NEWCHOD construction ideas I.M. 28/08/2013

Basic considerations: Ideally the NEWCHOD should be “ thin ” and be contained in a light tight box, occupying a limited longitudinal space. It should be built with plastic scintillator TILES of dimensions, dependent on the position relative to the beam, such that the maximum rate of particle crossing is < 500KHz in any tile with time resolution better than 1ns

The active area should cover a ring between the IRC, radius 140mm and the inner radius 1070mm of the LAV12. To limit interactions, the material seen by the particles should be minimized and this also in the ring (370mm radial width)from 1070 to 1440 occupied by the LAV12 crystals.

It is proposed that the BASIC TILE be of 30mm thickness and be read out by two sets of 9 WLS FIBERS suitably placed, so that any crossing particle would produce with high efficiency a pulse in each of the light sensors linked to the a fiber set. The average time of the two pulses would define the L0 time.

As an example I have chosen the basic TILE with dimensions: Height = 1070/10 = 107mm Width = 1070/4 = 265mm With this choice 37x4 = 148 tiles are needed (with 2 SiPM or PM each, 296 electronic channels) of which 40 of half size

The two sets (let say TOP and BOTTOM) of fibers could be separately brought together with a common length of about 400mm. At this point a SiPM could be attached to the TOP fibers and similarly to the BOTTOM. Otherwise, possibly with a junction of clear fibers), the light could be brought to SiPMs (or PMs) placed at the outside perimeter of NEWCHOD

A single plane of such scintillator tiles would be suitable for defining the ring (140 to 1070mm in radius) crossed by the charged particles for the L0 trigger trigger. In principle the electronic chain could be the same as for MUV3 (and by chance… with the same 296 number of channels with the chosen dimensions)

It is proposed to accurately place the TILES on SINGLE CENTRAL FOIL e.g 2mm thick G10 glass fiber with thin Cu lamination (thus allowing soft soldering), surrounded for 1440>radius>1070mm by a “ strong”, let say 30mm thick plate, with honey comb structure, to minimize interactions.

The rows of tiles should be placed on each side of the foil (e.g. the one seen by the incoming beam) at a vertical distance equal to their width. In this way the WLS fibers equipping any given tile would not significantly interfere with other tiles.

On the other side of the foil (e.g. the one towards the LKr calorimeter) the tiles will be placed vertically in the regions not occupied by tiles on the entry side. Considering the tiles on the two sides of the supporting foil, any particle passing through the annular region of interest will produce a signal in one tile with its two light detectors.

With a threshold at 1/3 Mip Spasimir has calculated the rates in KHz on the NEWCHOD surface, subdivided in either the standard or the small tiles. With the chosen example the maximum total rate (multiple hits in a tile counted as 1) in the standard tiles is 470KHz

For the validation of the NEWCHOD a FULL test of a Standard TILE will have to be carried out with cosmic rays and afterwards with a beam at high rate. The electronic chain up to the output of Constant Fraction Dicriminators should also be tested at the rate of > 500KHz.

Towards agreeing on the NEWCHOD requirements I.M.

Discussing with Augusto it was apparent that we should aim at agreeing, by the forthcoming June Collaboration Meeting, on the requirements of the NEWCHOD in term of: 1maximum loss of triggers by accidentals 2 efficiency and geometrical coverage 3 time resolution 4 electronic chain including hardware trigger

1_ Spasimir’s rates calculation allows to evaluate: SUM(RSIG*REAL)/SUM(RSIG) i.e. the average over the whole active surface of rate of hits per cell (REAL) weighted by the rate of signal (RSIG) in the SAME cell.

Taking e.g. the small cell subdivision and the K  pai nu antinu signal, the AVERAGE REAL is about 140KHZ. The AVERAGE LOSS of signal by trigger accidental vetoing (SAME cell)will hence be: TDEADns* = 1.0% for a reasonable TDEAD=70ns

More serious losses due to accidentals could be due to MULTIPLICITY requirements. The logical OR of the NEWCHOD hits (if reasonably well aligned in time) is expected to have a rate of about 8MHZ. This, for total vetoing time of 10ns, would already mean 8.0% LOSS when requiring no extra hit !

The rate of the OR of the MUV3 hits should be of about 4 MHZ and hence the loss due to MUV3 accidental Vetoing should be 50% of that caused by the multiplicity requirement, assuming the same accuracy in the time alignment of the hits from the NEWCHOD and MUV3 tiles.

The hits due to backsplash, coming late relative to the event, should not cause rejection due to multiplicity requirements. In order to reduce accidental losses it will be necessary to improve the time alignment of hits before defining the multiplicity.

In my view a HARDWARE trigger using uncorrected hits and a linear OR cannot be expected to do better than the 10ns quoted as example. It may nevertheless be useful to have it, in particular while setting up the proper SOFTWARE L0 trigger

At the level of the foreseen (but not yet fully defined) Software TELL62 L0 trigger the time alignment should be actually only slightly worse than the intrinsic time resolution of the signals from the tiles. It may thus be reasonable to aim at <1% accidental losses with 1ns time resolution.

2_ The geometrical coverage for the 140< Radius<1070mm should be easily of order <0.1%: With tiles 100mm high a lack of overlap of 1mm would result in 1% loss, but this can be avoided by actually overlapping a row in front with a row in the back of the middle plane by a small positive amount.

The electronic efficiency of the CFDs, looking at the two light sensors from the same tile, will be determined by the fluctuactions in the number of photo electrons (or of the exited pixels in case of SiPMs. With threshold at 1/3Mip for average N p.e.>16 Efficiency >.99% for both CFDs to be ON (but probably better to require N p.e.>25)

3_ Time resolution of the average of the two CFDs should be of the same order as the time resolution obtained with MUV3, i.e. about 0.5ns This can probably be achieved by using rather thick (but single plane) scintillator giving enough light e.g 30mm --> 30 p.e. or pixels ?

4_ There are at least two practically available choices: a) Use Frascati electronics, which digitizes time (with normal discriminators) and time over threshold of pulses. An on board OR chain allows for a multiplicity based L0 HARDWARE trigger.

b) Use the same MUV3 electronic chain with CFD discriminators. For each pair of light sensors the mean time is provided and with a suitable chain of ORs it can be used for producing a multiplicity based L0 HARDWARE trigger.

Summary of exchange of views on the NEWCHOD during the weekly Meetings May 16th and 23rd a) single plane of scintillator with subdivision of the active surface in TILES such that, following Spasimir suggestion, it would be possible to exclude from the trigger the region (roughly a rectangol displaced towards negative X)where less than a few per mille of K-->pai nu antinu (with the z and p "standard" cuts) have the pai crossing the scintillator plane. b) whether PMs or SiPM are used the time resolution for a tile should be better that 1ns.

c) the losses due to geometrical coverage should be << 1% d) the duration of the pulses reaching the discriminators should not exceed 70ns e) the discriminators should be of the Constant Fraction type f) from each pair of pulses from a tile the Mean Time should be made available towards providing a rough HARDWARE trigger from multiplicity requirements.

g) given the point e) the front end the electronic chain (after necessary amplification and shaping in case of SiPM) could be practically identical to the one existing for MUV3 ( the difference being that for MUV3 the best time information from the pair of pulses looking at one time is the output of their overlap coincidence rather than the mean time)

Since the beginning of July an intense effort has been put by all groups involved in the NEWCHOD i.e Mainz,Perugia,Pisa, Protvino with the collaboration of Frascati and the essential help of CERN in order to build and test a Standard Tile and TEST its RESPONSE.

After a (not so) BRIEF STORY (which could be the subject of an after dinner talk) FINALLY on 22 August a FIRST test has been possible with a TILE produced and machined in Protvino, sent to Cern and equipped with Bicron 1mm diameter fibers. The two bundle of fibers have been then connected to 3x3mm SiPMs by Hamamatsu.

Two typical examples of the response are shown in the following two transparencies. Keep in mind that the single pixel pulse heigth is 15-20mV. Also the preamplifiers were meant for 1x1mm2 SiPM and the output saturates at about 600mV (i.e pixels)

Now a few slides with images of the up to date preliminary design of the proposed NEWCHOD structure