Spsc protodune-sp preliminary tdr review

1.3.1. To meet the very tight schedule of taking data before LS2, some aspects of ProtoDUNE-SP are not identical to those planned for DUNE. The TDR should clearly state what are the exceptions that are not identical, and explain how this impacts the risk mitigation for the DUNE far detector provided by ProtoDUNE-SP as an engineering prototype. Key cold electronics components such as “the dedicated ASIC for serializing the data and providing a 1-GB/s link is not yet available, therefore an FPGA emulating its functionality is used; it is mounted on a dedicated mezzanine board…” It is not clear in the current version of the TDR why acquiring beam data is prioritized above an integrated full system test of the charge electronics in retiring risk.
The protoDUNE-SP detector provides a full prototype of: a large membrane cryostat, the High-Voltage system (field cage, cathode plane and HV feedthrough), the APAs, and options for the photon detector system. There will be future developments of the cold electronics. The front-end ASIC is essentially the final design. However, the cold ADC ASIC is not the final version and, as noted, the digital cold data ASIC for serializing the data is not yet available.
Nevertheless, the protoDUNE-SP programme will provide a full, large-scale prototype of the TPC; it is the only place where a full system test of this aspect will be available. It will allow DUNE to mitigate/retire many of the main risks related to the TPC (including HV) and will demonstrate the readout of the ionization using wrapped APAs. The timescale for the validation of the basic TPC design is driven by the schedule for the major LBNC and DOE reviews of the DUNE TDR that will take place in 2019. This sets the requirement for protoDUNE-SP data collection in 2018. The existing electronics chain is the only option for meeting this requirement.
A system-level test of the complete, final cold electronics chain (attached to an APA) will be made in a standalone facility. The validation and system-level characterization of the final electronics will be needed prior to CD-3b approval at the end of 2019. However, this time is approximately two years after the procurement of the electronics for the protoDUNE-SP detector.

1.3.1. Is it envisioned to upgrade the ProtoDUNE-SP electronics if there is further optimization to test these? The statement is made that “In the event further optimization of the ASICs is required based on the ProtoDUNE-SP findings, it can be implemented before the start of production in 2020.”
As part of our risk mitigation programme we are likely to pursue a second design for the cold electronics chain, as a backup, to address concerns over the robustness of the existing design. If the backup design goes forward, there may be an argument for testing it in protoDUNE-SP. At this stage, we would not want to exclude this possibility as we may require flexibility to adapt to future developments. The need for upgrading the protoDUNE-SP would very much depend on the differences between the final design and the current system. This will determine whether a standalone test or a more complete test in protoDUNE-SP would be needed. Assuming the final version is based on the existing architecture, a standalone test is likely to be sufficient. If a backup solution, based on a different architecture were adopted, there could be a case for upgrading the protoDUNE-SP electronics.
We have set up a task force to review the cold electronics system, and we will be in a better position to answer this question when the committee reports back in May.

1.3.2. Re “will be used for energy calibration in the low energy range of the SN neutrino events…” SN neutrino energies peak at 5 MeV, way below any energy accessible from the beams. Please explain how the SN physics would benefit from the beam calibration.
SN neutrino interactions in LAr produce electrons (from Charged-Current interactions on Ar) with energies in the 5-50 MeV range. The peak is at about 30 MeV and there are almost no interactions below 5 MeV. Muons make up a few percent of the beam and anti-muons (mu+) will stop in the detector and decay producing Michel electrons. The Michel electrons are mostly in the energy range 20-50 MeV, which is the energy range of relevance to SN neutrinos.
1.4: Initially 16 weeks of beam data are assumed. What is the minimum beam time needed to guarantee a successful physics programme that goes beyond calibration and characterization of the detector?
As described previously, there are many uncertainties associated with this question. We are working on quantifying the connection between the beam analysis and the final systematic uncertainties. The presented run plan provides sufficient statistics to control systematics at the sub-percent level. This would not be achievable with data sample which was a factor 10 smaller. What about a factor two, i.e. an eight-week beam run? Assuming a clean and well-understood beam with the currently expected fluxes and assuming stable detector operation, this would probably be sufficient.
Exactly what assumptions go into the event rate calculation? E.g. “Data collection efficiency = 50%” is assumed here. What are the main causes of efficiency loss? The quantitative beam time request should be updated before the annual review, in time for the MTP.
To our knowledge, the question whether H2 and H4 tertiary beamlines at EHN1 can run simultaneously has not yet been resolved. To our knowledge, it is also not clear whether during the run period from June to October 2018, the secondary beam (upstream the target of the H4 beamline) will be shared with other users. The rate at which useful data can be accumulated also depends on the efficiency of matching the particle track in the external beamline detectors to the observed particle in the protoDUNE-SP detector. This efficiency, which is currently being studied in MC, depends on the amount of material upstream of the TPC active volume (beam window and beam plug) and on the tracking resolution of the detectors in the beamline. For the above reasons, we assumed a “collection efficiency” of 50%.
1.4 What is the drift time of the signal in the TPC? How many events are expected per spill? What is the probability of pile-up events, and of these, the probability of over-lapping events?
For the nominal electric field, the maximum drift time is 2ms. The instantaneous particle rates (during the ~4.8s spill) vary depending on the beam momentum from ~20Hz at 1GeV/c to ~200Hz at 7GeV/c. Even at the highest rate, the cumulative probability for the pile-up two or more events within the same 2ms frame is rather small ~ 6%. This is not a major concern.

2 Detector components
2.2 Anode Plane Assemblies (APA)
2.2.1. The ProtoDUNE-SP APA will have 2 neighbors, while the actual DUNE APA will have 5, a brief comment should be included on how the cross-talk measurement here is extrapolated to the FD performance.
The capacitive coupling between APAs is expected to be larger between two side-by-side modules than that for the top and bottom modules for two main reasons:

1. 
   they share a larger boundary area (6m x 9cm vs. 2.3m x 9cm);
   
2. 
   the outermost wire plane on the side is the U plane, whereas on the top it is the non-readout grid plane.
   

We do not expect crosstalk in the DUNE Far Detector to be significantly different from that in the ProtoDUNE-SP detector.

2.2.1. How do the efficiency, energy, and angular resolution requirements in Table 2.2 compare with demonstrated performance, e.g. from ICARUS, MicroBooNE? What number of dead channels per APA can be tolerated and still achieve these requirements? How is this requirement met (in the QA plan)?
Answer will be provided soon.

Table 2.3.: What is the mesh tension? Is relaxation an issue with respect to maintaining mesh-wire spacing (and what is this number?) Has the relaxation been measured, and is it consistent with a duration comparable to the proposed operation of ProtoDUNE-SP? How does this compare with the frame planarity spec (5 mm)? Same question about relaxation w.r.t. The wires, and the mesh+epoxy mount planned. What happens in the event of a rip in the mesh? What is the breaking tension for the mesh, and how does this compare with the thermal shock scenario? These points should be discussed in 2.2.4.
The mesh is made from fine bronze wire, 94% copper and 6% tin.  This is similar to the bronze alloy C51000 that has a stress relaxation of about 7% in 11 years at room temperature.  The mesh is pulled taut when it is applied to the frame but is not tensioned to a measured value.  The low stress relaxation value means that it is not expected to loosen significantly over time.  No tension loss has been observed on the 40% APA prototype built several years ago or on the 35T APAs built three years ago.

2.2.2. What level of shielding is achieved? Re: “The mesh layer serves to shield the sense planes from pickup from the Photon Detection System and from “ghost” tracks that would otherwise be visible when ionizing particles have a trajectory that passes through the collection plane.” Is there a measurement of the pickup from the PDS? Has this shielding scheme been demonstrated?
Both ArgoNeut/LArIAT and MicroBooNE have seen these ghost tracks. Adding a mesh plane should eliminate these artifacts. ICARUS, MicroBooNE and LArIAT also see pickup noise in the wire channels coincident with the firing of the PMTs. This problem is reported to be solved in ICARUS where a shielding wire grid was placed over the PMTs. From the 35t prototype run, we do not have any indication of such pickup from the DUNE PDS (SiPMs instead of PMTs). Nevertheless, since the PDS design is still being prototyped, we believe the shielding provides a robust system to maintain isolation between the TPC and PDS.

2.2.2. ProtoDUNE-SP reads a single drift volume with one APA, while the DUNE FD APA views two drift regions. Can the proposed operation configuration of ProtoDUNE-SP test pickup from the 2nd drift region? Re: “In the current design of the DUNE-SP far detector module, a central row of APAs is flanked by drift-fields, requiring sensitivity on both sides. The wrapped APAs allow the induction plane wires to sense drifting ionization originating from either side of the APA. This double-sided feature is not strictly necessary for the ProtoDUNE-SP arrangement, which has APAs located against the cryostat walls and a drift field on one side only, but it is compatible with this setup as the grid layer facing the wall effectively blocks any ionization generated outside the TPC from drifting into the wires on that side of the APA.” Has the ProtoDUNE-SP shielding scheme described here been demonstrated to be effective (e.g. with cosmic rays in a small setup)?
The DUNE APA has a symmetric, double-sided design for the wire planes. The outermost wire plane (the grid plane) is at a nominal -650 V bias voltage. This negative bias voltage creates a reverse drift field when placed against the grounded cryostat wall. Consequently, it is not possible to see electrons between the APA and the cryostat.
The double-sided readout was demonstrated in the smaller 35t prototype TPC, where two drift volumes were implemented to test the APA concept.

2.2.3. In the thermal shock case considered for the wire tension, is deformation of the frame (e.g. out of the plane) considered? Re: “In the limiting case, with complete contraction of the wire and none in the frame, the tension would be expected to reach 11.7 N. This is still well under the 20 N yield tension.” In this case is there still a safety factor of 2?
To first order, a minor out of plane distortion on the wire frame does not change the wire length or by extrapolation the wire tension. We do however, set limits on the out of plane distortion of the APA frame in order to maintain good wire plane spacing.


2.2.3. If a single wire breaks, what will be the impact? What is the expected number of wires that will be affected?
It is difficult to predict the behavior of a broken wire and therefore estimate the number of wires affected. We have designed the APA to have frequent intermediate supports on each wire such that the longest unsupported length is about 1.3m to minimize the reach of a broken wire. Any wire that comes to contact with others will observe much higher noise on the front-end ASIC at a minimum, up to temporary or permanent malfunction. Shorted wires between wire planes will cause changes in local electron transparency and in “induction” vs. “collection” distribution along all impacted wires through the APA. Shorted grid plane wires to other wire planes would lead to minor short range distortions in the drift region near the APA.
To our knowledge, no wire breakages have been observed in ICARUS, MicroBooNE, ArgonNEUT, or LArIAT. These are not fine wires, and the main risk of breakages is likely to be associated with “nicks” produced in the production/installation process. This is addressed by our QA/QC measures.

2.2.4: What is the electron deflection technique used to measure the collection inefficiency due to the joint between APAs, and what were the results obtained?
The electron deflection technique uses one or two conductive strips placed over and above the gap between two APAs, which are biased more negatively than the natural potential at their location, to push the incoming electrons away from this gap. This is the same principle that allows the incoming electrons to move around the induction plane wires. The deflector is implemented as a row of FR4 boards with copper strips on their edges. Although one deflector board was installed in the 35t TPC, we did not record sufficient data to evaluate its performance.
2.2.5. has the noise performance / crosstalk been demonstrated in bench tests for this wire board assembly design?
No. The 35t TPC used nearly identical wire board assembly. However, it was not possible to establish the noise performance due to the 35t cryostat environment.
2.2.5 how is the wire connection to the wire board verified during assembly? (QA plan). Until what point in the installation is repair possible?
The physical connection to the board is verified visually.  There must be adequate solder on the pad and the wire must run along the surface of the board, approximately centered in the solder.  After soldering, but before adding boards for the next layer, electrical continuity is tested between the end of the wire at the APA head and the end at the foot.  Any open wires are repaired.  After adding the next layer of boards it is not possible to re-solder a wire; the pad is covered by the next board.
2.2.5. The CR board resistors are rated to 2 kV. Is this sufficient for spark protection of the CR boards if there is a spark on the wires? What is typical spark voltage under these operating conditions?
It is assumed that the dielectric strength of LAr is greater than that of air. So a 2 kV rated resistor in air should be safe in LAr. The bias voltages on the wires are expected to be below 1kV. Since the wires are running in ionization mode instead of proportional mode as in the case of MWPCs, sparking on the wires is not considered a likely risk.
2.2.5. Has the bias voltage RC filter component selection for the CR boards been finalized? This is not clear from the text.
Yes, the RC filter components on the CR boards have been finalized.

2.2.5. A procedure is described to snip away wire segments after soldering on the foot boards. How is QA done to ensure end treatment doesn’t result in sparking?
A razor blade or flush wire cutter is used to cut the wire flush of the solder surface so no sharp point sticks out of the solder joint.
More detailed answer will be provided soon.
2.2.5. Has the wire positioning been verified after thermal cycling given the final board material and positioning fixture choices?
The boards are located by more than friction.  The head and side boards have close fitting holes that fit over sleeves screwed to the frame. These prevent any significant movement between the boards and the frame (they would do this even if the hold down screws were to somehow loosen).  This approach wasn't possible on the foot boards but these are located with flat-head screws that prevent significant relative motion.

2.2.5. Has the aging of the epoxy been studied? (Think of the Big Dig!!) Is there a mechanical strength test of the solder and epoxy anchor in the QA procedure? Does any of this get repeated after delivery of the planes to CERN? What is the risk of shorting a whole plane of wires if one breaks?
We have several ongoing tests running at PSL with wires at 5 and 10 Newtons glued and/or soldered to boards.  The tension is measured every few months and some of these tests have been going on for several years.  Of the most interest are the tests using the solder-epoxy combination that we settled on for ProtoDUNE.  These show that either the epoxy or solder alone would retain the wires at 5N.  Having both gives an extra safety margin.

     As was mentioned earlier, it's difficult to predict all the possible outcomes if a wire were to break.  It depends on whether the end of the broken wire is still held by the epoxy or whether it's completely free and it depends on the length of any loose end formed when the wire breaks.

2.2.6. Has the electrical isolation scheme of one APA from its neighbor been demonstrated after thermal cycling? How is this verified during installation? (QA plan) What level of vibration is expected in situ, and is the isolation scheme robust at this level?
The electrical isolation scheme has not been demonstrated as the first APA is being fabricated. (Note the pin depicted in Fig. 2.12 should have an insulating sleeve around it.) This isolation can be easily verified since each APA is electrically isolated from its mechanical suspension. A simple resistance measurement between the two APA frames is all one has to do at installation. We cannot allow any vibration large enough to move the wires by a few microns for the low noise operation of the TPC, so we don’t expect the robust isolation features to fail due to vibration.

2.3 Cathode Plane Assembly (CPA)
2.3.2. Has it been demonstrated that the scheme for redundant connections in the HV bus between CPA columns doesn’t introduce noise?
We have interpreted this question as: does the loop formed by the HV bus introduce noise?
At high frequencies, the membrane cryostat and the outer warm structure form a good faraday cage. We are not expecting noticeable EM interference entering the cryostat. At low frequencies, the cryostat is less opaque to magnetic flux. Changing magnetic flux will induce a current in the HV bus loop. However, since the bus is low impedance, the current driven by EMF does not result in significant voltage swing, which can be capacitively coupled to the sense wires. Had the HV bus not been a closed loop, the EMF could drive voltage swings at the ends of the bus and therefore couple to the sense wires. In any case, any EM interference from internal or external sources would have stronger direct coupling to the 7m long sense wires compared to the coupling to the CPA followed by capacitive coupling to the sense wires.
We can open the HV bus loop before the cryostat is sealed and perform an in-air to check for noise issues related to the HV bus
2.3.2. How will the resistivity of the ground-facing surfaces be verified after installation? How will the breakdown voltage of the HV cup be verified? If a problem is found, what is the latest point in the installation at which it can be fixed?
There is no ground facing cathode surface in the protoDUNE-SP detector.
The HV cup is modeled in 3D with E_max~ 20kV/cm(?). It is currently being tested in the 35t cryostat HV test setup. The HV cup can be replaced by setting up scaffold in the east aisle outside of the TPC.
2.3.2. What is the peak energy dumped in the case of a discharge? How does this compare with the level at which damage to the readout electronics occurs (assuming capacitive coupling)?

The total energy stored in the protoDUNE-SP electric-field is about 30J. The impact of a discharge on the front-end ASICs is documented in dune-doc-1320.

2.3.2. Can you quantify by how much this design reduces the instantaneous charge injection to the front-end electronics?
With an all metallic cathode, the time constant of discharging the cathode is assumed to be tens of ns. With a resistive cathode of 1 GOhm/sq, the discharge time constant is approximately 1s. Our current choice for the resistive cathode surface has ~ 4 MOhm/sq, so the estimated instantaneous current injection into FEE should be more than 2 orders of magnitudes lower than a conductive cathode.
2.3.4. What voltage drop is acceptable across the planes to achieve the physics requirements? How does this compare with the <<1% value given here? What level of field uniformity is achieved at the edge of the planes, given this field shaping profile design? What level is required to meet the physics goals?
The uniformity of drift field inside the boundary of the active volume is better than 2%, except at field cage module boundaries. The uniformity of drift field inside the fiducial volume is better than 1%, except at the corner edges of the field cage where larger gaps between field cage modules are present to accommodate TPC installation. To create 1% changes in drift field, the voltage of the cathode needs to change 1.8kV. This should be compared to the E-field distortion from space charge buildup, which in protoDUNE is expected to be larger than +/- 5%. Provided the space charge effects can be understood, they can in principal be largely calibrated out in software. This will need to be demonstrated in practice.

2.4 Field Cage (FC)
Re 1.3.1. Has the full operating high voltage (HV) been achieved in the prototype feedthrough? It is not clear that the test described in Section 1.3.1. Is conservative, if the test didn’t achieve full HV.
The feedthrough to be used in protoDUNE-SP is an exact copy of the Dual-Phase feedthrough. This has been tested at 300kV in ultrapure LAr (purified through commercial Hydrosorb Oxisorb which purifies at the 0.1 ppb level) in a cryostat that was previously evacuated to minimize the residual air contamination of LAr.
2.4.2. Given the 20 cm distance between ground planes and field profiles, in the event of breakdown what is the surface current on the supports? Can sparking follow this path? Has this been tested?
Previous discharge studies seem to suggest that discharge paths along insulator surfaces are random and non-repeating, even after markings have been made from prior sparks.
2.4.2. What is the peak energy dumped in the case of a discharge? How does this compare with the level at which damage to the readout electronics occurs (assuming capacitive coupling)?
The field cage voltage is linearly graded by the divider. In a discharge, the sudden voltage swing on the divider will be distributed along the field cage at the worst case linearly (heavily concentrated on the CPA end, assuming a discharge occurs there).  Each field cage module has about 1J of stored energy, and its coupling to the FEE is less than that of the CPA, due to the much smaller solid angle and the divider network. In any case, the field cage and CPAs are electrically connected. In a discharge, eventually all stored energy could be dumped if the discharge does not have an extinction voltage. A full simulation with the field cage equivalent circuits included is under way for a more complete study of the field cage discharge behavior.