From: Brown, RM (Bob) - PPD [R.M.Brown@rl.ac.uk]
Sent: 14
December 2004 20:02
To: Cockerill, DJA (David);
Shepherd-Themistocleous, CH (Claire); Bell, KW (Ken)
Cc: Kennedy, BW
(Bruce)
Subject: MGPA Dynamic range
Dear
all,
Prompted by a brief coffee lounge discussion, I
have taken a first look at the question of dynamic range choice for the
MGPA.
Following the discussion of Chris Seez (FPPA Memo,
August 2002) (thanks to DJAC for pointing me to
this):
The deciding factor, is the desire to keep the
precision physics within the high gain range of the pre-amp, to avoid switching
problems. (This approach was introduced to address FPPA problems, but
applies also to the MGPA.)
From figure 3 of the Seez note, the cut-off energy for
photons in the endcaps from the decay of a 150 GeV Higgs is 300
GeV.
Thus if 300 GeV is taken as full scale for Gain = 12,
then full scale for Gain = 1 is 3.6 TeV.
According to the BWK data for the acceptance rig,
normalised to Test Beam measurements, the mean sensitivity of a VPT + BTCP
crystal is
34 electrons/MeV.
Thus full scale of 3.6 TeV is equivalent to
3.6x3.4x10^7x1.6x10^ -7 pC = 19.6 pC
Note that there is a wide spread in VPT
sensitivity. Most tubes lie within a range of +/-30% about the mean, but
approximately 6% have even lower sensitivity and 10% have even higher
sensitivity. However, this can be largely managed by sorting the tubes and
deploying those with the highest gains at low eta and those with the lowest
gains at high eta.
The BWK sensitivity is the angle-averaged value at
1.8T. This will be almost unchanged at 4T. (Previous discussions
assumed a 10% drop from 1.8 to 4T.)
On the other hand, as others have pointed out, various
losses will rapidly occur:
Irreversible darkening of the crystal and the VPT
face plate, and 'burn-in' of the VPTs.
In addition, the maximum energy deposited in a single
crystal is of order 80%, thus the range-change cut-off should be taken as
240 GeV, rather than 300 GeV.
Furthermore, we plan to operate the tubes at 800V
rather than 1000V, reducing the gain to 90% of the value measured on the
Rig.
Thus even ignoring the 'aging' losses, the full scale
charge should be taken as 0.7x19.6 pC = 13.7
pC.
If an early aging factor of 0.75 is assumed, then full
scale drops to ~10 pC. The current 'base-line' is
16pC.
Following the discussion of Chris Seez, we can assume
that 'full scale' in the ADC corresponds to 3500 channels (allowing for the
pedestal).
If this is set equal to 240 GeV, then the LSB on the
high gain range is equivalent to ~70 MeV.
According to Mark Raymond, the MGPA noise on the x12
gain range should be less than 3500 electrons, for full scale of 16 pC.
Reducing the full scale by ~30% would reduce the noise by
~15%.
Assuming 3000 electrons noise, and allowing for a
degradation of the mean VPT response from 34 to 20 electrons/MeV, gives an
equivalent noise of 150 MeV/channel on the high gain range. For eta ~2,
when expressed as 'Transverse noise', this matches the Barrel performance of 40
MeV/channel.
For an LSB of 70 MeV, the pedestal is a few channels
wide, which is a desirable situation.
In summary, if one assumes the first gain switch at 240
GeV and 20 electrons/MeV, then full scale is 9.2
pC.
Thus 12.6 pC would seem a reasonably safe specification
(corresponding to external components of C(f) = 8.2 pC, R(f) = 4.7k and RF noise
= 2236 electrons).
Comments invited.
Bob.
__________________________
Bob
Brown
Particle Physics
Department
Rutherford Appleton
Laboratory
Chilton, Didcot,
OX110QX
Tel (+44) 1235
446216
Fax (+44) 1235
446733
__________________________