes that are far from the norm for that eta bin
  4. compare individual slopes to 2006 absolute gains (from mips) for each eta bin, to test robustness and stability of gain determinations
  5. for tower gains far from ideal (as determined with slopes and/or mips) consider adjusting HV
  6. look for anything else that seems out of whack!

Definitions:

For the gain calibration of towers, we will use

  • x = channel number = ADC - ped
  • E = full e.m. energy (GeV) = (deposited energy / sampling fraction) for e.m. particles
  • G = absolute gain (channels / GeV) including sampling fraction

So:   E = x / G

For slopes, raw spectra (y vs. x) are fit to:   y = A e-bx

Thus, one expects that for a given eta bin:   G  ~  1 / b


Results:

1.  Two new tubes are fine!   Slopes of recently replaced PMT's 04TB12 and 12TE06 are very close to those of neighboring towers at same eta, or those of same subsector in the neighboring sector:

towerID integral slope error
03TB12 2003 -0.04144 0.00181
04TA12 2081 -0.04527 0.00177
04TB12 2022 -0.04400 0.00173
04TC12 2195 -0.03825 0.00170
05TB12 2056 -0.04465 0.00177

towerID integral slope error
11TE06 2595 -0.04157 0.00162
12TD06 1965 -0.05977 0.00185
12TE06 2535 -0.04516 0.00165
01TA06 2124 -0.05230 0.00179
01TE06 2070 -0.05342 0.00190

More global comparisons to all the tower slopes in the same eta bin are given below.   For both tubes, the gain is 5-10% lower than average, but well within useful range.


2.  Change of base (same PMT) has little effect on tower gains.  This has been confirmed for the six bases that were changed (03TA09, 06TB04, 10TE01, 12TA01, 12TC11, 12TE06), using the same comparisons to neighboring towers used in step 1 above.


3.  For all 720 towers, comparison of 2007 slopes to 2006 mip-based absolute gains indicates about 6 problem towers (most "well known")

  • 06TA03 - no useful mip results, fitted slope was positive!  Spectra never make much sense.
  • 08TC05 - didn't work last year, still not working!  Spectra shows only a pedestal.
  • 07TC05 - no gain determined in 2005 or 2006.  Has largest slope of all towers, probably useless.
  • 06TD11 - each year, everything gets replaced; each year it continues to be 'flakey,' sometimes working, sometimes not.
  • 12TD01 - seemed okay last year, now has a very small slope.  Maybe PMT is dying fast?
  • 10TA11 - worked fine last year, recently died.  HV off, only a pedestal.

In addition, 09TE01 seems to be working now, though it failed the mip gain analysis last year, and hasn't been 'fixed.'

All of these cases are easily seen in the following correlation plot:


4.  See clear correlations, within each eta bin, between new (2007) slope analysis, last year's mip analysis    ->   gains are stable, methods are robust!   On vertical scale, solid magenta line = ideal gain for that bin, dashed = +/- 15%

eta bin

correlation plot comments
1 .gif one high gain tube (10TA01), reasonable correlation, no obvious problems
2 .gif looks okay, all within +/- 20% of ideal gains
3 .gif pretty ratty - several towers ~15% off 'correlation' curve
4 .gif one very low gain tube (01TA04), one with very small slope (02TD04), otherwise all okay
5 .gif a couple of high-gain towers, correlation is very good
6 .gif one low gain, a few high-gain, but good correlation.  New PMT 12TE06 looks reasonable
7 .gif overall gains a bit high compare to ideal, no real problems
8 .gif no problems
9 .gif no problems
10 .gif strong correlations, tight clustering in both gain sets
11 .gif odd shape, but okay.  Only problem (lower left corner) is 06TD11
12 .gif everything a bit noisier, gains ~7% high overall.  New PMT 04TB12 fits right in!


5.  Number of 'gain outliers' is quite small, deviation of average from ideal always < 10%.  Because the endcap towers are not used for trigger decisions, no obvious advantage in making HV adjustments to large number of towers.


Conclusion:  Endcap towers are in good shape!  A very small number (~6 / 720) are not working well, but for these few, HV adjustment would not solve the problem.  No strong argument for changing HV on any particular tube at this point.


N.B.   For each eta bin, one can calculate the ratio   R = G / (1/b)   as a 'conversion' of slope data to absolute gains.   Using the 2006 mip calibration and the 2007 slopes, one gets a fairly smooth curve, though something seems to be happening around eta bin 8.