a Xi :hep-ex/0305022 1 13 May 2003
DIRAC: A High Resolu ion Spec ome e o
Pionium De ec ion
B. Ade a p,∗, L. A anasye ℓ, M. Benayoun e, A. Benelli q,
Z. Be ka b, V. B ekho skikh o, G. Ca agheo gheopol m,
T. Cechak b, M. Chiba k, E. Cima , S. Cons an inescu m,
C. De az a, D. D eossi g, D. D ija d a, A. Duda e ℓ,
I. E angelou d, M. Fe o-Luzzi a, M.V. Gallas p,a, J. Ge nd b,
R. Giacomich g, P. Giano i , M. Gia doni , D. Goldin q,
F. G´omez p, A. Go in o, O. Go chako ℓ, C. Gua aldo ,
M. Hans oul a, M. Iliescu ,m, M. Zhabi sky ℓ, V. Ka pukhin ℓ,
J. Kluson b, M. Kobayashi h, P. Kokkas d, V. Koma o ℓ,
V. K uglo ℓ, L. K uglo a ℓ, A. Kuliko ℓ, A. Kup so ℓ,
V. Ku ochkin o, K.-I. Ku oda ℓ, A.Lambe o g, A. Lana o a, ,
V. Lapshin o, R. Lednicky c, P. Le us e e, P. Le isand i ,
A. Lopez Ague a p, V. Luche ini , T. Maki j, N. Man hos d,
I. Manuilo o, L. Mon ane a, J.-L. Na joux e, L. Nemeno a,ℓ,
M. Niki in ℓ, T. N´u˜nez Pa do p, K. Okada i, V. Olche skii ℓ,
D. O ecchini , A. Pazos p, M. Pen ia m, A. Penzo g,
J.-M. Pe eau a, C. Pe ascu ,m, M. Pl´o p, T. Pon a m, D. Pop m,
G.F.Rappazzo g, A. Riazan se o, J.M. Rod iguez p,
A. Rod iguez Fe nandez p, A. Rome o p, V. Rykalin o,
C. San ama ina p,q, J. Sabo ido p, J. Schache , Ch.P. Schue z q,
A. Sido o o, J. Smolik c, M. S einache q, F. Takeu chi i,
A. Ta aso ℓ, L. Tausche q, M.J. Toba p, F. T ian is d,
S. T uso n, V. U kin ℓ, O. V´azquez Doce p, P. V´azquez p,
S. Vlachos q, V. Yazko n, Y. Yoshimu a h, P. Z elo ℓ
aCERN, Gene a, Swi ze land
bCzech Technical Uni e si y, P ague, Czech Republic
cIns i u e o Physics ACSR, P ague, Czech Republic
dIoannina Uni e si y, Ioannina, G eece
eLPNHE des Uni e si es Pa is VI/VII, IN2P3-CNRS, F ance
INFN - Labo a o i Nazionali di F asca i, F asca i, I aly
P ep in submi ed o Nuclea Ins umen s and Me hods A 2 No embe 2018
gINFN-T ies e and T ies e Uni e si y, T ies e, I aly
hKEK, Tsukuba, Japan
iKyo o Sangyou Uni e si y, Japan
jUOEH-Kyushu, Japan
kTokyo Me opoli an Uni e si y, Japan
ℓJINR Dubna, Russia
mNa ional Ins i u e o Physics and Nuclea Enginee ing IFIN-HH Bucha es ,
Romania
nSkobel sin Ins i u e o Nuclea Physics o Moscow S a e Uni e si y Moscow,
Russia
oIHEP P o ino, Russia
pSan iago de Compos ela Uni e si y, Spain
qBasel Uni e si y, Swi ze land
Be n Uni e si y, Swi ze land
Abs ac
The DIRAC spec ome e has been commissioned a CERN wi h he aim o de-
ec ing π+π−a oms p oduced by a 24 GeV/chigh in ensi y p o on beam in hin
oil a ge s. A challenging appa a us is equi ed o cope wi h he high in e ac ion
a es in ol ed, he igge ing o pion pai s wi h e y low ela i e momen um, and
he measu emen o he la e wi h esolu ion a ound 0.6 MeV/c. The gene al cha -
ac e is ics o he appa a us a e explained and each pa is desc ibed in some de ail.
The main ea u es o he igge sys em, da a-acquisi ion, moni o ing and se up
pe o mances a e also gi en.
Key wo ds: DIRAC expe imen , double a m spec ome e , pion sca e ing,
expe imen al echniques, elemen a y a om
∗Depa amen o de F´ısica de Pa ´ıculas, Uni e sidade de San iago de Compos ela,
E-15782 San iago, Spain. Tel.: 34-981-563100, e-mail: ade [email p o ec ed]
2
1 In oduc ion
The DIRAC expe imen aims o measu e he g ound s a e li e ime o π+π−
a oms wi h 10% p ecision, using he 24 GeV/cp o on beam o he CERN
P o on Synch o on. The a om li e ime is a consequence o he s ong in e -
ac ion a low ene gy and i is de e mined by he cha ge exchange ampli ude
π+π−→π0π0 e y close o h eshold. The p obabili y o his p ocess is p o-
po ional o he squa e o he di e ence o S-wa e ππ sca e ing leng hs wi h
iso opic spin 0 and 2, |a0−a2|2. The ela ion be ween he li e ime and |a0−a2|
is model-independen [1]. The pion sca e ing leng hs ha e been calcula ed in
he amewo k o chi al pe u ba ion heo y wi h a p ecision o a ew pe cen :
a0= 0.220 ±0.005 and a2=−0.0444 ±0.0010. Using hese alues one can
p edic he pionium li e ime [2]: (2.9±0.1) ×10−15s. In o de o de e mine
|a0−a2|down o 5%, he li e ime has o be measu ed wi hin 10% accu acy.
Such a measu emen would p o ide a c ucial es o he unde s anding o
chi al symme y b eaking in QCD.
Pionium a oms (A2π) a e p oduced in p o on-nucleus in e ac ions. A e p o-
duc ion hese ela i is ic a oms may ei he decay in o π0π0o ge exci ed o
highe quan um numbe s, o b eak up (be ionised) in he a ge ma e ial whe e
hey a e p oduced. In he case o b eak-up, cha ac e is ic pion pai s (“a omic”
pai s) eme ge. These pai s ha e a low ela i e momen um in hei cen e o
mass sys em (Q < 3 MeV/c), e y small opening angle (θ < 3 m ad) and
nea ly iden ical ene gies in he labo a o y sys em. A high esolu ion magne ic
spec ome e is hen equi ed [3] o spli up he pai s and measu e hei el-
a i e momen um wi h su icien p ecision (0.6 MeV/c) o de ec he pionium
signal supe imposed on he subs an ial backg ound o “ ee” π+π−pai s p o-
duced in inclusi e p o on-nucleus in e ac ions. A p e ious expe imen , using
in e nal p o on beam, has epo ed obse a ion o pionium a oms [4].
The o al numbe o p oduced π+π−a oms is ela ed by an exac exp ession
o he numbe o ee pion pai s wi h low ela i e momen a. Fo a gi en a ge
ma e ial and hickness he a io o obse ed a omic pai s o he o al numbe
o p oduced a oms, i.e. he a om b eakup p obabili y, depends on he li e ime
in a unique way [5].
2 Gene al layou o he expe imen al se up
The DIRAC expe imen al se up [6,7] is loca ed a he T8 p o on beam line
o 24 GeV/cmomen um in he Eas Hall o he PS accele a o a CERN.
The isome ic iew o he se up is shown in Fig.1. The DIRAC appa a us
3
is designed o de ec cha ged pion pai s wi h high esolu ion o e he pai
ela i e momen um. I became ope a ional a he end o 1998 and has been
collec ing da a since he middle o 1999.
Fig. 1. Isome ic iew o he DIRAC se up. The adia ion shielding bounda ies a e
shown on he loo (each di ision ma ked on he bounda y co esponds o 1 me e ).
Fig. 2. Side iew o he DIRAC se up. The seconda y pa icle channel is inclined
by 5.7◦wi h espec o he p ima y p o on beam.
The se up consis s o he p o on beam line, a ge s a ion, seconda y pa icle
acuum channel, spec ome e magne and de ec o s placed ups eam and
downs eam he analysing magne . F ee and a omic π+π−pai s p oduced in
he a ge en e he seconda y pa icle channel which is il ed upwa ds by 5.7◦
wi h espec o he p o on beam (Fig.2). A he end o he seconda y pa icle
channel he spec ome e magne is ins alled, also il ed by 5.7◦ oge he wi h
all he downs eam de ec o s.
The op iew o he se up is shown in Fig.3. The ups eam sec ion o he
seconda y pa icle channel be ween he a ge s a ion and he spec ome e
magne is ins umen ed wi h he ollowing de ec o s: mic os ip gas chambe s
(GEM/MSGC), scin illa ing ib e de ec o (SFD) and scin illa ion ionisa ion
hodoscope (IH).
Downs eam he spec ome e magne he se up spli s in o wo iden ical a ms
o de ec ion and iden i ica ion o posi i e and nega i e cha ged pa icles. The
angle be ween each a m and he spec ome e symme y axis is 19◦. Along
each a m he ollowing de ec o s a e loca ed: d i chambe sys em (DC),
e ical scin illa ion hodoscope (VH), ho izon al scin illa ion hodoscope (HH),
gas Che enko coun e (CH), p eshowe de ec o (PSH) and muon de ec o
(MU).
3 Beam lines, spec ome e magne and adi-
a ion shielding
3.1 P o on beam and a ge s a ion
To ex ac p o ons om he PS o he T8 beam line a slow ejec ion mode is
used. The beam is ex ac ed in spills o ≈400–500 ms du a ion. Du ing da a
aking, be ween 1 o 5 cycles pe PS supe -cycle o 14.4÷19.2 s du a ion a e
4
Fig. 3. Schema ic op iew o he DIRAC spec ome e . Mo ing om he a -
ge s a ion owa ds he magne he e a e 4 planes o mic os ip gas chambe s
(GEM/MSGC), 3 planes o scin illa ing ib e de ec o s (SFD) and 4 planes o ionisa-
ion hodoscope (IH). Downs eam he dipole magne , on each a m o he spec ome-
e , he e a e 4 s a ions o d i chambe s (DC), e ical and ho izon al scin illa ion
hodoscopes (VH, HH), gas Che enko coun e (CH), p eshowe de ec o (PSH) and,
behind he i on abso be , muon de ec o (MU).
deli e ed o DIRAC. The p o on beam in ensi y was se o (0.6−1.0) ·1011
p o ons pe spill, depending on he a ge used.
The PS p o on beam line includes wo bending magne s de lec ing he beam a
an angle o 76 m ad owa ds he inal s aigh T8 sec ion, co ec o magne s
pe o ming ho izon al and e ical s ee ing and quad upole magne s which
ocus he beam on he expe imen a ge . The dimension o he beam spo
a he a ge loca ion a e x= 1.6 mm, y= 3.2 mm a 2σle el. The di e -
gence o he beam is abou 1 m ad. The nominal momen um o he ex ac ed
beam is 24 GeV/c, wi h ins an aneous momen um sp ead close o 0.08% a
2σ. The design o he p o on beam line op ics has been op imised using he
TRANSPORT simula ion code [8] by he CERN PS Di ision.
Downs eam he a ge he p o on beam a els in a acuum channel below
he spec ome e magne and de ec o s and inally is abso bed by a beam
dump.
To measu e he beam in ensi y and o une he beam posi ion on he a ge ,
he beam line is equipped a se e al loca ions wi h seconda y emission cham-
be s and luminescen sc eens wi h TV came as. One addi ional beam posi ion
de ec o (cen oid) [9] is ins alled close o he a ge s a ion.
5

The a ge s a ion houses a emo e con olled de ice wi h 12 holde s o he
a ge s, including an emp y holde and a luminescence sc een. Du ing da a
aking, a ge s made o P (28 µm hick), Ni (94 µm and 98 µm hick) and
Ti (250 µm hick) we e used.
The DIRAC expe imen is sensi i e o pa icles ou side he beam co e (halo),
because he a ge is e y hin (nuclea a ge e iciency is <10−3), and he
ups eam de ec o s a e placed e y close (18–26 cm in he e ical di ec ion)
o he p ima y p o on beam. The halo is o igina ed om sca e ing o p ima y
p o ons on he spli e blades, and a special op ics has been designed o de-
c ease he backg ound halo o a negligible le el. The a io o de ec o coun ing
a es wi h he a ge in place o hose wi h an emp y holde was measu ed o
be ∼25.
3.2 Seconda y pa icle channel and spec om-
e e magne
The seconda y pa icle channel [6] [7] is placed a an angle o 5.7◦ ela i e
o he p o on beam and consis s o wo acuum olumes, as shown in Fig. 3.
The i s one is a 2 m long, 611 mm diame e ube, loca ed immedia ely
downs eam he a ge s a ion, common o bo h he p o on beam line and
he seconda y pa icle channel. Seconda y pa icles exi his ube h ough
a 200 mm diame e window, made o 250 µm hick myla ilm. The second
olume, loca ed a ∼3.5 m om he a ge , consis s o a cylind ical acuum
sec ion, con aining a collima o , a ached o a 2.7 m long la acuum chambe
placed be ween he spec ome e magne poles. A 1.5 m long ai gap be ween
he wo acuum sec ions allows inse ion o he ups eam de ec o s. The angu-
la ape u e o he seconda y pa icle channel is de e mined by he collima o
and is equal o ±1◦in ho izon al and e ical di ec ions esul ing in a solid
angle accep ance o 1.2·10−3s . The la chambe is ended wi h a 0.68 mm
hick Al ou le window o 2.0×0.4 m2dimensions (W×H).
The spec ome e dipole magne (magne ic ield B=1.65 T, ield in eg al BL =
2.2 T·m) has an ape u e o 1.55×0.50 m2(W×H). To educe he s ay ield,
wo magne ic sc eens a e ixed nea i s en ance and exi .
Table 1 summa ises he ma e ial hicknesses (in uni s o adia ion leng h
×10−4) encoun e ed by seconda y pa icles be o e hey each he DC sys em
whe e hei momen a a e measu ed.
6
Table 1
Ma e ial con ibu ions along he seconda y pa icle channel in uni s o adia ion
leng h ×10−4.
Ni- a ge 33.5
Myla window 8.7
4 planes GEM/MSGC 224.1
3 planes SFD 260.0
4 planes IH 153.1
ai gap 34.7
Myla window 8.7
Al-window 76.4
To al 765.7
3.3 Beam dump and adia ion shielding
Neu on and gamma luxes may cause a se ious p oblem o sensi i e elemen s
o he se up. To es ima e hei e ec a simula ion o he backg ound adia ion
lux in he ull expe imen al appa a us has been pe o med [10]. The esul s
we e used o op imise he design o he adia ion shielding which is shown in
Fig. 4.
Fig. 4. The DIRAC se up on he T8 PS ex ac ion line and he adia ion shielding.
A he end o he T8 beam line p ima y p o ons a e abso bed by an i on beam
dump. To dec ease he backg ound gamma and neu on luxes om he beam
dump owa ds he de ec o s, a dedica ed adia ion shielding has been adop ed.
I includes ins alla ion o a g aphi e co e in o he beam dump a ea, a conc e e
wall nea he beam dump and, a 3 m dis ance, ano he i on-conc e e-i on
wall, bo h wi h holes o allow passage o he p o on beam pipe.
The downs eam de ec o s a e shielded in addi ion om backg ound seconda y
pa icles p oduced on he p ima y p o on pipe and su ounding elemen s. Fo
his pu pose a 1 m hick i on wall is ins alled be ween he ups eam de ec o
7
egion and he spec ome e magne . In addi ion, collima o s a e inse ed bo h
in he p ima y p o on beam pipe and in he seconda y pa icle channel. The
p esence o he collima o in he p o on beam line de e mines a educ ion o
he backg ound a e by a ac o o 2.
A adia ion shielding encloses he whole DIRAC expe imen al appa a us o
p o ec he su ounding Eas Hall a ea om i adia ion. I has been designed
acco ding o he maximum lux o 2.7×1010 inciden p o ons pe second,
in espec o he CERN sa e y egula ions. Being he appa a us loca ed in a
ully enclosed a ea, i has become necessa y o p o ide he expe imen al a ea
wi h cooling and en ila ion equipmen o p e en o e hea ing o de ec o s
and elec onics.
4 La ge De ec o s ups eam he magne
Two acking de ices ha e been ins alled in he seconda y pa icle channel
be o e he magne and he collima o : he GEM/MSGC and he SFD. They
a e used o imp o e he esolu ion on he measu emen o he longi udinal and
ans e se componen s o he ela i e momen um o pion pai s as de e mined
by he d i chambe s acking sys em and he nominal posi ion o he beam
a he a ge cen e .
A he same ime hese de ec o s allow o selec pa icle pai s o igina ed by
p ima y in e ac ions a he a ge om he backg ound o seconda y in e -
ac ions and pa icle decays. The ensemble o MSGC/GEM+SFD cons i u es
a acking sys em wi h 7 de ec o planes wi h 2 s e eo angles which p o ides
adequa e space esolu ion o each he limi o mul iple sca e ing in he a ge
ma e ial.
To inc ease he de ec ion capabili y on close-lying acks, an Ionisa ion Ho-
doscope (IH) is ins alled downs eam he SFD, wi h he pu pose o de ec ing
pion pai s wi h a oo small opening angle o be esol ed by he acking de ec-
o s. This is achie ed by a de ailed pulse-high analysis o he double ionisa ion
p oduced by he pa icle pai s in 4 laye s o scin illa ion coun e s.
A gene al pic u e o he abo e-men ioned de ec o s, as hey a e ins alled be-
ween he i s acuum chambe and he seconda y pa icle channel, can be
seen in Figu e 5.
8
Fig. 5. Pho og aphy o he h ee de ec o s ins alled ups eam he magne , be ween
he i s acuum chambe ( igh -hand side) and he seconda y pa icle channel.
F om igh o le , he GEM/MSGC, SFD and IH de ec o s can be ound. The
p ima y p o on beam line can be app ecia ed a he bo om.
5 The GEM/MSGC de ec o
This de ec o pe o ms pa icle acking a a dis ance o 2.4 m om he in e -
ac ion poin . I is a p opo ional gas de ec o , based on he p inciple o he
Gas Elec on Ampli ie (GEM) [11] [12], complemen ed wi h a second ampli-
ica ion and eadou s age p o ided by Mic o S ip Gas Chambe s (MSGC)
[13] [14]. A mo e comple e desc ip ion o he de ec o and i s pe o mance is
being p epa ed in a sepa a e publica ion [15].
I measu es pa icle coo dina es in 4 planes along he di ec ion o he incoming
pa icle: X,Y,U,V, wi h o ien a ions 0, 90, 5, 85 deg ees, espec i ely, whe e
he 0 deg ees a e de ined by mic os ips unning e ically (X-coo dina e). The
s e eo angles allow esolu ion o ghos combina ions o wo o mo e pa icles.
Wi h a single-hi space esolu ion close o 54 µm, his de ec o p o ides a
p ecise measu emen o he pion pai angula opening, ul ima ely limi ed by
mul iple sca e ing in he hin a ge .
5.1 De ec o concep
Each chambe has ac i e a ea 10.24×10.24 cm2, and consis s o a d i elec-
ode, a GEM oil and a MSGC senso . The GEM plane is e enly spaced om
he o he wo wi h a uni o m gap o 3 mm, as indica ed in Fig. 6. The d i
elec ode is made o a Ch omium-coa ed hin glass (200 µm). The GEM is a
50 µm hick kap on oil coppe -cladded on bo h sides wi h a 4 µm hick Cu
laye . The e ching pa e n is cha ac e ised by 50 µm wide holes, 140 µm apa
[16]. Applica ion o a po en ial di e ence o 400 V be ween he wo me al
laye s (V1=−1600 V, V2=−2000 V) p oduces elec on ampli ica ion by a
ac o o 30 [12].
The MSGC senso consis s o 200 µm pi ch al e na ing Ch omium s ips, wi h
9µm and 100 µm anode and ca hode wid h, espec i ely. They a e implan ed
on a ba e DESAG D263 subs a e. Applied ol ages a e: −410 V on ca hodes,
and −3000 V on he d i elec ode, whe eas anode s ips a e se o g ound.
The gas employed is a mix u e o A -DME (60/40). Unde hese conse a i e
condi ions an o e all de ec o gain o app oxima ely 3000 is achie ed.
9
solu ion does no equi e addi ional delay lines o adjus he iming o he
SFD wi h espec o he igge iming.
6.3 Pe o mance in he expe imen
Due o he high lux o pa icles a he posi ion o he de ec o close o he
a ge , and o he p esence o a non-negligible amoun o inclined acks associ-
a ed wi h seconda y in e ac ions in he channel, he SFD pe o mances in he
expe imen sligh ly di e om he abo e men ioned. The de ec ion e iciency
is s ill high (a ound 98%), bu he a e age hi mul iplici y is nea 5 in he
50 ns ime window o TDC (a a nominal beam in ensi y o 1011 p o ons pe
spill impinging on a 94 µm Ni a ge ). The aw ime spec a, ob ained om
e+e−and π+π−e en s, a e shown in Fig. 10 and Fig. 11, espec i ely, o wo
a bi a y SFD channels. The wid h o he dis ibu ions is domina ed by he
ime ji e o he igge signal. A e o -line decon olu ion o he igge ime
ji e he esolu ion o he SFD is ound o be σ=0.8 ns.
7 The Ionisa ion Hodoscope
Cha ged pions o igina ed om pionium b eakup c oss he ups eam de ec o s
a a he small ela i e dis ances. When he dis ance is less han he double
ack esolu ion o he ups eam acking de ices, hen only one hi is de-
ec ed, hus making he e en econs uc ion ambiguous. Tha is why ano he
echnique based on a measu emen o he ionisa ion loss is used as well.
A dedica ed Ionisa ion Hodoscope (IH) [23] has been buil o sepa a e double
ionisa ion signals p oduced by close pion pai s inciden on he same scin illa -
ing slab, om single ionisa ion signals p oduced by one pa icle. In his way,
he unce ain ies esul ing om he ine iciency in de ec ing wo acks wi h
ela i e dis ance app oaching ze o can be signi ican ly educed.
The Ionisa ion Hodoscope desc ibed he e was ins alled in 2001 o eplace a
p e ious e sion o a simila de ec o ype [24], consis ing o only wo planes
wi h 16 slabs o 2 mm hickness, o ien ed in he e ical di ec ion.
The p esen IH de ec o is a scin illa ion hodoscope consis ing o 4 planes o
11×11 cm2sensi i e a ea placed no mally o he axis o he se up (Fig. 12), 3 m
downs eam he a ge . Two planes ha e e ically o ien ed slabs (planes X-A
and X-B) whe eas he o he wo ha e ho izon al slabs (planes Y-A and Y-B).
They a e a anged in he ollowing sequence, mo ing along he beam di ec ion:
16

Fig. 12. Design o he IH scin illa ion plane. 1 – scin illa o s, 2 – ligh -guides, 3 – PM
pho oca hodes.
Fig. 13. Isome ic iew o he Ionisa ion Hodoscope. 1 – scin illa o s, 2 – ligh -guides,
3 – pho omul iplie s wi h shielding.
X-A, Y-A, X-B, Y-B. This o de ing has been chosen o minimise possible c oss-
co ela ions be ween signals in he planes (e.g. due o δ-elec ons). Each plane
is assembled om 16 plas ic scin illa ing slabs made o as scin illa o (BC-
408). Planes wi h he same slab o ien a ion a e shi ed by a hal -slab-wid h
wi h espec o each o he . The slabs a e 11 cm long, 7 mm wide and 1 mm
hick. They a e connec ed o he PM pho oca hodes ia 2 mm hick and 7 mm
wide luci e ligh guides ( ig. 13).
The on and ea su aces o a slab a e co e ed by a millipo e ilm [25] o
e icien ligh collec ion. A he la e al su ace o he slab, ligh is e lec ed by
a hin (30 µm) aluminised black myla ilm, which is used ins ead o millipo e
ilm in o de o minimise he gaps be ween adjacen slabs. A ypical gap
be ween wo adjacen slabs in his con igu a ion is less han 70 µm wide.
Scin illa ion ligh is de ec ed by FEU-85 pho omul iplie s wi h 25 mm diame-
e pho oca hodes. Pho omul iplie s a e assembled by 16 uni s in o a compac
se , allowing independen eplacemen o each PM. Pho oca hodes a e in op-
ical con ac wi h he wide side o a ligh guide ins ead o he adi ional
bu -end eadou . This imp o es he ligh collec ion e iciency by abou 50%.
17
Fig. 14. Typical ADC spec a o single (solid line) and double (dashed line) ioni-
sa ion loss om pa icles c ossing one IH scin illa ing slab.
Losses o double ampli udes
0 0.2 0.4 0.6 0.8 1
Con amina ion o single ampli udes
0
0.05
0.1
0.15
0.2
0.25
Fig. 15. Con amina ion o single ionisa ion ampli udes as a unc ion o losses o
double ionisa ion as ob ained om he analysis o he spec a o Fig. 14.
As he de ec o is highly loaded by in ense pa icle lux, he las 4 PM-dynodes
a e ed by an addi ional powe supply o ensu e a cons an PM ampli ica ion
h oughou he spill.
Signal ampli ude and ime a e digi ised by LeC oy ADC 4300B and TDC 3377
modules, espec i ely. The ime esolu ion o he IH de ec o is be e han
1 ns. The ypical esponse o one IH channel o close pa icle pai s inciden on
one scin illa ing slab and o single pa icle is shown in Fig. 14. I a h eshold is
se o e ain 90% o he double ionisa ion signal om pai s, he con amina ion
om single pa icle ampli udes is less han 15% (Fig. 15).
18
8 D i Chambe s
8.1 Gene al layou and cha ac e is ics
The d i chambe sys em is used o pe o m pa icle acking downs eam
he dipole magne . The sys em is designed o sus ain a high pa icle luency
in he seconda y channel, eaching 10 kHz/cm2a he inne mos egion.
A wo-a m solu ion has been chosen, excep o he i s chambe which is a
single la ge module (DC-1) designed wi h wo sepa a ed sensi i e a eas 0.8×
0.4 m2each. This chambe p o ides 6 successi e measu emen s o he pa icle
ajec o y along he coo dina es X,Y,W,X,Y,W, whe e W is a s e eo angle
wi h inclina ion 11.3owi h espec o he X-coo dina e. DC-1 is ins umen ed
wi h 800 elec onic channels.
Each o he wo a ms consis s o 3 chambe modules, o iden ical design, mea-
su ing coo dina es X,Y (DC-2), X,Y (DC-3) and X,Y,X,Y (DC-4) ollowing
he di ec ion o he ou going pa icle. Thei dimensions a e 0.8×0.4 m2(DC-
2), 1.12×0.4 m2(DC-3), and 1.28×0.4 m2(DC-4). Bo h a ms oge he con ain
1216 elec onic channels.
The dis ance be ween he cen e o he i s hal o DC1 and he cen e o DC4
p o ides a le e -a m o 1.6 m along he a e age pa icle pa h, ha ing uni o m
spacing o chambe s DC-2 and DC-3 along his pa h. Cha ac e is ics o he
d i chambe sys em a e summa ised in Table 3.
8.2 D i chambe elec odes
A schema ic d awing o he sensi i e elemen is shown in Fig. 16. The anode
wi es pi ch is 10 mm, he dis ance L be ween he anode and ca hode planes is
5 mm. The ca hode planes and po en ial wi es a e a equal ol ages. As seen
in he igu e, a sensi i e a ea, co esponding o each anode wi e and limi ed
by he ca hode planes and po en ial wi es, has a squa e (10 ×10 mm2) shape.
In his case, wi h a sui able gas mix u e, i is possible o achie e a linea
beha iou o he d i unc ion, excep in a small egion nea he po en ial
wi e.
Ca hode planes a e made o 20 µm hick ca bon-coa ed myla oils wi h a
su ace esis i i y o abou 400 Ω pe squa e. Such ca hode oils p o ide s able
chambe ope a ion due o a high wo k unc ion o he ca bon coa ing and,
being hin, add only small amoun o ma e ial along he pa icle pa h.
19
Table 3
Gene al p ope ies o he DC modules.
Module Sensi i e Measu ed Numbe o
ype a ea, cm2coo dina e planes
DC-1 40 ×80 X2
le a m Y2
W2
40 ×80 X2
igh a m Y2
W2
DC-2 40 ×80 X1
Y1
DC-3 40 ×112 X1
Y1
DC-4 40 ×128 X2
Y2
Fig. 16. Schema ic iew o he wi e chambe elec odes: AW – anode wi es, P W –
po en ial wi es, C– ca hode oils. Dimensions a e in mm.
Anode and po en ial wi es o 50 µm and 100 µm diame e , espec i ely, a e
made o a coppe -be yllium alloy. The a he la ge diame e o he anode wi es
has been chosen in o de o ope a e he chambe s a high cu en a alanche
ampli ica ion mode.
20
8.3 Chambe design
The chambe design is shown in Fig. 17 o he case o he DC-2 module. The
module is a s ack o aluminium and ib eglass ames, each o 5 mm hick-
ness, ixed by sc ews. The ib eglass ames a e he suppo s o he chambe
elec odes (anode and po en ial wi es and ca hode oils). The wo ou e alu-
minium ames in he s ack a e used o ix he myla window, and he inne
ones a e he space s be ween he ib eglass ames. Rigidi y o he module is
en o ced by aluminium ec angula ubes sc ewed o he su ace o he ame
package.
Fig. 17. Design o he DC-2 module. Uppe igu e: gene al iew. Lowe igu e:
s uc u e o he ame s ack; X – X-plane, Y – Y-plane, C – ca hode oils.
Gas igh ness o he chambe module is p o ided by ubbe o- ings glued
along he inne edges o he aluminium ames. Wi hin a module, gas lows
sequen ially in he sub- olumes de ined be ween ca hode oils, by means o
holes d illed on opposi e sides o he ib eglass ames.
The design o module DC-1 di e s om he one shown in Fig. 17. The main
di e ence, illus a ed in Fig. 18, consis s in he ac ha DC-1 comp ises, in
a single gas olume, wo se s o sensi i e planes, placed symme ically o he
le and igh hand side o he spec ome e axis. The middle zone, which is
s ongly i adia ed by pa icles (mos ly as p o ons om a ge agmen a-
ion), is made insensi i e o he pa icle lux. The limi ing edge o he sensi i e
21

zones, close o he axis, can be a ied. This is possible by means o a s ipped
s uc u e o he neighbou ing ca hodes, which allows s epwise applica ion o
ol age. This design o he DC-1 module ensu es li le amoun o ma e ial, by
a oiding ames in he small angle egion.
Fig. 18. Schema ic iew o he DC-1 module. Do ed a eas show he sensi i e
egions o he X-, Y- and W-planes. Ha ched a eas ma k he zones o he ca hode
s ips which allow o change he wid h o he insensi i e a ea in he cen al egion.
8.4 Chambe ope a ion and pe o mance
The d i chambe s ope a e in a high cu en a alanche mode. This mode is
cha ac e ised by high pulse ampli ude (abou 1 mA), small pulse wid h (20 ns),
and s able ope a ion due o an e iciency pla eau la ge han 1 kV. The single
hi e iciency is abo e 96% when he pa icle lux is abou 10 kHz/cm2. The
employed gas mix u e is A (∼50%) + iC4H10(∼50%) + H2O(0.5%), and he
chambe ope a ion ol age is 3.85 kV.
A space- o- ime ela ionship was ex ac ed om he ime spec um and i s
in eg al dis ibu ion shown in Fig. 19, o a sample o clean e en s wi h a small
amoun o backg ound hi s. The in eg al dis ibu ion has been pa ame e ised
by a second o de polynomial o he ype:
l=a1× ⋆+a2× ⋆2.
In his o mula ⋆= T DC − 0−δ , whe e δ is he signal p opaga ion ime
along he anode wi e.
S udy o he d i unc ion pa ame e s o di e en chambe planes a di e en
beam in ensi ies shows good s abili y o he abo e ela ion. Fo his eason he
22
same d i unc ion pa ame e s (a1and a2) we e used o all chambe planes
du ing he o -line ack econs uc ion p ocedu e.
0
10000
20000
25000
30000
5000
15000
0 50 100 150 200 250 300
D i ime in eg al
50
100
150
200
250
00 50 100 150 200 250 300
D i ime spec um
Fig. 19. Dis ibu ion o he d i ime (uppe ) and i s in eg al spec um (lowe ) o
he X4-plane. Ho izon al scale is in TDC channels, bin wid h is 0.5ns.
Coo dina e esolu ion o he DC sys em is illus a ed in Fig. 20, whe e he
dis ibu ion o di e ences be ween he p edic ed posi ion and measu ed co-
o dina es in one o he planes is shown (X4-plane igh a m). The measu ed
s anda d de ia ion, σ= 100 µm, is de ined no only by he in insic chambe
plane esolu ion, bu also by he accu acy o he p edic ed ack coo dina es.
Taking he la e in o accoun he measu ed in insic space esolu ion o one
plane is be e han 90 µm.
T acking e iciency o he d i chambe sys em as a whole is abou 99%, due
o he ac ha he eques ed numbe o hi s pe econs uc ed ack is less
han he o al numbe o sensi i e planes c ossed by a pa icle.
8.5 Readou elec onics
The eadou elec onics o he d i chambe s, which is a cus om-made sys em
[26], p o ides da a eadou in o he da a collec ion memo ies and inpu o he
igge p ocesso (see sec ion “T igge sys em”).
The sensi i e wi e signals a e digi ised in he 16-channel mul i-hi ime- o-
digi al con e e boa ds (TDC), which a e plugged in he connec o s moun ed
on o he chambe ames. This solu ion esul s in educed numbe o elec onic
23
−1.5 −0.5 0.5−2.0 −1.0 0 1.0 1.5 2.0
0
400
1400
σ=0.1 mm
200
600
800
1000
1200
Del a X, mm
Fig. 20. Dis ibu ion o di e ences be ween he measu ed and p edic ed
X-coo dina e o one d i chambe plane (X4).
uni s, small numbe o cables and high noise immuni y. The de ec ion h esh-
old in he TDC boa d can a y om 0.05 o 2 mA, he maximum numbe o
hi s pe channel is 16. Leas coun o TDC is 0.5 ns, which co esponds o a
d i dis ance o 25 µm a a cons an d i eloci y o 50 µm/ns, well below
he chambe in insic space esolu ion.
A comple e eadou chain consis s o TDC boa ds, bus d i e s, eadou con-
olle and VME memo y. Up o 8 TDC boa ds can be connec ed o he bus
d i e ia common da a and con ol buses, o ming a segmen o he DC ead-
ou sys em. Simila ly, up o 8 segmen s a e connec ed o a eadou con olle
o ming a eadou b anch. The da a o an e en a e s o ed in local da a bu e s
un il he highe le el igge decision is issued. I he e en is accep ed, he
da a a e se ially ans e ed ia he eadou con olle o he VME memo ies.
To ead ou all he DC da a 3 eadou b anches a e used. The accep ed da a
a e ans e ed o he VME bu e s wi hin 5 µs on a e age. The DC eadou
is as enough compa ed o he global eadou ime o he expe imen .
Fo igge pu poses he DC eadou sys em is equipped wi h as da a po s
which di ec ly ansmi he hi wi e numbe s o a igge ack p ocesso . This
p o ides a minimum access ime o he da a, hus educing he la ency o he
igge sys em. T ansmission o he ack p ocesso is pe o med in pa allel.
24
9 The Ve ical Hodoscopes (Time-o -Fligh De-
ec o )
The e ical hodoscope (VH) consis s o an a ay o e ical scin illa ing slabs
placed downs eam he DC sys em. The VH sys em, oge he wi h he ho i-
zon al hodoscope, p o ides as coincidence signals be ween he spec ome e
a ms necessa y o he i s le el igge . I is also used, in co ela ion wi h
o he de ec o s, in he de ini ion o dedica ed igge s o calib a ion pu poses
and o a highe le el igge o he selec ion o low Qe en s (see igge
sec ion la e ).
Fig. 21. Time di e ence be ween cha ged posi i e and nega i e pa icles ob ained
om s anda d had on igge da a. The cen al peak has a gaussian wid h o 193 ps,
and he shaded a ea ep esen s a 2σcu used o selec p omp ππ e en s. The la
backg ound is o igina ed om acciden al pai s, no belonging o he same beam
in e ac ion. No e he shoulde on he igh -hand side o he peak, due o π−pp omp
pai s.
A key unc ion o his de ec o , which mo i a ed a special design, is o p o ide
a e y accu a e ime de ini ion o pion pai s o igina ed om he same p o on
in e ac ion (p omp pai s), in o de o pe o m a clean sepa a ion (in o -line
analysis) wi h espec o pai s in which he pions a e p oduced a di e en
imes (acciden al pai s).
25
men um egion) a e 10 mm hick, whe eas he es a e 25 mm hick (a ound
2 and 5 uni s o adia ion leng h, espec i ely). The scin illa o used is BI-
CRON ype BC-408, wi h slab dimensions 35 ×75 cm2and 1 cm hickness.
The scin illa ion ligh is ansmi ed o pho omul iplie s EMI 9954-B, placed
a one end only, by 10 mm hick Plexiglas ligh -guides ending wi h Plexiglas
cylinde s o ma ch wi h he PM pho oca hodes. Since he maximum pa icle
lux on each PSH elemen is as high as 2 MHz, an addi ional boos e powe
supply is used o eed he las PM dynodes.
The de ec o signals a e linea ly spli in o wo b anches, one used o igge
pu poses and ano he o ADC analysis. In he o me , a leading edge LeC oy
4416 disc imina o is used wi h a h eshold co esponding o e icien de ec ion
o minimum ionising pa icles. In he la e he signals a e ed in o 20 dB
a enua o s in o de o i he elec on signals in o he dynamical ange o he
LeC oy 4300B ADC.
The single a m de ec o e iciency is 99.5% o pions. Fig. 30 shows he pulse-
heigh spec a om one elemen o he PSH o pions and elec ons selec ed by
he igge sys em ( his selec ion is based on he Che enko de ec o esponse).
Fig. 30. Pulse-heigh spec a o pions and elec ons in one elemen o PSH.
As seen om Fig. 30, he pion spec um has a ail (o igina ing om nuclea
in e ac ion o pions in he con e e ) ex ending o he elec on ampli ude
egion. The o -line s udy o he e/π sepa a ion e iciency o he PSH showed
ha ejec ion o elec ons eaches 85% wi h less han 5% losses o pions. As
only a e y small ac ion o elec ons escapes he on-line iden i ica ion in he
Che enko coun e s, he combined use o he Che enko and PSH de ec o s
p o ides almos 100% elec on ejec ion powe a he o -line analysis s age.
32

13 Muon de ec o
Admix u e o muons in he ππ e en s can be a se ious sou ce o backg ound.
Fo his eason a muon de ec ion sys em is implemen ed o p o ide e icien
muon agging. Muons come almos en i ely om pion decays wi h a small
admix u e om o he decays and di ec µ+µ−pai p oduc ion.
The muon de ec o consis s o scin illa ion coun e s placed behind a hick i on
abso be which almos en i ely abso bs had ons and ela ed had onic showe s.
This de ec o is placed a he downs eam end o he DIRAC appa a us, ew
me e s om he in ense p ima y p o on beam dump. As a esul , he muon
scin illa ion coun e s may unde go a high lux o backg ound adia ion om
he beam dump a ea. This has equi ed a special design o he coun e a ays
and elec onics and has p e en ed om using muon in o ma ion du ing on-line
da a educ ion.
The coun e s a e loca ed behind i on abso be blocks wi h hickness anging
om 60 o 140 cm (see Fig. 25). The hickness is la ge in he egion close
o he spec ome e symme y axis, in o de o compensa e o he ha de
pion momen um spec um. A double laye s uc u e has been en isaged o
he coun e s, each laye on each a m consis ing o 28 coun e s wi h equal
scin illa ing slabs o 75 ×12 cm2 on a ea and 0.5 cm hickness. The muon
de ec o da a a e ead ou only i simul aneous signals om a pai o co e-
sponding coun e s in he wo laye s a e de ec ed. This essen ially educes he
backg ound coun ing a e induced by he neu on lux om he beam dump.
PM
PM
CFD
CFD
MT o TDC
µ
Fig. 31. Readou scheme o he muon de ec o elemen .
33
Fig. 32. Schema ic layou o muon coun e s on hei suppo s uc u e, indica ing
ligh guides and pho omul iplie s.
A special eadou a chi ec u e is ealised o comp omise be ween a cos -sa ing
solu ion and he need o achie e a easonably high ime esolu ion. Scin illa-
ion ligh is de ec ed by 25 mm diame e bialkaline FEU-85 pho omul iplie s
placed a one scin illa o end, in he wo laye s a opposi e ends, as shown in
Fig. 31. Signals om a pai o coun e s a e ed in o cons an ac ion disc im-
ina o s (CFD) ollowed by mean ime (MT). CAEN modules C808 and C561
a e used, espec i ely, o his pu pose. In such a scheme he ou pu signal is
gene a ed only i bo h coun e s a e hi , and co ec iming occu s only i he
same pa icle c osses bo h coun e s.
Fish- ail ligh guides a e used o couple he PM pho oca hodes o he scin il-
la o s, excep when his is impeded by he p esence o he conc e e loo (see
Fig. 32).
In his case ( o 20 coun e s o he second laye in each a m), pho omul ipli-
e s a e di ec ly coupled o he scin illa ing slabs, which a e hen made wice
hicke o compensa e o he loss o ligh yield.
In Fig. 33 he measu ed ime di e ence be ween he signals o he muon de-
ec o and he e ical hodoscope is shown o an indi idual muon elemen o
he posi i e a m. A global ime esolu ion o 1.3 ns is obse ed, wi h a e y
small backg ound le el.
The way o supp ess muons a he igge le el would be o include he muon
coun e signals in o he an i-coincidence igge logic. Because o he high
backg ound load o he muon de ec o , one migh dange ously supp ess use ul
pion e en s i hey happen o occu on- ime wi h backg ound signals in he
muon de ec o . Tha is why he op ion o use o -line he muon de ec o in-
o ma ion has been chosen. In he o -line analysis only he e en s wi h muon
34
Muon ime, ns
σ=1.3 ns
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
-20 -15 -10 -5 0 5 10 15 20
Fig. 33. Time di e ence be ween he signals o he muon de ec o and he e ical
hodoscope.
coun e hi s co ela ed in ime wi h hose o o he de ec o s a e agged as
“muon” e en s and hus ejec ed [30]. F om he analysis o expe imen al da a
we ha e in e ed ha he ac ion o such e en s, con aining a leas one muon,
is abou 10% [31]. Such muon-e en s o igina e o a la ge ex en (∼80%) om
π±-decays in he pa h be ween he DC and he muon coun e sys ems (de-
cays ups eam he DC a e mos ly supp essed by he igge sys em and hus
con ibu e less o he collec ed e en sample).
14 T igge sys em
The igge sys em was designed o p o ide a educ ion o he e en a e
o a le el accep able o he da a acquisi ion sys em which is a ound 2000
e en s/spill. Pion pai s a e p oduced in he a ge mainly in a ee s a e wi h
a wide dis ibu ion o e hei ela i e momen um Q, whe eas a omic pai s
om A2πdisin eg a ion ha e e y low Q, ypically below 3 MeV/c. The on-
line da a selec ion ejec s e en s wi h pion pai s ha ing QL>30 MeV/co
Qx>3 MeV/co Qy>10 MeV/c, keeping a he same ime high e iciency
o de ec ion o pai s wi h Qcomponen s below hese alues, (QL, Qxand
Qya e longi udinal and ans e sal componen s o he ela i e momen um,
espec i ely.)
35
A mul ile el igge is used in DIRAC [32]. I comp ises a simple and as i s
le el igge and highe le el igge p ocesso s which apply selec ion c i e ia
o di e en componen s o he ela i e momen um o pion pai s.
Due o he equi emen s o he da a analysis p ocedu e, he on-line selec ion
o only ime co ela ed (p omp ) pion pai s, o igina ing om a single p o on-
a ge in e ac ion and de ec ed simul aneously by bo h spec ome e a ms,
is no enough. In addi ion, a la ge numbe o unco ela ed, acciden al, pion
pai s is also necessa y. These acciden al pai s a e used in he o -line analy-
sis o desc ibe he ela i e momen um dis ibu ion o ee (non-a omic) pion
pai s wi hou Coulomb in e ac ion in he inal s a e. The e o e, he igge
sys em should apply e y simila selec ion c i e ia o p omp and acciden-
al e en s, wi hin a p eselec ed coincidence ime window cen ed a ound he
peak o p omp e en s. The s a is ical e o o he A2πli e ime measu emen
depends on he numbe o bo h p omp and acciden al de ec ed pai s. In
s anda d expe imen al condi ions, he op imal a io o p omp o acciden al
e en s is ob ained using a 40 ns wide coincidence ime window be ween he
imes measu ed in he le (VH1) and igh (VH2) e ical hodoscopes.
Since 1999, when he expe imen has s a ed, he igge a chi ec u e was
upg aded se e al imes o achie e a la ge educ ion o he backg ound e en
a e (p omp and acciden al pai s wi h la ge alues o Q). In he p esen a icle
we b ie ly desc ibe he mos ecen e sion 2.
A block diag am o he igge a chi ec u e is p esen ed in Fig. 34. The i s
le el igge T1 s a s digi isa ion o he de ec o signals in he da a acquisi-
ion (DAQ) modules (ADC, TDC, e c.). A he nex le el he neu al ne wo k
igge DNA/RNA (DIRAC Neu al A omic and Re ised Neu al A omic ig-
ge ) ejec s he e en s wi h high Q alues. A he las s age, a powe ul d i
chambe igge p ocesso T4 imposes addi ional cons ain s o he ela i e
momen um and akes he inal decision o accep o o ejec he e en .
In addi ion o he main igge designed o de ec pionic a oms, se e al cali-
b a ion igge s a e un in pa allel. They a e applied o DAQ di ec ly, wi hou
supp ession by highe le el igge s ages bu wi h app op ia e p escaling ac-
o s.
2Be o e 2001 he igge sys em included T2 and T3 s ages ollowing he i s le el
igge (T1). T2 selec ed pa icle pai s wi h a small ∆xdis ance in he ups eam
spec ome e egion ( o ejec ion o high Qx) using he da a om SFD and IH. T3
analysed hi pa e ns o he ups eam IH and downs eam VH de ec o s imposing
selec ion c i e ia o QL alues. A de ailed desc ip ion o T2 and T3 is gi en in [32]
and [33]. Wi h he implemen a ion o a new a chi ec u e o he igge sys em hese
igge s ages we e emo ed.
36
s a FERA and DC
DNA
RNA
clea FERA and DC
s a MSGC
T4 eadou all
T0
T1
clea all
and deno e posi i e and nega i e decisions o he co esponding
igge le el
+
+
+
+
Fig. 34. Gene al block diag am o he DIRAC mul ile el igge .
14.1 Fi s le el igge (T1)
The i s le el igge (a de ailed desc ip ion is gi en in [34]) ul ils he ollowing
asks:
— Selec s e en s wi h signals in bo h de ec o a ms downs eam he magne .
— Classi ies he pa icle in each a m as π±o e±depending on he p esence
o he Che enko coun e signal. P o ons, kaons and muons a e equally
included in he “pion” class, hei iden i ica ion is pe o med in he o -line
analysis.
— A anges he coincidences be ween he signals de ec ed in he wo a ms.
The wid h o he coincidence ime window de ines he a io be ween yields
o p omp and acciden al e en s in he collec ed da a.
— Applies a coplana i y c i e ion o pa icle pai s: he di e ence be ween he
hi slab numbe s in he ho izon al hodoscopes in he wo a ms (HH1 and
HH2) should be ≤2. This c i e ion o ces a selec ion on he Qycomponen
o he ela i e momen um and p o ides a a e educ ion by a ac o o 2.
— Selec s in pa allel e en s om se e al physics p ocesses needed o he se up
calib a ion: e+e−pai s, Λ →p+π−decays, K±decays o h ee cha ged
pions.
The physics and calib a ion igge signals pass h ough he mask egis e
and, a e p ope p escaling, a e combined wi h an OR unc ion. Any igge
ype can be enabled o disabled by p ope p og amming o he mask egis e .
Independen p escaling o each sub- igge channel allows o adjus hei el-
a i e a e wi h espec o he a e o he main igge . A speci ic igge ma k
is eco ded o e e y e en o allow so ing he da a by igge ype du ing
o -line analysis and on-line moni o ing.
All T1 modules a e ECL line p og ammable mul ichannel CAMAC uni s. Mos
o hem a e comme cial modules, excep o he dedica ed coplana i y p oces-
so which has been cus om-de eloped a JINR. Mean ime uni s a e used in
37

all VH and HH channels o emo e he dependence o he ime measu emen
on he hi loca ion, hus educing he o al igge ime ji e .
14.2 Neu al ne wo k igge (DNA/RNA)
The DNA/RNA igge [35] is a p ocessing sys em using a neu al ne wo k
algo i hm. I s ha dwa e is based on he cus om-buil e sion o he neu al
igge used in he CPLEAR expe imen [36].
DNA/RNA ecei es (see Fig. 35) he hi pa e ns om he e ical hodoscopes
VH1, VH2 and he X-planes o he ups eam de ec o s: he ionisa ion ho-
doscope (IH) and he scin illa ing ib e de ec o (SFD). Fo low Qe en s he
hi s in hese de ec o s a e co ela ed.
Fig. 35. DIRAC de ec o s used o he neu al ne wo k igge s DNA and RNA.
Numbe s o signal channels om each de ec o a e gi en in pa en heses.
38
The neu al ne wo k was ained o selec pa icle pai s wi h low ela i e mo-
men a: Qx<3 MeV/c,Qy<10 MeV/cand QL<30 MeV/c. The e en s
which do no sa is y any o hose condi ions a e conside ed “bad” and e-
jec ed.
The DNA/RNA logic is s a ed by a as p e igge , T0, and in 250 ns e alua es
an e en . Ra e educ ion by a ac o o 2 wi h espec o T1 is achie ed wi h
DNA/RNA.
14.3 D i chambe p ocesso (T4)
T4 is he inal igge s age. T4 p ocesso econs uc s s aigh acks in he
X-p ojec ion o he d i chambe s and analyses hem o de e mine he alue
o he ela i e momen um ( he algo i hm is desc ibed in [32]).
magne
X2X1 X3
hi wi e numbe s
hi wi e numbe s inde 2
T ack
T ack
analyse
X4
X6
X5
inde 1
T ack
A m 1
A m 2
ack iden i ie s ack iden i ie s
Fig. 36. T4 ope a ion block diag am. Only he d i chambe X-planes in ol ed in
T4 a e shown.
The block diag am o he T4 ope a ion is shown in Fig. 36. The d i chambe
p ocesso includes wo s ages: he ack inde and he ack analyse . The
ack inde (an iden ical p ocesso is used o each a m) ecei es he numbe s
o he hi wi es om all d i chambe X-planes. D i ime alues a e no
used in he T4 logic. A unique numbe , “ ack iden i ie ”, which con ains he
encoded numbe s o he hi wi es, is associa ed o he ound ack.
39
I acks a e ound in bo h a ms, he ack analyse con inues he e en e al-
ua ion. The ack analyse ecei es he ack iden i ie s om bo h a ms and
compa es hem wi h he con en o a look-up memo y able which con ains all
possible combina ions o ack iden i ie s o pion pai s wi h QL<30 MeV/c
and Qx<3 MeV/c. These “allowed” combina ions a e ob ained om a ded-
ica ed simula ion using he p ecise geome y o he se up. I a ele an com-
bina ion is ound, he T4 p ocesso gene a es a posi i e decision signal which
s a s he da a ans e o he VME bu e memo ies. O he wise, he Clea
and Rese signals a e applied o he DAQ and igge sys ems.
The T4 decision ime depends on he complexi y o he e en and is a ound
3.5 µs on a e age. The ejec ion ac o o T4 is a ound 5 wi h espec o he
T1 a e and a ound 2.5 wi h espec o DNA/RNA.
The whole igge sys em is ully compu e con olled: no ha dwa e in e en-
ion is needed in o de o modi y he igge con igu a ion. Wi h all selec ion
s ages enabled he e en a e a he ypical expe imen al condi ions is a ound
700 pe spill, ha is well below he limi s o he DAQ a e capabili y.
15 Da a acquisi ion sys em
The a chi ec u e o ha dwa e and so wa e pa s o he da a acquisi ion sys em
akes in o accoun he ime s uc u e o he p o on beam. The machine supe -
cycle o he CERN PS accele a o has 15–20 s du a ion. Wi hin his pe iod
DIRAC ecei es p o ons in spills o 400–450 ms wid h, om 1 o 5 spills pe
supe -cycle. The in e al be ween consecu i e spills can be as sho as 1 s.
Du ing he accele a o bu s he da a om all de ec o s a e ead ou in o
VME bu e memo ies (comme cial and dedica ed elec onic modules) wi h-
ou any so wa e in e en ion. The da a ans e o VME p ocesso boa ds,
e en building, da a ans e o he main hos compu e and o he ela i ely
slow ope a ions a e pe o med du ing he pause be ween bu s s. This p o ides
he maximum ope a ion a e o DAQ. The in o ma ion comes om 2048 chan-
nels o mic os ip gas chambe s, 800 channels o scin illa ing ib e de ec o ,
2016 channels o d i chambe s and 224 channels o o he scin illa ion and
Che enko de ec o s. Fo e e y channel he ime o ampli ude in o ma ion is
eco ded o bo h. In addi ion o a main eadou mode, he eadou o scale s
a he end o e e y spill is a anged ia a CAMAC bus. The coun ing a es
o all he de ec o s, igge a es a di e en igge le els and o di e en
sub- igge modes a e eco ded oge he wi h he alue o he beam in ensi y
p o ided by he PS complex.
40
15.1 DAQ ha dwa e
The da a eadou [37] is a anged wi h 12 eadou b anches: 4 b anches o
GEM/MSGC, 3 b anches o DC and 5 FERA [38] b anches o all o he
de ec o s. In FERA and DC b anches he VME modules CES HSM 1870 and
LeC oy 1190 a e used as bu e memo ies. In GEM/MSGC b anches he bu e
memo ies a e inco po a ed in o dedica ed VME modules [39]. FERA b anches
include di e en FERA compa ible LeC oy modules: ADC 4300B, mul i-hi
TDC 3377, uni e sal logic modules 2366 con igu ed in his applica ion like
FERA egis e s and scale s. Peculia i ies o FERA eadou in DIRAC, such
as mul i-ga e and Fas Clea ope a ion, a e desc ibed in [40].
The logic o eadou is he ollowing. The i s le el igge T1 s a s digi i-
sa ion in ADC and TDC o FERA and DC b anches (see Fig. 34 in Sec ion
“T igge sys em”). Readou is inhibi ed unless a posi i e decision o he high-
es le el igge T4 is ecei ed. I he T4 p ocesso decides posi i ely, he
Inhibi Readou s a us is eleased and he con e ed e en da a a e ans-
e ed o bu e memo ies. I he decision o DNA/RNA o T4 is nega i e,
hen a Fas Clea signal is gene a ed [40] which disca ds he da a in all FERA
modules and DC egis e s.
In con as o FERA and DC subsys ems, he p ocessing o GEM/MSGC da a
is s a ed by he nex le el DNA/RNA igge . This educes he dead ime in-
oduced by he Clea p ocess which in GEM/MSGC elec onics akes longe .
Thus, he nega i e decisions o T4 only lead o clea ing o he GEM/MSGC
da a. I no Clea signal is ecei ed, he con e ed da a a e ans e ed o bu e
memo ies.
The eadou o he whole e en akes 45 µs and is de ined by a ixed acquisi ion
ime o GEM/MSGC which exceeds he eadou ime in o he b anches.
15.2 DAQ so wa e
The main pa o he DAQ so wa e [41] is unning on wo VME p ocesso
boa ds and on he main DAQ hos . Powe PC-based VME p ocesso boa ds
con ol he igge and FERA eadou elec onics ia wo CAMAC b anch
d i e s, ope a e VME modules, ead da a om bu e memo ies and ans e
hem o he main DAQ hos . The main DAQ hos pe o ms e en building,
eco ds he buil da a and dis ibu es hem o o he compu e s o on-line
moni o ing and analysis.
41
[8] K. B own, D. Ca ey, C. Iselin, F. Ro hacke . “TRANSPORT: a compu e
p og am o designing cha ged-pa icle beam anspo sys ems”, CERN-80-04,
Gene a, CERN, 1980.
[9] J. Bosse , G. Molina i, J-M. Nongla on, V. P ie o, R. S ee enbe g, “Tes o a
dispe sion sweep co ec ion sys em using a cen oid in he DIRAC beam line”,
CERN/PS 2002-041 (OP), Gene a, CERN, 2002.
[10] I.A. Ku ochkin, “P elimina y calcula ions o neu on backg ound in he DIRAC
expe imen ”, DIRAC no e 96-7, CERN, 1996.
[11] F. Sauli, NIM A386 (1997) 531.
[12] “De elopmen o a Mic o S ip De ec o o he DIRAC Expe imen ”, Te esa
N´u˜nez, Doc o al Thesis, Uni e si y o San iago de Compos ela, 1999.
[13] A. Oed, NIM A263 (1988) 351.
[14] F. G´omez e al., Nucl. Ins . and Me h. A384 (1997) 351.
[15] B. Ade a e al., “Pe o mance o he GEM/MSGC de ec o in he DIRAC
Expe imen ”, Nucl. Ins . and Me h. A, in p epa a ion.
[16] Manu ac u ing echnology by he CERN EST DEM g oup.
[17] “De elopmen and Cons uc ion o he Readou Elec onics o Mic o S ip
De ec o s”, Pablo V´azquez, Doc o al Thesis, Uni e si y o San iago de
Compos ela, 2000.
[18] R. Ho isbe ge and D. Pi zl, Nucl. Ins . and Me h. A326 (1993) 92.
[19] V. Ago i sas e al., Nucl. Ins . and Me h., A411 (1998) 17-30.
[20] V. Ago i sas e al., “Ul a as eadou o scin illa ing ib es using upg aded
posi ion-sensi i e pho omul iplie s” , DRDC P oposal, DRDC/P-25, CERN,
1991.
[21] M. Fe o-Luzzi e al., No e Dame 1997, “Scin illa ing ib e de ec o s”, AIP
Con .P oc. Vol.450, p.278–285, New Yo k, USA.
[22] A. Go in e al., Nucl. Ins . and Me h., A452 (2000) 280-288.
[23] V. B ekho skikh e al., “New Ionisa ion Hodoscope: design and
cha ac e is ics”, DIRAC no e 2002-09, CERN, 2002.
[24] C. De az, D. D ija d, M. Fe o-luzzi and V. Koma o , “The ionisa ion
hodoscope: pe o mance and cha ac e is ics o he i s module”, DIRAC
In e nal No e 1997-09.
[25] Millipo e pape ype HAWP po e size 0.45 made by MILLIPORE Co p.,
Bed o d, Ma 01730 USA.
[26] L. A anasye , V. Ka pukhin, NIM A492 (2002) 351.
[27] B. Ade a e al., Nucl. Ins . and Me h. A491 (2002) 41.
48

[28] M. B agadi eanu e al., NIM 426 (1999) 254.
[29] M. Pen ia, Gh. Ca agheo gheopol, M. Ciobanu, D. Pop and C. Rusu,
“P eshowe de ec o commissioning along wi h DIRAC se -up”, DIRAC no e
99-03.
[30] V. B ekho skikh, M.V. Gallas, “Muon iden i ica ion in DIRAC expe imen ”,
DIRAC No e 2001-02.
[31] A. Lana o, “Fea u es o µ-candida e e en s”, DIRAC No e 2002-02.
[32] L. A anasye e al., NIM A491 (2002) 376.
[33] M. Gallas, NIM A482 (2002) 222.
[34] L. A anasye , M. Gallas, V. Ka pukhin and A. Kuliko , NIM A479 (2002) 407.
[35] P. Kokkas, M. S einache , L. Tausche and S. Vlachos, NIM A471 (2001) 358.
[36] F.R. Leimg ube e al., NIM A365 (1995) 198.
[37] V. Ka pukhin, A. Kuliko , V. Olshe sky, S. T uso . “Readou logic and i s
ha dwa e implemen a ion in he DIRAC expe imen .” P ep in JINR E10-2003-
2, Dubna, 2003. Submi ed o NIM.
[38] FERA – Fas Encoding and Readou ADC Sys em Possibili ies. Applica ion
No e AN-4004A, LeC oy Co po a ion.
[39] F. Gomez, P. Vazquez, “MSGC/GEM De ec o Elec onics”, DIRAC In e nal
No e 00-01, CERN, 2000.
[40] V. Ka pukhin, A. Kuliko , “Mul iga e and Fas Clea Logic in FERA Readou ”,
P ep in JINR E10-2001-2, Dubna, 2001.
[41] V.G. Olshe sky, S.V. T uso , NIM A469 (2001) 216.
[42] G.D. Badhwa , S.A. S ephens, R.L. Golden Phys Re . D15, 820 (1977).
[43] A. Lana o, “Pa ame e isa ion o co ela ed and acciden al pa icle spec a”,
DIRAC In e nal No e 01-01, CERN, 2001.
49