Note: PostScript version of this proposal will be available soon.

A Proposal for E831 Inner Muon First-level Trigger

Yangling Zhang; Dec. 1, 95;

Major Revision Jan. 5, 96;
Modification Jan. 12, 96;

General Discussions
HxV Module
Timing
Inelastic J/psi Acceptance
Dimuon Rates
Comparison with the Sextant Geometry
IM1V & IM2H vs All 3 Planes
Modification History
- Jan. 5, 96
References

General Discussions

The proposal is to build a quadrant geometry from the existing inner muon trigger counters. We will use an HxV logic module from UI together with discriminators and logic fan-in/fan-out units from LeCroy to form a single-muon and a di-muon signal from the inner muon trigger counters. We will be setting up the initial trigger using just IM1V & IM2H. We will have the possibility of including IM1H in the trigger if the trigger time permits and if there is a need to reduce the trigger rate. Most of the discussions assume IM1V & IM2H only.

The aircored cable runs from penetration 1 to the current 3rd cable tray from the top and over the rails of MH2/IM1 and down the eastern wall of the last filter. The length of this aircored cable is 85 +/- 5 ns with the end about 3 ft from the ground. This cable could be made shorter, up to 10 ft shorter, by raising the trigger crate.

The speed of the aircored cable is about 0.90c.

We plan to run a Gore-Tex cable, speed 0.83c, from each dynode to the trigger crate. The Gore-Tex cables should take a direct route from dynodes to the east end of the plane at the height of the trigger crate. If we leave the crate height at 4 ft from the floor, the maximum cable length is 15 ft or 18 ns on Gore-Tex cables. A rear view of these cables and a side view of these cables are provided.

For the ease of connecting and disconnecting the Gore-Tex cables when we want to roll the trigger planes, we plan to use Lemo B240 connectors. Each B240 connector would take 10 coaxial cables. Due to the size of B240, We will have to use Gore-Tex C06C028 cables, which has an outer diameter of 0.09" and attenuation is 17dB per 100 ft at 400 MHz. For a 15-foot Gore-Tex C06C028 cable, the attenuation is 1.53 dB or 92% in terms of the output voltage. We will use 2 Lemo B240 for IM2H and 2 Lemo B240 for IM1V, plus a single normal lemo connector for IM1V as shown in the drawing of the trigger crate.

The trigger crate is a standard NIM crate with 5 LeCroy 623B discriminators, 1 LeCroy 821 discriminator, 4 LeCroy 429A logic units and one UI HxV logic unit. This trigger scheme uses UI HxV logic unit to give signals for single muon and dimuon events.

In the future, we hope to develop a new logic unit just for IM trigger.

IM1V and IM2H channels are combined logically to form the proper inputs for the HxV logic unit. It is noted that some V-view inputs are overlapping while the H-view inputs are not. Because the two V-view signals for each quadran(sp) of the detector are not overlapping, a single muon would not produce a dimuon trigger.

The HxV Module

I previously thought that it was necessary to modify the UI HxV logic module. It turns out that that the existing HxV module has already the output that I wanted. Through this memo, I still refer to the HxV module either as the original HxV or the modified HxV. I refer the Y+Z output as the original HxV and the Y output as modified HxV for lack of better words.

Timing

Using the time particles passing through HxV as t0, the 1st level trigger signals should be in MG 255 +/- 10 ns after t0. Under this scheme, the IM 1st level trigger signal would arrive in MG no later than 232 +/- 11 ns after t0. The individual contributions to the IM 1st level trigger timing is itemized as follows:

IM2H is 27+/-0.5 ft downstream of HxV. Muons would cross IM2H 27+/-0.5 ns after crossing HxV.
It takes about 50+/-10 ns for scintillator light to be collected and amplified by the PMT.
The longest cable from a dynode to trigger crate is 17 +/- 1 ft. It takes 20 +/- 2 ns on a Gore-Tex cable.
Delays in the modules of the trigger crate add up to 29 ns. Given that the inter connecting cables are 4 ns total, we have a 33ns delay in the trigger crate.
The aircored cable from the trigger crate to the counting room is no longer than 85 +/- 5 ns.
The cable from the aircored to the MG module is not longer than 15 ft. It will be 18 ns on Gore-Tex cables.

Most of the uncertainties on the IM 1st level trigger timing comes from the variations in PMT. In principle, we could adjust PMT HVs if some PMTs are too slow. The IM 1st level trigger signal should be able to reach MG module in time.

A detailed listing of the NIM modules used and time delays of different schemes can be found in the following table.

Table 1

Timing and Complexity for Dimuon Triggers
Schemes	# of Input Channels at each Stage						NIM Slots	Time(ns)
HxV w 1V & 2H	1/821 5/623B DISC 41	4/429A OR 41	1/HxV HxV 20	1	NA	NA	12	33.5
Sextant w 1V & 2H	6/623B DISC 41	4/429A OR 41	2/622 AND 12	5/465 AND 6	1/429A OR 15	1	17	54.5
Quad w/o HxV, 1V & 2H	6/623B DISC 41	4/429A OR 41	2/622 AND 8	1/365AL AND 4	1	NA	13	43
HxV w 3 Planes	8/623B DISC 61	6/429A OR 61	3/622 AND 32	1/HxV 20	1	NA	19	46
Sextant 3 Planes	8/623B DISC 61	5/429A OR 61	6/465 AND 18	5/465 AND 6	1/429A OR 15	1	25	54.5
Quad w/o HxV, 3 Planes	8/623B DISC 61	6/429A OR 61	2/365AL AND 12	1/429A AND 4	1	NA	17	40

Notes on Table 1:

1/821 means one LeCroy 821 NIM module. 429A, 365AL and 465 etc are all referring to LeCroy NIM modules.
NIM Slots is the total number of NIM slots needed.
Time(ns) is the total time delay in the trigger crate. It includes the time delay in the modules, as well as 2 ns for the cables interconnecting modules.
The items under # of Input Channels at each Stage is a little hard to explain. It shows the NIM modules used at each stage of the trigger logic, as well as the number of input and output channels. For example, the first row shows that we need one LRS 821 and 5 LRS 623B discriminators. There are 41 input channels and 41 output channels for the discriminators. Four LRS 429A fan-in/fan-out modules follow the discriminators. These 429A modules combine the 41 channels into 20 channels. Finally there is an UI HxV module which gives a dimuon signal from the 20 channels.

Inelastic J/psi Acceptance

A study was done using MC. The color-singlet model was used to generate the j/psi->mu+mu- portion of the event. The jetset routine LU1ENT() was used to generate the gluon jet. E687 SROGUE package was used to simulate the detector responses. Table 1 shows the acceptance for different schemes. It is sort of surprising that the Sextant geometry does not have a higher acceptance than the quadrant geometry. I will have to check this out more carefully.

Taking the numbers as is, the acceptance is 13% if we use all 3 planes, and is 17% if we only use IM1V & IM2H.

Table 2

Acceptance Study From Different Schemes. Listed are the percentage of events that would give IM Dimuon signals per J/psi->mumu event.
IM Planes	E687	Sextant	Org HxV	Mod HxV
1V, 1H & 2H	0.140+/-0.002	0.130+/-0.002	0.135+/-0.002	0.131+/-0.002
1V & 2H	N/A	0.154+/-0.003	0.176+/-0.003	0.172+/-0.003

Dimuon Rates

The absolute muon rates, both single muon and dimuon, have not been carefully studied yet. The hand-waving argument is that because of the added filter the hadron punch through in IM1V in E831 should be equivalent to that of IM2H in E687. The hadron punch through in IM2H in E831 is even smaller. Because there is a 2-feet steel filter in between IM1V and IM2H, the accidental coincidence between there two planes due to single delta-rays would be non-existing.

We did a study using E687 single-muon runs. We checked each event against different trigger schemes. The results are summaried in table 3. The scheme using the quadrant geometry and the "modified" HxV module has 5 times smaller dimuon rates than the scheme using the sextant geometry. The probability of the random noise was found to be about 1.2% per counter per event on this tape, YD7686.

Table 3

Dimuon Rates from E687 Single Muon Runs - YD7686.
1V, 1H & 2H
E687	Sextant	Org.HxV	Mod.HxV
0.0198+/-0.0003	0.0328	0.0456	0.0064

A similar study was done using E687 MG triggers. The results are summaried in table 4. The quadrant scheme has 3 to 4 times smaller dimuon rate than the sextant geometry. The IM dimuon trigger rate will be about 0.00068 +/- 0.00010 per E687 MG trigger. Assuming that the E687 MG was at 70 kHz and that the E831 rate will be 3 times higher, the E831 IM dimuon trigger rate will be about 142 +/- 20 Hz. Because of the added muon filter in E831, the actual rate should be lower.

Table 4

Dimuon Rates from E687 MG Triggers
Tape	Events	MGs	IM1V, IM1H & IM2H				IM1V & IM2H
Tape	Events	MGs	E687	Org HxV	Mod HxV	Sextant	Org HxV	Mod HxV	Sextant
YD7125	913,592	41,350		0.0038+/-3	0.0007+/-1	0.0021+/-2
YD7279	583,943	13,336	0.0014+/-3	0.0031+/-5	0.0006+/-2	0.0022+/-4	N/A

Notes

Dimuon Rates refers to number of simulated dimuon triggers divided by the number of MGs.

For completeness, we also list rate study using MC generated events. Each event contains a single-muon and some random noise of given probability.

Table 5

Dimuon Rates From MC. Listed are the percentage of the events that would give a dimuon trigger. The events have a single-muon through the IM and some random noise hits. The random hits are independent of the single-muons. The probability of having a noise hit was assumed to be either 5% per counter per event, or 1% as indicated in the table.
IM Planes	Noise Rate: 5%				Noise Rate: 1%
IM Planes	E687	Sextant	Org HxV	Mod HxV	E687	Sextant	Org HxV	Mod HxV
1V & 2H	28	19	48	23	3.1+/-0.2	2.2+/-0.2	9.8+/-0.3	3.4+/-0.2
1V, 1H & 2H	N/A	6.4	21	2.7	N/A	0.85+/-0.1	4.1+/-0.2	0.10+/-0.03

Don't know how to estimate the effect of particles passing through edge of the filter.

One might get some estimation of the upper limits on rates from E687 minimum bias data.

Comparison with the Sextant Geometry

To summarize the comparison studies between the quadrant geometry and the sextant geometry proposed by John Cumalat in Thoughts on An Inner Muon Trigger

Timing: My current estimate is that the trigger logic is allowed no more than 55 ns. Implementing the sextant geometry will take 55 to 58 ns using LeCroy modules. While using the HxV module and a quadrant geometry would take 34 ns if we use only IM1V & IM2H, and 46 ns if we use all 3 planes.
Acceptance: The current best estimate (I will double check this) is that the sextant geometry has the same acceptance as the quadrant geometry.
Rate: Sextant scheme has a 3 times to 6 times higher rate than the quadrant scheme, when all 3 IM planes are taken into account. If we only use IM1V & IM2H, the sextant scheme has a lower rate.

IM1V & IM2H vs All 3 Planes

Whether to use just IM1V & IM2H or to use all 3 planes (IM1V, IM1H & IM2H) in the first level trigger is a question of trade-off between acceptance and fake dimuons. Using only IM1V & IM2H, we gain 34% in J/psi acceptance while the fake dimuon rates are 10 to 20 times higher, depending on the noise rate.

It will take 46 ns to form a dimuon trigger using all 3 planes, which is 12 ns longer than using just IM1V & IM2H.

My previous thought was to start with just IM1V & IM2H. We should bring IM1H in the trigger if the rate is too high. Right now it looks very likely that the rate would be too high without IM1H.

References

E.Berger & D.Jones: Inelastic photoproduction of J/psi and Epsilon by gluons Phy.Rev.D23 (1981)1521.

Modification History

Jan. 5, 96 This is a major revision. Too many changes to list.
Jan. 12, 96
- Found a bug in HxV trigger simulation. This would change the fake rates and acceptance and the comparisons between sextant and quadrant geometry.
- A new fake dimuon rate study using E687 muon runs.
- A new fake dimuon rate study using E687 MG triggers.
- Modification of the HxV is no longer needed. The existing HxV has enough outputs to choose from.

A Proposal for E831 Inner Muon First-level Trigger

Contents