1. Design of the He and HB ODU
M. McKenna, M. Wayne
Notre Dame

2. Schedule Highlights The ODU project is done in two stages
Preproduction of up to 32 ODUs (16 HB and 16 HE)
Mechanical unit design for HB/HE completed by 26-June-2014
Assembly of HE begins 28-July-2014, assembly of HB begins 25-Aug-2014
HE (HB) units at CERN and ready for mounting in RMs by 15-Feb-2015 (15-Mar-2015)
Full production of 360 ODUs (180 HB and 180 HE)
Design modifications based on test beam results completed by 04-Sep-2015
Assembly (done in 5 batches) of HE begins 28-Dec-2015, ends 18-May-2016
Last batch of HE checked out at CERN, ready to mount on 12-Sep-2016
Assembly (done in 5 batches) of HB begins 19-May-2016, ends 10-Oct-2016
Last batch of HB checked out at CERN, ready to mount on 8-Feb-2017

3. Initial design assumptions
Mechanical dimensions for both HE and HB ODU does not change from test beam designs
Patch panel, brackets, etc. are not changed, cookie design accommodates different numbers of SiPMs on HE/HB and new SiPM package dimensions
Threading of fiber pigtails changes accordingly
HE ODU will have 48 channels
Depth segmentation based on scheme developed by Vitaly Smirnov
HB ODU will have 64 channels
Depth segmentation based on 4 depth segments per tower
Interleaving of back two segments for some redundancy
Optical mixers – not sure if needed, would simplify design and fabrication if these are not necessary (more on this later)

4. First prototype of the HE ODU
Up to 7 layers in depth
96 channels for each pair of ODU
Threading a bit tricky but achievable
With drawings and color-coding, can be done in “mass production” by students and techs
But there are a few issues to resolve

5. HE ODU design issues
With the current design for depth segmentation, one ODU of each pair requires 49 SiPMs while the other requires 47 SiPMs
Consequence of how cables bring in the fibers to the patch panels
We propose combining L-1 with L0,L1,L2 in tower 18 - open to other suggestions
Will result in 2 unused SiPM channels per pair of HE ODU
The patch panel has a connector for “Layer 00”
Second sigma loop in the 0th layer (?)
Not clear if these fibers are actually brought to the ODU – eliminate?
We propose threading fibers from Layer 00 together with Layer 0
There is no input for tower 17, L0 – we will use L00 instead
In 44 of the 48 channels the SiPM receives 4 fibers or less (2.8 mm device), the other 4 channels receive 5-7 fibers (3.3 mm device) vs vs
We propose this

6. Other HE ODU points, questions
The “cookie” will have 44 holes with 2.37 mm diameter (for 2,3,4 fibers) and 4 holes with 2.87 mm diameter (for 6,7 fibers)
The holes for attaching the patch plate and bottom plate to the cookie holder have been moved to provide increased clearance for a v-groove needed for the polishing fixture
Calibration inputs
We propose 7 calibration channels with 8 fibers per channel
Calibration fibers are 0.25 mm clear PMMA, one fiber to each SiPM with 8 extra fibers
Will springs be used to press the cookie to the SiPMs?
Will black anodizing be required?
If “yes” we need to know which areas to keep clear for electrical ground contact
There are unused connector slots – do we need blank connectors to cover them up?

7. 4 mm thick – 3.2 mm diameter fiber elements
Optical Mixers
Optical mixers are designed to spread large amounts of light across the SiPM active area
Effectively increases the dynamic range at the high end, but fast recovery time can also increase the dynamic range
Do we need the mixers for HE and/or HB?
We propose instrumenting half the preproduction ODU with mixers, test in the beam and on the bench, then make the final decision for full production
Note: overall design must account for additional thickness of mixer plate
Optical Mixer plate
4 mm thick – 3.2 mm diameter fiber elements

8. HB ODU Design issues
To our knowledge, no real study of the depth segmentation has been done yet – only design is the following:
4 segments in depth per tower
Identical for towers 1-15, tower 16 a bit different
Interleaving of back two segments, then combined in the electronics
ODUs will require 64 SiPM each – 32 small, 32 large diameter
Threading looks to be challenging because of the interleaving

9. HB ODU – 1st attempts at threading
This is doable but fibers are tight once the ODU is compacted to appropriate size
The threading above was done using the same “hybrid” package as in the HE ODU
Might be easier with 4 packages with all smaller diameter, 4 with all larger diameter